Release 3.0.0 (#2545)

* bump version
* update release notes
* removed 2.7, 3.4, 3.5, and 3.6 from supported versions in setup.py
* switched GH actions back to ubuntu-latest, removed tests with 2.7 and 3.6, and added 3.11
* some little fixes in Citus documentation and behave tests

.github/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,48 @@
+---
+name: Bug report
+about: Create a report to help us improve
+title: ''
+labels: ''
+assignees: ''
+
+---
+
+**Describe the bug**
+A clear and concise description of what the bug is.
+
+**To Reproduce**
+Steps to reproduce the behavior:
+
+**Expected behavior**
+A clear and concise description of what you expected to happen.
+
+**Screenshots**
+If applicable, add screenshots to help explain your problem.
+
+**Environment**
+-   Patroni version:
+-   PostgreSQL version:
+-   DCS (and its version):
+
+**Patroni configuration file**
+```
+Please copy&paste your Patroni configuration file here
+```
+
+**patronictl show-config**
+```
+Please copy&paste the output of "patronictl show-config" command here
+```
+
+**Have you checked Patroni logs?**
+Please provide a snippet of Patroni log files here
+
+**Have you checked PostgreSQL logs?**
+Please provide a snippet here
+
+**Have you tried to use GitHub issue search?**
+Maybe there is already a similar issue solved.
+
+
+**Additional context**
+Add any other context about the problem here.

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1 @@
+blank_issues_enabled: false

.github/workflows/install_deps.py

@@ -0,0 +1,143 @@
+import inspect
+import os
+import shutil
+import subprocess
+import stat
+import sys
+import tarfile
+import zipfile
+
+
+def install_requirements(what):
+    old_path = sys.path[:]
+    w = os.path.join(os.getcwd(), os.path.dirname(inspect.getfile(inspect.currentframe())))
+    sys.path.insert(0, os.path.dirname(os.path.dirname(w)))
+    try:
+        from setup import EXTRAS_REQUIRE, read
+    finally:
+        sys.path = old_path
+    requirements = ['mock>=2.0.0', 'flake8', 'pytest', 'pytest-cov'] if what == 'all' else ['behave']
+    requirements += ['coverage']
+    # try to split tests between psycopg2 and psycopg3
+    requirements += ['psycopg[binary]'] if sys.version_info > (3, 7, 0) and\
+        (sys.platform != 'darwin' or what == 'etcd3') else ['psycopg2-binary']
+    for r in read('requirements.txt').split('\n'):
+        r = r.strip()
+        if r != '':
+            extras = {e for e, v in EXTRAS_REQUIRE.items() if v and any(r.startswith(x) for x in v)}
+            if not extras or what == 'all' or what in extras:
+                requirements.append(r)
+
+    subprocess.call([sys.executable, '-m', 'pip', 'install', '--upgrade', 'pip'])
+    subprocess.call([sys.executable, '-m', 'pip', 'install', '--upgrade', 'wheel'])
+    r = subprocess.call([sys.executable, '-m', 'pip', 'install'] + requirements)
+    s = subprocess.call([sys.executable, '-m', 'pip', 'install', '--upgrade', 'setuptools'])
+    return s | r
+
+
+def install_packages(what):
+    from mapping import versions
+
+    packages = {
+        'zookeeper': ['zookeeper', 'zookeeper-bin', 'zookeeperd'],
+        'consul': ['consul'],
+    }
+    packages['exhibitor'] = packages['zookeeper']
+    packages = packages.get(what, [])
+    ver = versions.get(what)
+    subprocess.call(['sudo', 'apt-get', 'update', '-y'])
+    return subprocess.call(['sudo', 'apt-get', 'install', '-y', 'postgresql-' + ver, 'expect-dev'] + packages)
+
+
+def get_file(url, name):
+    try:
+        from urllib.request import urlretrieve
+    except ImportError:
+        from urllib import urlretrieve
+
+    print('Downloading ' + url)
+    urlretrieve(url, name)
+
+
+def untar(archive, name):
+    with tarfile.open(archive) as tar:
+        f = tar.extractfile(name)
+        dest = os.path.basename(name)
+        with open(dest, 'wb') as d:
+            shutil.copyfileobj(f, d)
+            return dest
+
+
+def unzip(archive, name):
+    with zipfile.ZipFile(archive, 'r') as z:
+        name = z.extract(name)
+        dest = os.path.basename(name)
+        shutil.move(name, dest)
+        return dest
+
+
+def unzip_all(archive):
+    print('Extracting ' + archive)
+    with zipfile.ZipFile(archive, 'r') as z:
+        z.extractall()
+
+
+def chmod_755(name):
+    os.chmod(name, stat.S_IRUSR | stat.S_IWUSR | stat.S_IXUSR |
+             stat.S_IRGRP | stat.S_IXGRP | stat.S_IROTH | stat.S_IXOTH)
+
+
+def unpack(archive, name):
+    print('Extracting {0} from {1}'.format(name, archive))
+    func = unzip if archive.endswith('.zip') else untar
+    name = func(archive, name)
+    chmod_755(name)
+    return name
+
+
+def install_etcd():
+    version = os.environ.get('ETCDVERSION', '3.4.23')
+    platform = {'linux2': 'linux', 'win32': 'windows', 'cygwin': 'windows'}.get(sys.platform, sys.platform)
+    dirname = 'etcd-v{0}-{1}-amd64'.format(version, platform)
+    ext = 'tar.gz' if platform == 'linux' else 'zip'
+    name = '{0}.{1}'.format(dirname, ext)
+    url = 'https://github.com/etcd-io/etcd/releases/download/v{0}/{1}'.format(version, name)
+    get_file(url, name)
+    ext = '.exe' if platform == 'windows' else ''
+    return int(unpack(name, '{0}/etcd{1}'.format(dirname, ext)) is None)
+
+
+def install_postgres():
+    version = os.environ.get('PGVERSION', '15.1-1')
+    platform = {'darwin': 'osx', 'win32': 'windows-x64', 'cygwin': 'windows-x64'}[sys.platform]
+    if platform == 'osx':
+        return subprocess.call(['brew', 'install', 'expect', 'postgresql@{0}'.format(version.split('.')[0])])
+    name = 'postgresql-{0}-{1}-binaries.zip'.format(version, platform)
+    get_file('http://get.enterprisedb.com/postgresql/' + name, name)
+    unzip_all(name)
+    bin_dir = os.path.join('pgsql', 'bin')
+    for f in os.listdir(bin_dir):
+        chmod_755(os.path.join(bin_dir, f))
+    return subprocess.call(['pgsql/bin/postgres', '-V'])
+
+
+def main():
+    what = os.environ.get('DCS', sys.argv[1] if len(sys.argv) > 1 else 'all')
+
+    if what != 'all':
+        if sys.platform.startswith('linux'):
+            r = install_packages(what)
+        else:
+            r = install_postgres()
+
+        if r == 0 and what.startswith('etcd'):
+            r = install_etcd()
+
+        if r != 0:
+            return r
+
+    return install_requirements(what)
+
+
+if __name__ == '__main__':
+    sys.exit(main())

.github/workflows/mapping.py

@@ -0,0 +1 @@
+versions = {'etcd': '9.6', 'etcd3': '14', 'consul': '13', 'exhibitor': '12', 'raft': '11', 'kubernetes': '15'}

.github/workflows/release.yaml

@@ -0,0 +1,41 @@
+name: Publish Patroni distributions to PyPI and TestPyPI
+on:
+  push:
+    tags:
+      - 'v[0-9]+.[0-9]+.[0-9]+'
+  release:
+    types:
+      - published
+jobs:
+  build-n-publish:
+    name: Build and publish Patroni distributions to PyPI and TestPyPI
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@master
+
+    - name: Set up Python 3.9
+      uses: actions/setup-python@v4
+      with:
+        python-version: 3.9
+
+    - name: Install dependencies
+      run: python .github/workflows/install_deps.py
+
+    - name: Run tests and flake8
+      run: python .github/workflows/run_tests.py
+
+    - name: Build a binary wheel and a source tarball
+      run: python setup.py sdist bdist_wheel
+
+    - name: Publish distribution to Test PyPI
+      if: github.event_name == 'push'
+      uses: pypa/[email protected]
+      with:
+        password: ${{ secrets.TEST_PYPI_API_TOKEN }}
+        repository_url: https://test.pypi.org/legacy/
+
+    - name: Publish distribution to PyPI
+      if: github.event_name == 'release'
+      uses: pypa/[email protected]
+      with:
+        password: ${{ secrets.PYPI_API_TOKEN }}

.github/workflows/run_tests.py

@@ -0,0 +1,48 @@
+import os
+import shutil
+import subprocess
+import sys
+import tempfile
+
+
+def main():
+    what = os.environ.get('DCS', sys.argv[1] if len(sys.argv) > 1 else 'all')
+
+    if what == 'all':
+        flake8 = subprocess.call([sys.executable, 'setup.py', 'flake8'])
+        test = subprocess.call([sys.executable, 'setup.py', 'test'])
+        version = '.'.join(map(str, sys.version_info[:2]))
+        shutil.move('.coverage', os.path.join(tempfile.gettempdir(), '.coverage.' + version))
+        return flake8 | test
+    elif what == 'combine':
+        tmp = tempfile.gettempdir()
+        for name in os.listdir(tmp):
+            if name.startswith('.coverage.'):
+                shutil.move(os.path.join(tmp, name), name)
+        return subprocess.call([sys.executable, '-m', 'coverage', 'combine'])
+
+    env = os.environ.copy()
+    if sys.platform.startswith('linux'):
+        from mapping import versions
+
+        version = versions.get(what)
+        path = '/usr/lib/postgresql/{0}/bin:.'.format(version)
+        unbuffer = ['timeout', '900', 'unbuffer']
+    else:
+        if sys.platform == 'darwin':
+            version = os.environ.get('PGVERSION', '15.1-1')
+            path = '/usr/local/opt/postgresql@{0}/bin:.'.format(version.split('.')[0])
+            unbuffer = ['unbuffer']
+        else:
+            path = os.path.abspath(os.path.join('pgsql', 'bin'))
+            unbuffer = []
+    env['PATH'] = path + os.pathsep + env['PATH']
+    env['DCS'] = what
+    if what == 'kubernetes':
+        env['PATRONI_KUBERNETES_CONTEXT'] = 'k3d-k3s-default'
+
+    return subprocess.call(unbuffer + [sys.executable, '-m', 'behave'], env=env)
+
+
+if __name__ == '__main__':
+    sys.exit(main())

.github/workflows/tests.yaml

@@ -0,0 +1,157 @@
+name: Tests
+
+on:
+  pull_request:
+  push:
+    branches:
+    - master
+
+env:
+  CODACY_PROJECT_TOKEN: ${{ secrets.CODACY_PROJECT_TOKEN }}
+  SECRETS_AVAILABLE: ${{ secrets.CODACY_PROJECT_TOKEN != '' }}
+
+jobs:
+  unit:
+    runs-on: ${{ matrix.os }}-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu, windows, macos]
+
+    steps:
+    - uses: actions/checkout@v3
+
+    - name: Set up Python 3.7
+      uses: actions/setup-python@v4
+      with:
+        python-version: 3.7
+    - name: Install dependencies
+      run: python .github/workflows/install_deps.py
+    - name: Run tests and flake8
+      run: python .github/workflows/run_tests.py
+
+    - name: Set up Python 3.8
+      uses: actions/setup-python@v4
+      with:
+        python-version: 3.8
+    - name: Install dependencies
+      run: python .github/workflows/install_deps.py
+    - name: Run tests and flake8
+      run: python .github/workflows/run_tests.py
+
+    - name: Set up Python 3.9
+      uses: actions/setup-python@v4
+      with:
+        python-version: 3.9
+    - name: Install dependencies
+      run: python .github/workflows/install_deps.py
+    - name: Run tests and flake8
+      run: python .github/workflows/run_tests.py
+
+    - name: Set up Python 3.10
+      uses: actions/setup-python@v4
+      with:
+        python-version: '3.10'
+    - name: Install dependencies
+      run: python .github/workflows/install_deps.py
+    - name: Run tests and flake8
+      run: python .github/workflows/run_tests.py
+
+    - name: Set up Python 3.11
+      uses: actions/setup-python@v4
+      with:
+        python-version: 3.11
+    - name: Install dependencies
+      run: python .github/workflows/install_deps.py
+    - name: Run tests and flake8
+      run: python .github/workflows/run_tests.py
+
+    - name: Combine coverage
+      run: python .github/workflows/run_tests.py combine
+
+    - name: Install coveralls
+      run: python -m pip install coveralls
+
+    - name: Upload Coverage
+      env:
+        COVERALLS_FLAG_NAME: unit-${{ matrix.os }}
+        COVERALLS_PARALLEL: 'true'
+        GITHUB_TOKEN: ${{ secrets.github_token }}
+      run: python -m coveralls --service=github
+
+  behave:
+    runs-on: ${{ matrix.os }}-latest
+    env:
+      DCS: ${{ matrix.dcs }}
+      ETCDVERSION: 3.4.23
+      PGVERSION: 15.1-1  # for windows and macos
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu]
+        python-version: [3.7, '3.10']
+        dcs: [etcd, etcd3, consul, exhibitor, kubernetes, raft]
+        include:
+        - os: macos
+          python-version: 3.8
+          dcs: raft
+        - os: macos
+          python-version: 3.9
+          dcs: etcd
+        - os: macos
+          python-version: 3.11
+          dcs: etcd3
+
+    steps:
+    - uses: actions/checkout@v3
+    - name: Set up Python
+      uses: actions/setup-python@v4
+      with:
+        python-version: ${{ matrix.python-version }}
+    - uses: nolar/setup-k3d-k3s@v1
+      if: matrix.dcs == 'kubernetes'
+    - name: Add postgresql apt repo
+      run: |
+        sudo apt-get update -y
+        sudo apt-get install -y wget ca-certificates gnupg
+        sudo sh -c 'echo "deb http://apt.postgresql.org/pub/repos/apt $(lsb_release -cs)-pgdg main" > /etc/apt/sources.list.d/pgdg.list'
+        sudo sh -c 'wget -qO - https://www.postgresql.org/media/keys/ACCC4CF8.asc | gpg --dearmor > /etc/apt/trusted.gpg.d/apt.postgresql.org.gpg'
+      if: matrix.os == 'ubuntu'
+    - name: Install dependencies
+      run: python .github/workflows/install_deps.py
+    - name: Run behave tests
+      run: python .github/workflows/run_tests.py
+    - name: Upload logs if behave failed
+      uses: actions/upload-artifact@v3
+      if: failure()
+      with:
+        name: behave-${{ matrix.os }}-${{ matrix.dcs }}-${{ matrix.python-version }}-logs
+        path: |
+          features/output/*_failed/*postgres?.*
+          features/output/*.log
+        if-no-files-found: error
+        retention-days: 5
+    - name: Generate coverage xml report
+      run: python -m coverage xml -o cobertura.xml
+    - name: Upload coverage to Codacy
+      run: bash <(curl -Ls https://coverage.codacy.com/get.sh) report -r cobertura.xml -l Python --partial
+      if: ${{ env.SECRETS_AVAILABLE == 'true' }}
+
+  coveralls-finish:
+    name: Finalize coveralls.io
+    needs: unit
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/setup-python@v4
+    - run: python -m pip install coveralls
+    - run: python -m coveralls --service=github --finish
+      env:
+        GITHUB_TOKEN: ${{ secrets.github_token }}
+
+  codacy-final:
+    name: Finalize Codacy
+    needs: behave
+    runs-on: ubuntu-latest
+    steps:
+    - run: bash <(curl -Ls https://coverage.codacy.com/get.sh) final
+      if: ${{ env.SECRETS_AVAILABLE == 'true' }}

.travis.yml

@@ -1,157 +0,0 @@
-sudo: true
-dist: trusty
-language: python
-addons:
-  apt:
-    packages:
-      - expect-dev # for unbuffer
-env:
-  global:
-  - ETCDVERSION=3.0.17 ZKVERSION=3.4.11 CONSULVERSION=0.7.4
-  - PYVERSIONS="2.7 3.5 3.6"
-  - EXCLUDE_BEHAVE="3.5"
-  - BOTO_CONFIG=/doesnotexist
-matrix:
-  include:
-  - python: "3.5"
-    env: TEST_SUITE="python setup.py"
-  - python: "3.6"
-    env: DCS="etcd" TEST_SUITE="behave"
-  - python: "3.6"
-    env: DCS="exhibitor" TEST_SUITE="behave"
-  - python: "3.6"
-    env: DCS="consul" TEST_SUITE="behave"
-  - python: "3.6"
-    env: DCS="kubernetes" TEST_SUITE="behave"
-branches:
-  only:
-  - master
-  - /^v\d+\.\d+(\.\d+)?$/
-cache:
-  directories:
-  - $HOME/mycache
-before_cache:
-- |
-  rm -fr $HOME/mycache/python*
-  for pv in $PYVERSIONS; do
-      if [[ $TEST_SUITE != "behave" || $pv != $EXCLUDE_BEHAVE ]]; then
-          fpv=$(basename $(readlink $HOME/virtualenv/python${pv}))
-          mv $HOME/virtualenv/${fpv} $HOME/mycache/${fpv}
-      fi
-  done
-install:
-  - |
-    set -e
-
-    if [[ $TEST_SUITE == "behave" ]]; then
-        function get_consul() {
-            CC=~/mycache/consul_${CONSULVERSION}
-            if [[ ! -x $CC ]]; then
-                curl -L https://releases.hashicorp.com/consul/${CONSULVERSION}/consul_${CONSULVERSION}_linux_amd64.zip \
-                    | gunzip > $CC
-                [[ ${PIPESTATUS[0]} == 0 ]] || return 1
-                chmod +x $CC
-            fi
-            ln -s $CC consul
-        }
-
-        function get_etcd() {
-            EC=~/mycache/etcd_${ETCDVERSION}
-            if [[ ! -x $EC ]]; then
-                curl -L https://github.com/coreos/etcd/releases/download/v${ETCDVERSION}/etcd-v${ETCDVERSION}-linux-amd64.tar.gz \
-                    | tar xz -C . --strip=1 --wildcards --no-anchored etcd
-                [[ ${PIPESTATUS[0]} == 0 ]] || return 1
-                mv etcd $EC
-            fi
-            ln -s $EC etcd
-        }
-
-        function get_kubernetes() {
-            wget -O localkube "https://storage.googleapis.com/minikube/k8sReleases/v1.7.0/localkube-linux-amd64"
-            chmod +x localkube
-            sudo nohup ./localkube --logtostderr=true --enable-dns=false > localkube.log 2>&1 &
-
-            echo "Waiting for localkube to start..."
-            if ! timeout 120 sh -c "while ! curl -ks http://127.0.0.1:8080/ >/dev/null; do sleep 1; done"; then
-                sudo cat localkube.log
-                echo "localkube did not start"
-                exit 1
-            fi
-            echo "Check certificate permissions"
-            sudo chmod 644 /var/lib/localkube/certs/*
-            sudo ls -altr /var/lib/localkube/certs/
-
-            echo "Set up .kube/config"
-            mkdir ~/.kube
-            echo -e "apiVersion: v1\nclusters:\n- cluster:\n    certificate-authority: /var/lib/localkube/certs/ca.crt\n    server: https://127.0.0.1:8443\n  name: local\ncontexts:\n- context:\n    cluster: local\n    user: myself\n  name: local\ncurrent-context: local\nkind: Config\npreferences: {}\nusers:\n- name: myself\n  user:\n    client-certificate: /var/lib/localkube/certs/apiserver.crt\n    client-key: /var/lib/localkube/certs/apiserver.key\n" > ~/.kube/config
-        }
-
-        function get_exhibitor() {
-            ZC=~/mycache/zookeeper-${ZKVERSION}
-            if [[ ! -d $ZC ]]; then
-                curl -L http://www.apache.org/dist/zookeeper/zookeeper-${ZKVERSION}/zookeeper-${ZKVERSION}.tar.gz | tar xz
-                [[ ${PIPESTATUS[0]} == 0 ]] || return 1
-                mv zookeeper-${ZKVERSION}/conf/zoo_sample.cfg zookeeper-${ZKVERSION}/conf/zoo.cfg
-                mv zookeeper-${ZKVERSION} $ZC
-            fi
-            $ZC/bin/zkServer.sh start
-            # following lines are 'emulating' exhibitor REST API
-            while true; do
-                echo -e 'HTTP/1.0 200 OK\nContent-Type: application/json\n\n{"servers":["127.0.0.1"],"port":2181}' \
-                    | nc -l 8181 &> /dev/null
-            done&
-        }
-
-        attempt_num=1
-        until get_${DCS}; do
-            [[ $attempt_num -ge 3 ]] && exit 1
-            echo "Attempt $attempt_num failed! Trying again in $attempt_num seconds..."
-            sleep $(( attempt_num++ ))
-        done
-    fi
-
-    for pv in $PYVERSIONS; do
-        if [[ $TEST_SUITE != "behave" || $pv != $EXCLUDE_BEHAVE ]]; then
-            fpv=$(basename $(readlink $HOME/virtualenv/python$pv))
-            if [[ -d ~/mycache/${fpv} ]]; then
-                mv ~/virtualenv/${fpv} ~/virtualenv/${fpv}.bckp
-                mv ~/mycache/${fpv} ~/virtualenv/${fpv}
-            fi
-            source ~/virtualenv/python${pv}/bin/activate
-            # explicitly install all needed python modules to cache them
-            for p in '-r requirements.txt' 'psycopg2-binary behave codacy-coverage coverage coveralls flake8 mock pytest-cov pytest setuptools'; do
-                pip install $p --upgrade
-            done
-        fi
-    done
-script:
-  - |
-    for pv in $PYVERSIONS; do
-        if [[ $TEST_SUITE == "behave" && $pv == $EXCLUDE_BEHAVE ]]; then
-            continue
-        fi
-
-        source ~/virtualenv/python${pv}/bin/activate
-
-        if [[ $TEST_SUITE != "behave" ]]; then
-            echo Running unit tests using python${pv}
-            unbuffer $TEST_SUITE test
-            $TEST_SUITE flake8
-        elif [[ $pv != $EXCLUDE_BEHAVE ]]; then
-            echo Running acceptance tests using python${pv}
-            if ! PATH=.:/usr/lib/postgresql/9.6/bin:$PATH unbuffer $TEST_SUITE; then
-                # output all log files when tests are failing
-                grep . features/output/*_failed/*postgres?.*
-                exit 1
-            fi
-        fi
-    done
-
-    set +e
-after_success:
-  # before_cache is executed earlier than after_success, so we need to restore one of virtualenv directories
-  - fpv=$(basename $(readlink $HOME/virtualenv/python3.6)) && mv $HOME/mycache/${fpv} $HOME/virtualenv/${fpv}
-  - coveralls
-  - if [[ $TEST_SUITE != "behave" ]]; then python-codacy-coverage -r coverage.xml; fi
-  - if [[ $DCS == "exhibitor" ]]; then ~/mycache/zookeeper-${ZKVERSION}/bin/zkServer.sh stop; fi
-  - sudo kill $(jobs -p)

CODEOWNERS

@@ -0,0 +1,2 @@
+# global owners
+*        @CyberDem0n @hughcapet

Dockerfile

@@ -1,6 +1,6 @@
 ## This Dockerfile is meant to aid in the building and debugging patroni whilst developing on your local machine
 ## It has all the necessary components to play/debug with a single node appliance, running etcd
-ARG PG_MAJOR=10
+ARG PG_MAJOR=15
 ARG COMPRESS=false
 ARG PGHOME=/home/postgres
 ARG PGDATA=$PGHOME/data
@@ -14,7 +14,7 @@ ARG PGDATA
 ARG LC_ALL
 ARG LANG
 
-ENV ETCDVERSION=2.3.8 CONFDVERSION=0.16.0
+ENV ETCDVERSION=3.3.13 CONFDVERSION=0.16.0
 
 RUN set -ex \
     && export DEBIAN_FRONTEND=noninteractive \
@@ -25,11 +25,12 @@ RUN set -ex \
             | grep -Ev '^python3-(sphinx|etcd|consul|kazoo|kubernetes)' \
             | xargs apt-get install -y vim curl less jq locales haproxy sudo \
                             python3-etcd python3-kazoo python3-pip busybox \
+                            net-tools iputils-ping --fix-missing \
     && pip3 install dumb-init \
 \
     # Cleanup all locales but en_US.UTF-8
     && find /usr/share/i18n/charmaps/ -type f ! -name UTF-8.gz -delete \
-    && find /usr/share/i18n/locales/ -type f ! -name en_US ! -name en_GB ! -name i18n ! -name iso14651_t1 ! -name iso14651_t1_common ! -name 'translit_*' -delete \
+    && find /usr/share/i18n/locales/ -type f ! -name en_US ! -name en_GB ! -name i18n* ! -name iso14651_t1 ! -name iso14651_t1_common ! -name 'translit_*' -delete \
     && echo 'en_US.UTF-8 UTF-8' > /usr/share/i18n/SUPPORTED \
 \
     # Make sure we have a en_US.UTF-8 locale available
@@ -42,18 +43,18 @@ RUN set -ex \
     && echo 'syntax on\nfiletype plugin indent on\nset mouse-=a\nautocmd FileType yaml setlocal ts=2 sts=2 sw=2 expandtab' > /etc/vim/vimrc.local \
 \
     # Prepare postgres/patroni/haproxy environment
-    && mkdir -p $PGHOME/.config/patroni /patroni /run/haproxy \
+    && mkdir -p "$PGHOME/.config/patroni" /patroni /run/haproxy \
-    && ln -s ../../postgres0.yml $PGHOME/.config/patroni/patronictl.yaml \
+    && ln -s ../../postgres0.yml "$PGHOME/.config/patroni/patronictl.yaml" \
     && ln -s /patronictl.py /usr/local/bin/patronictl \
     && sed -i "s|/var/lib/postgresql.*|$PGHOME:/bin/bash|" /etc/passwd \
     && chown -R postgres:postgres /var/log \
 \
     # Download etcd
-    && curl -sL https://github.com/coreos/etcd/releases/download/v${ETCDVERSION}/etcd-v${ETCDVERSION}-linux-amd64.tar.gz \
+    && curl -sL "https://github.com/coreos/etcd/releases/download/v$ETCDVERSION/etcd-v$ETCDVERSION-linux-$(dpkg --print-architecture).tar.gz" \
             | tar xz -C /usr/local/bin --strip=1 --wildcards --no-anchored etcd etcdctl \
 \
     # Download confd
-    && curl -sL https://github.com/kelseyhightower/confd/releases/download/v${CONFDVERSION}/confd-${CONFDVERSION}-linux-amd64 \
+    && curl -sL "https://github.com/kelseyhightower/confd/releases/download/v$CONFDVERSION/confd-$CONFDVERSION-linux-$(dpkg --print-architecture)" \
             > /usr/local/bin/confd && chmod +x /usr/local/bin/confd \
 \
     # Clean up all useless packages and some files
@@ -89,7 +90,7 @@ RUN set -ex \
     && find /usr/bin -xtype l -delete \
     && find /var/log -type f -exec truncate --size 0 {} \; \
     && find /usr/lib/python3/dist-packages -name '*test*' | xargs rm -fr \
-    && find /lib/x86_64-linux-gnu/security -type f ! -name pam_env.so ! -name pam_permit.so ! -name pam_unix.so -delete
+    && find /lib/$(uname -m)-linux-gnu/security -type f ! -name pam_env.so ! -name pam_permit.so ! -name pam_unix.so -delete
 
 # perform compression if it is necessary
 ARG COMPRESS
@@ -98,8 +99,10 @@ RUN if [ "$COMPRESS" = "true" ]; then \
         # Allow certain sudo commands from postgres
         && echo 'postgres ALL=(ALL) NOPASSWD: /bin/tar xpJf /a.tar.xz -C /, /bin/rm /a.tar.xz, /bin/ln -snf dash /bin/sh' >> /etc/sudoers \
         && ln -snf busybox /bin/sh \
-        && files="/bin/sh /usr/bin/sudo /usr/lib/sudo/sudoers.so /lib/x86_64-linux-gnu/security/pam_*.so" \
+        && arch=$(uname -m) \
-        && libs="$(ldd $files | awk '{print $3;}' | grep '^/' | sort -u) /lib/x86_64-linux-gnu/ld-linux-x86-64.so.* /lib/x86_64-linux-gnu/libnsl.so.* /lib/x86_64-linux-gnu/libnss_compat.so.*" \
+        && darch=$(uname -m | sed 's/_/-/') \
+        && files="/bin/sh /usr/bin/sudo /usr/lib/sudo/sudoers.so /lib/$arch-linux-gnu/security/pam_*.so" \
+        && libs="$(ldd $files | awk '{print $3;}' | grep '^/' | sort -u) /lib/ld-linux-$darch.so.* /lib/$arch-linux-gnu/ld-linux-$darch.so.* /lib/$arch-linux-gnu/libnsl.so.* /lib/$arch-linux-gnu/libnss_compat.so.* /lib/$arch-linux-gnu/libnss_files.so.*" \
         && (echo /var/run $files $libs | tr ' ' '\n' && realpath $files $libs) | sort -u | sed 's/^\///' > /exclude \
         && find /etc/alternatives -xtype l -delete \
         && save_dirs="usr lib var bin sbin etc/ssl etc/init.d etc/alternatives etc/apt" \
@@ -116,7 +119,7 @@ RUN if [ "$COMPRESS" = "true" ]; then \
 FROM scratch
 COPY --from=builder / /
 
-LABEL maintainer="Alexander Kukushkin <alexander.kukushkin@zalando.de>"
+LABEL maintainer="Alexander Kukushkin <akukushkin@microsoft.com>"
 
 ARG PG_MAJOR
 ARG COMPRESS
@@ -149,7 +152,8 @@ RUN sed -i 's/env python/&3/' /patroni*.py \
     && sed -i 's/^    \(replication\|superuser\|rewind\|unix_socket_directories\|\(\(  \)\{0,1\}\(username\|password\)\)\):/#&/' postgres?.yml \
     && sed -i 's/^      parameters:/      pg_hba:\n      - local all all trust\n      - host replication all all md5\n      - host all all all md5\n&\n        max_connections: 100/'  postgres?.yml \
     && if [ "$COMPRESS" = "true" ]; then chmod u+s /usr/bin/sudo; fi \
-    && chown -R postgres:postgres $PGHOME /run /etc/haproxy
+    && chmod +s /bin/ping \
+    && chown -R postgres:postgres "$PGHOME" /run /etc/haproxy
 
 USER postgres
 

Dockerfile.citus

@@ -0,0 +1,173 @@
+## This Dockerfile is meant to aid in the building and debugging patroni whilst developing on your local machine
+## It has all the necessary components to play/debug with a single node appliance, running etcd
+ARG PG_MAJOR=15
+ARG COMPRESS=false
+ARG PGHOME=/home/postgres
+ARG PGDATA=$PGHOME/data
+ARG LC_ALL=C.UTF-8
+ARG LANG=C.UTF-8
+
+FROM postgres:$PG_MAJOR as builder
+
+ARG PGHOME
+ARG PGDATA
+ARG LC_ALL
+ARG LANG
+
+ENV ETCDVERSION=3.3.13 CONFDVERSION=0.16.0
+
+RUN set -ex \
+    && export DEBIAN_FRONTEND=noninteractive \
+    && echo 'APT::Install-Recommends "0";\nAPT::Install-Suggests "0";' > /etc/apt/apt.conf.d/01norecommend \
+    && apt-get update -y \
+    # postgres:10 is based on debian, which has the patroni package. We will install all required dependencies
+    && apt-cache depends patroni | sed -n -e 's/.*Depends: \(python3-.\+\)$/\1/p' \
+            | grep -Ev '^python3-(sphinx|etcd|consul|kazoo|kubernetes)' \
+            | xargs apt-get install -y vim curl less jq locales haproxy sudo \
+                            python3-etcd python3-kazoo python3-pip busybox \
+                            net-tools iputils-ping --fix-missing \
+    && curl https://install.citusdata.com/community/deb.sh | bash \
+    && apt-get -y install postgresql-$PG_MAJOR-citus-11.1 \
+    && pip3 install dumb-init \
+\
+    # Cleanup all locales but en_US.UTF-8
+    && find /usr/share/i18n/charmaps/ -type f ! -name UTF-8.gz -delete \
+    && find /usr/share/i18n/locales/ -type f ! -name en_US ! -name en_GB ! -name i18n* ! -name iso14651_t1 ! -name iso14651_t1_common ! -name 'translit_*' -delete \
+    && echo 'en_US.UTF-8 UTF-8' > /usr/share/i18n/SUPPORTED \
+\
+    # Make sure we have a en_US.UTF-8 locale available
+    && localedef -i en_US -c -f UTF-8 -A /usr/share/locale/locale.alias en_US.UTF-8 \
+\
+    # haproxy dummy config
+    && echo 'global\n        stats socket /run/haproxy/admin.sock mode 660 level admin' > /etc/haproxy/haproxy.cfg \
+\
+    # vim config
+    && echo 'syntax on\nfiletype plugin indent on\nset mouse-=a\nautocmd FileType yaml setlocal ts=2 sts=2 sw=2 expandtab' > /etc/vim/vimrc.local \
+\
+    # Prepare postgres/patroni/haproxy environment
+    && mkdir -p $PGHOME/.config/patroni /patroni /run/haproxy \
+    && ln -s ../../postgres0.yml $PGHOME/.config/patroni/patronictl.yaml \
+    && ln -s /patronictl.py /usr/local/bin/patronictl \
+    && sed -i "s|/var/lib/postgresql.*|$PGHOME:/bin/bash|" /etc/passwd \
+    && chown -R postgres:postgres /var/log \
+\
+    # Download etcd
+    && curl -sL https://github.com/coreos/etcd/releases/download/v${ETCDVERSION}/etcd-v${ETCDVERSION}-linux-$(dpkg --print-architecture).tar.gz \
+            | tar xz -C /usr/local/bin --strip=1 --wildcards --no-anchored etcd etcdctl \
+\
+    # Download confd
+    && curl -sL https://github.com/kelseyhightower/confd/releases/download/v${CONFDVERSION}/confd-${CONFDVERSION}-linux-$(dpkg --print-architecture) \
+            > /usr/local/bin/confd && chmod +x /usr/local/bin/confd \
+    # Prepare client cert for HAProxy
+    && cat /etc/ssl/private/ssl-cert-snakeoil.key  /etc/ssl/certs/ssl-cert-snakeoil.pem > /etc/ssl/private/ssl-cert-snakeoil.crt \
+\
+    # Clean up all useless packages and some files
+    && apt-get purge -y --allow-remove-essential python3-pip gzip bzip2 util-linux e2fsprogs \
+                libmagic1 bsdmainutils login ncurses-bin libmagic-mgc e2fslibs bsdutils \
+                exim4-config gnupg-agent dirmngr libpython2.7-stdlib libpython2.7-minimal \
+    && apt-get autoremove -y \
+    && apt-get clean -y \
+    && rm -rf /var/lib/apt/lists/* \
+        /root/.cache \
+        /var/cache/debconf/* \
+        /etc/rc?.d \
+        /etc/systemd \
+        /docker-entrypoint* \
+        /sbin/pam* \
+        /sbin/swap* \
+        /sbin/unix* \
+        /usr/local/bin/gosu \
+        /usr/sbin/[acgipr]* \
+        /usr/sbin/*user* \
+        /usr/share/doc* \
+        /usr/share/man \
+        /usr/share/info \
+        /usr/share/i18n/locales/translit_hangul \
+        /usr/share/locale/?? \
+        /usr/share/locale/??_?? \
+        /usr/share/postgresql/*/man \
+        /usr/share/postgresql-common/pg_wrapper \
+        /usr/share/vim/vim80/doc \
+        /usr/share/vim/vim80/lang \
+        /usr/share/vim/vim80/tutor \
+#        /var/lib/dpkg/info/* \
+    && find /usr/bin -xtype l -delete \
+    && find /var/log -type f -exec truncate --size 0 {} \; \
+    && find /usr/lib/python3/dist-packages -name '*test*' | xargs rm -fr \
+    && find /lib/$(uname -m)-linux-gnu/security -type f ! -name pam_env.so ! -name pam_permit.so ! -name pam_unix.so -delete
+
+# perform compression if it is necessary
+ARG COMPRESS
+RUN if [ "$COMPRESS" = "true" ]; then \
+        set -ex \
+        # Allow certain sudo commands from postgres
+        && echo 'postgres ALL=(ALL) NOPASSWD: /bin/tar xpJf /a.tar.xz -C /, /bin/rm /a.tar.xz, /bin/ln -snf dash /bin/sh' >> /etc/sudoers \
+        && ln -snf busybox /bin/sh \
+        && arch=$(uname -m) \
+        && darch=$(uname -m | sed 's/_/-/') \
+        && files="/bin/sh /usr/bin/sudo /usr/lib/sudo/sudoers.so /lib/$arch-linux-gnu/security/pam_*.so" \
+        && libs="$(ldd $files | awk '{print $3;}' | grep '^/' | sort -u) /lib/ld-linux-$darch.so.* /lib/$arch-linux-gnu/ld-linux-$darch.so.* /lib/$arch-linux-gnu/libnsl.so.* /lib/$arch-linux-gnu/libnss_compat.so.* /lib/$arch-linux-gnu/libnss_files.so.*" \
+        && (echo /var/run $files $libs | tr ' ' '\n' && realpath $files $libs) | sort -u | sed 's/^\///' > /exclude \
+        && find /etc/alternatives -xtype l -delete \
+        && save_dirs="usr lib var bin sbin etc/ssl etc/init.d etc/alternatives etc/apt" \
+        && XZ_OPT=-e9v tar -X /exclude -cpJf a.tar.xz $save_dirs \
+        # we call "cat /exclude" to avoid including files from the $save_dirs that are also among
+        # the exceptions listed in the /exclude, as "uniq -u" eliminates all non-unique lines.
+        # By calling "cat /exclude" a second time we guarantee that there will be at least two lines
+        # for each exception and therefore they will be excluded from the output passed to 'rm'.
+        && /bin/busybox sh -c "(find $save_dirs -not -type d && cat /exclude /exclude && echo exclude) | sort | uniq -u | xargs /bin/busybox rm" \
+        && /bin/busybox --install -s \
+        && /bin/busybox sh -c "find $save_dirs -type d -depth -exec rmdir -p {} \; 2> /dev/null"; \
+    else \
+        /bin/busybox --install -s; \
+    fi
+
+FROM scratch
+COPY --from=builder / /
+
+LABEL maintainer="Alexander Kukushkin <[email protected]>"
+
+ARG PG_MAJOR
+ARG COMPRESS
+ARG PGHOME
+ARG PGDATA
+ARG LC_ALL
+ARG LANG
+
+ARG PGBIN=/usr/lib/postgresql/$PG_MAJOR/bin
+
+ENV LC_ALL=$LC_ALL LANG=$LANG EDITOR=/usr/bin/editor
+ENV PGDATA=$PGDATA PATH=$PATH:$PGBIN
+
+COPY patroni /patroni/
+COPY extras/confd/conf.d/haproxy.toml /etc/confd/conf.d/
+COPY extras/confd/templates/haproxy-citus.tmpl /etc/confd/templates/haproxy.tmpl
+COPY patroni*.py docker/entrypoint.sh /
+COPY postgres?.yml $PGHOME/
+
+WORKDIR $PGHOME
+
+RUN sed -i 's/env python/&3/' /patroni*.py \
+    # "fix" patroni configs
+    && sed -i 's/^\(  connect_address:\|  - host\)/#&/' postgres?.yml \
+    && sed -i 's/^  listen: 127.0.0.1/  listen: 0.0.0.0/' postgres?.yml \
+    && sed -i "s|^\(  data_dir: \).*|\1$PGDATA|" postgres?.yml \
+    && sed -i "s|^#\(  bin_dir: \).*|\1$PGBIN|" postgres?.yml \
+    && sed -i 's/^  - encoding: UTF8/  - locale: en_US.UTF-8\n&/' postgres?.yml \
+    && sed -i 's/^scope:/log:\n  loggers:\n    patroni.postgresql.citus: DEBUG\n#&/' postgres?.yml \
+    && sed -i 's/^\(name\|etcd\|  host\|  authentication\|  pg_hba\|  parameters\):/#&/' postgres?.yml \
+    && sed -i 's/^    \(replication\|superuser\|rewind\|unix_socket_directories\|\(\(  \)\{0,1\}\(username\|password\)\)\):/#&/' postgres?.yml \
+    && sed -i 's/^postgresql:/&\n  basebackup:\n    checkpoint: fast/' postgres?.yml \
+    && sed -i 's|^      parameters:|      pg_hba:\n      - local all all trust\n      - hostssl replication all all md5 clientcert=verify-ca\n      - hostssl all all all md5 clientcert=verify-ca\n&\n        max_connections: 100\n        shared_buffers: 16MB\n        ssl: "on"\n        ssl_ca_file: /etc/ssl/certs/ssl-cert-snakeoil.pem\n        ssl_cert_file: /etc/ssl/certs/ssl-cert-snakeoil.pem\n        ssl_key_file: /etc/ssl/private/ssl-cert-snakeoil.key\n        citus.node_conninfo: "sslrootcert=/etc/ssl/certs/ssl-cert-snakeoil.pem sslkey=/etc/ssl/private/ssl-cert-snakeoil.key sslcert=/etc/ssl/certs/ssl-cert-snakeoil.pem sslmode=verify-ca"|'  postgres?.yml \
+    && sed -i 's/^#\(ctl\|  certfile\|  keyfile\)/\1/' postgres?.yml \
+    && sed -i 's|^#  cafile: .*$|  verify_client: required\n  cafile: /etc/ssl/certs/ssl-cert-snakeoil.pem|' postgres?.yml \
+    && sed -i 's|^#  cacert: .*$|  cacert: /etc/ssl/certs/ssl-cert-snakeoil.pem|' postgres?.yml \
+    && sed -i 's/^#  insecure: .*/  insecure: on/' postgres?.yml \
+    # client cert for HAProxy to access Patroni REST API
+    && if [ "$COMPRESS" = "true" ]; then chmod u+s /usr/bin/sudo; fi \
+    && chmod +s /bin/ping \
+    && chown -R postgres:postgres $PGHOME /run /etc/haproxy
+
+USER postgres
+
+ENTRYPOINT ["/bin/sh", "/entrypoint.sh"]

MAINTAINERS

@@ -1,3 +1,2 @@
-Alexander Kukushkin <alexander.kukushkin@zalando.de>
+Alexander Kukushkin <akukushkin@microsoft.com>
-Feike Steenbergen <feike.steenbergen@zalando.de>
+Polina Bungina <polina.bungina@zalando.de>
-Oleksii Kliukin <[email protected]>

README.rst

@@ -1,4 +1,4 @@
-|Build Status| |Coverage Status|
+|Tests Status| |Coverage Status|
 
 Patroni: A Template for PostgreSQL HA with ZooKeeper, etcd or Consul
 --------------------------------------------------------------------
@@ -12,6 +12,10 @@ Patroni is a template for you to create your own customized, high-availability s
 
 We call Patroni a "template" because it is far from being a one-size-fits-all or plug-and-play replication system. It will have its own caveats. Use wisely.
 
+Currently supported PostgreSQL versions: 9.3 to 15.
+
+**Note to Citus users**: Starting from 3.0 Patroni nicely integrates with `Citus <https://www.citusdata.com>`__. Please check `Citus support <https://github.com/zalando/patroni/blob/master/docs/citus.rst>`__ page for more information.
+
 **Note to Kubernetes users**: Patroni can run natively on top of Kubernetes. Take a look at the `Kubernetes <https://github.com/zalando/patroni/blob/master/docs/kubernetes.rst>`__ chapter of the Patroni documentation.
 
 .. contents::
@@ -45,7 +49,7 @@ We report new releases information `here <https://github.com/zalando/patroni/rel
 Community
 =========
 
-There are two places to connect with the Patroni community: `on github <https://github.com/zalando/patroni>`__, via Issues and PRs, and on channel #patroni in the `PostgreSQL Slack <https://postgres-slack.herokuapp.com/>`__.  If you're using Patroni, or just interested, please join us.
+There are two places to connect with the Patroni community: `on github <https://github.com/zalando/patroni>`__, via Issues and PRs, and on channel `#patroni <https://postgresteam.slack.com/archives/C9XPYG92A>`__ in the `PostgreSQL Slack <https://postgresteam.slack.com/>`__.  If you're using Patroni, or just interested, please join us.
 
 ===================================
 Technical Requirements/Installation
@@ -59,7 +63,7 @@ To install requirements on a Mac, run the following:
 
     brew install postgresql etcd haproxy libyaml python
 
-**Psycopg2**
+**Psycopg**
 
 Starting from `psycopg2-2.8 <http://initd.org/psycopg/articles/2019/04/04/psycopg-28-released/>`__ the binary version of psycopg2 will no longer be installed by default. Installing it from the source code requires C compiler and postgres+python dev packages.
 Since in the python world it is not possible to specify dependency as ``psycopg2 OR psycopg2-binary`` you will have to decide how to install it.
@@ -86,6 +90,12 @@ There are a few options available:
 
     pip install psycopg2>=2.5.4
 
+4. Use psycopg 3.0 instead of psycopg2
+
+::
+
+    pip install psycopg[binary]
+
 **General installation for pip**
 
 Patroni can be installed with pip:
@@ -96,7 +106,7 @@ Patroni can be installed with pip:
 
 where dependencies can be either empty, or consist of one or more of the following:
 
-etcd
+etcd or etcd3
     `python-etcd` module in order to use Etcd as DCS
 consul
     `python-consul` module in order to use Consul as DCS
@@ -106,6 +116,8 @@ exhibitor
     `kazoo` module in order to use Exhibitor as DCS (same dependencies as for Zookeeper)
 kubernetes
     `kubernetes` module in order to use Kubernetes as DCS in Patroni
+raft
+    `pysyncobj` module in order to use python Raft implementation as DCS
 aws
     `boto` in order to use AWS callbacks
 
@@ -115,7 +127,7 @@ For example, the command in order to install Patroni together with dependencies
 
     pip install patroni[etcd,aws]
 
-Note that external tools to call in the replica creation or custom bootstap scripts (i.e. WAL-E) should be installed independently of Patroni.
+Note that external tools to call in the replica creation or custom bootstrap scripts (i.e. WAL-E) should be installed independently of Patroni.
 
 =======================
 Running and Configuring
@@ -124,7 +136,7 @@ Running and Configuring
 To get started, do the following from different terminals:
 ::
 
-    > etcd --data-dir=data/etcd
+    > etcd --data-dir=data/etcd --enable-v2=true
     > ./patroni.py postgres0.yml
     > ./patroni.py postgres1.yml
 
@@ -167,7 +179,7 @@ Applications Should Not Use Superusers
 
 When connecting from an application, always use a non-superuser. Patroni requires access to the database to function properly. By using a superuser from an application, you can potentially use the entire connection pool, including the connections reserved for superusers, with the ``superuser_reserved_connections`` setting. If Patroni cannot access the Primary because the connection pool is full, behavior will be undesirable.
 
-.. |Build Status| image:: https://travis-ci.org/zalando/patroni.svg?branch=master
+.. |Tests Status| image:: https://github.com/zalando/patroni/actions/workflows/tests.yaml/badge.svg
-   :target: https://travis-ci.org/zalando/patroni
+   :target: https://github.com/zalando/patroni/actions/workflows/tests.yaml?query=branch%3Amaster
 .. |Coverage Status| image:: https://coveralls.io/repos/zalando/patroni/badge.svg?branch=master
-   :target: https://coveralls.io/r/zalando/patroni?branch=master
+   :target: https://coveralls.io/github/zalando/patroni?branch=master

TODO.md

@@ -1,12 +0,0 @@
-Failover
-========
-- When determining who should become master, include the minor version of PostgreSQL in the decision.
-
-Configuration
-==============
-- Provide a way to change pg_hba.conf of a running cluster on the Patroni level, without changing individual nodes.
-- Provide hooks to store and retrieve cluster-wide passwords without exposing them in a plain-text form to unauthorized users.
-
-Documentation
-==============
-- Document how to run cascading replication and possibly initialize the cluster without an access to the master node.

docker-compose-citus.yml

@@ -0,0 +1,139 @@
+# docker compose file for running a Citus cluster
+# with 3-node etcd v3 cluster as the DCS and one haproxy node.
+# The Citus cluster has a coordinator (3 nodes)
+# and two worker clusters (2 nodes).
+#
+# Before starting it up you need to build the docker image:
+# $ docker build -f Dockerfile.citus -t patroni-citus .
+# The cluster could be started as:
+# $ docker-compose -f docker-compose-citus.yml up -d
+# You can read more about it in the:
+# https://github.com/zalando/patroni/blob/master/docker/README.md#citus-cluster
+version: "2"
+
+networks:
+    demo:
+
+services:
+    etcd1: &etcd
+        image: patroni-citus
+        networks: [ demo ]
+        environment:
+            ETCDCTL_API: 3
+            ETCD_LISTEN_PEER_URLS: http://0.0.0.0:2380
+            ETCD_LISTEN_CLIENT_URLS: http://0.0.0.0:2379
+            ETCD_INITIAL_CLUSTER: etcd1=http://etcd1:2380,etcd2=http://etcd2:2380,etcd3=http://etcd3:2380
+            ETCD_INITIAL_CLUSTER_STATE: new
+            ETCD_INITIAL_CLUSTER_TOKEN: tutorial
+        container_name: demo-etcd1
+        hostname: etcd1
+        command: etcd -name etcd1 -initial-advertise-peer-urls http://etcd1:2380
+
+    etcd2:
+        <<: *etcd
+        container_name: demo-etcd2
+        hostname: etcd2
+        command: etcd -name etcd2 -initial-advertise-peer-urls http://etcd2:2380
+
+    etcd3:
+        <<: *etcd
+        container_name: demo-etcd3
+        hostname: etcd3
+        command: etcd -name etcd3 -initial-advertise-peer-urls http://etcd3:2380
+
+    haproxy:
+        image: patroni-citus
+        networks: [ demo ]
+        env_file: docker/patroni.env
+        hostname: haproxy
+        container_name: demo-haproxy
+        ports:
+            - "5000:5000"  # Access to the coorinator primary
+            - "5001:5001"  # Load-balancing across workers primaries
+        command: haproxy
+        environment: &haproxy_env
+            ETCDCTL_API: 3
+            ETCDCTL_ENDPOINTS: http://etcd1:2379,http://etcd2:2379,http://etcd3:2379
+            PATRONI_ETCD3_HOSTS: "'etcd1:2379','etcd2:2379','etcd3:2379'"
+            PATRONI_SCOPE: demo
+            PATRONI_CITUS_GROUP: 0
+            PATRONI_CITUS_DATABASE: citus
+            PGSSLMODE: verify-ca
+            PGSSLKEY: /etc/ssl/private/ssl-cert-snakeoil.key
+            PGSSLCERT: /etc/ssl/certs/ssl-cert-snakeoil.pem
+            PGSSLROOTCERT: /etc/ssl/certs/ssl-cert-snakeoil.pem
+
+    coord1:
+        image: patroni-citus
+        networks: [ demo ]
+        env_file: docker/patroni.env
+        hostname: coord1
+        container_name: demo-coord1
+        environment: &coord_env
+            <<: *haproxy_env
+            PATRONI_NAME: coord1
+            PATRONI_CITUS_GROUP: 0
+
+    coord2:
+        image: patroni-citus
+        networks: [ demo ]
+        env_file: docker/patroni.env
+        hostname: coord2
+        container_name: demo-coord2
+        environment:
+            <<: *coord_env
+            PATRONI_NAME: coord2
+
+    coord3:
+        image: patroni-citus
+        networks: [ demo ]
+        env_file: docker/patroni.env
+        hostname: coord3
+        container_name: demo-coord3
+        environment:
+            <<: *coord_env
+            PATRONI_NAME: coord3
+
+
+    work1-1:
+        image: patroni-citus
+        networks: [ demo ]
+        env_file: docker/patroni.env
+        hostname: work1-1
+        container_name: demo-work1-1
+        environment: &work1_env
+            <<: *haproxy_env
+            PATRONI_NAME: work1-1
+            PATRONI_CITUS_GROUP: 1
+
+    work1-2:
+        image: patroni-citus
+        networks: [ demo ]
+        env_file: docker/patroni.env
+        hostname: work1-2
+        container_name: demo-work1-2
+        environment:
+            <<: *work1_env
+            PATRONI_NAME: work1-2
+
+
+    work2-1:
+        image: patroni-citus
+        networks: [ demo ]
+        env_file: docker/patroni.env
+        hostname: work2-1
+        container_name: demo-work2-1
+        environment: &work2_env
+            <<: *haproxy_env
+            PATRONI_NAME: work2-1
+            PATRONI_CITUS_GROUP: 2
+
+    work2-2:
+        image: patroni-citus
+        networks: [ demo ]
+        env_file: docker/patroni.env
+        hostname: work2-2
+        container_name: demo-work2-2
+        environment:
+            <<: *work2_env
+            PATRONI_NAME: work2-2

docker-compose.yml

@@ -1,62 +1,43 @@
 # docker compose file for running a 3-node PostgreSQL cluster
 # with 3-node etcd cluster as the DCS and one haproxy node
+#
+# requires a patroni image build from the Dockerfile:
+# $ docker build -t patroni .
+# The cluster could be started as:
+# $ docker-compose up -d
+# You can read more about it in the:
+# https://github.com/zalando/patroni/blob/master/docker/README.md
 version: "2"
 
 networks:
     demo:
 
 services:
-    etcd1:
+    etcd1: &etcd
         image: patroni
         networks: [ demo ]
-        env_file: docker/etcd.env
+        environment:
+            ETCD_LISTEN_PEER_URLS: http://0.0.0.0:2380
+            ETCD_LISTEN_CLIENT_URLS: http://0.0.0.0:2379
+            ETCD_INITIAL_CLUSTER: etcd1=http://etcd1:2380,etcd2=http://etcd2:2380,etcd3=http://etcd3:2380
+            ETCD_INITIAL_CLUSTER_STATE: new
+            ETCD_INITIAL_CLUSTER_TOKEN: tutorial
         container_name: demo-etcd1
         hostname: etcd1
         command: etcd -name etcd1 -initial-advertise-peer-urls http://etcd1:2380
 
     etcd2:
-        image: patroni
+        <<: *etcd
-        networks: [ demo ]
-        env_file: docker/etcd.env
         container_name: demo-etcd2
         hostname: etcd2
         command: etcd -name etcd2 -initial-advertise-peer-urls http://etcd2:2380
 
     etcd3:
-        image: patroni
+        <<: *etcd
-        networks: [ demo ]
-        env_file: docker/etcd.env
         container_name: demo-etcd3
         hostname: etcd3
         command: etcd -name etcd3 -initial-advertise-peer-urls http://etcd3:2380
 
-    patroni1:
-        image: patroni
-        networks: [ demo ]
-        env_file: docker/patroni.env
-        hostname: patroni1
-        container_name: demo-patroni1
-        environment:
-            PATRONI_NAME: patroni1
-
-    patroni2:
-        image: patroni
-        networks: [ demo ]
-        env_file: docker/patroni.env
-        hostname: patroni2
-        container_name: demo-patroni2
-        environment:
-            PATRONI_NAME: patroni2
-
-    patroni3:
-        image: patroni
-        networks: [ demo ]
-        env_file: docker/patroni.env
-        hostname: patroni3
-        container_name: demo-patroni3
-        environment:
-            PATRONI_NAME: patroni3
-
     haproxy:
         image: patroni
         networks: [ demo ]
@@ -67,3 +48,37 @@ services:
             - "5000:5000"
             - "5001:5001"
         command: haproxy
+        environment: &haproxy_env
+            ETCDCTL_ENDPOINTS: http://etcd1:2379,http://etcd2:2379,http://etcd3:2379
+            PATRONI_ETCD3_HOSTS: "'etcd1:2379','etcd2:2379','etcd3:2379'"
+            PATRONI_SCOPE: demo
+
+    patroni1:
+        image: patroni
+        networks: [ demo ]
+        env_file: docker/patroni.env
+        hostname: patroni1
+        container_name: demo-patroni1
+        environment:
+            <<: *haproxy_env
+            PATRONI_NAME: patroni1
+
+    patroni2:
+        image: patroni
+        networks: [ demo ]
+        env_file: docker/patroni.env
+        hostname: patroni2
+        container_name: demo-patroni2
+        environment:
+            <<: *haproxy_env
+            PATRONI_NAME: patroni2
+
+    patroni3:
+        image: patroni
+        networks: [ demo ]
+        env_file: docker/patroni.env
+        hostname: patroni3
+        container_name: demo-patroni3
+        environment:
+            <<: *haproxy_env
+            PATRONI_NAME: patroni3

docker/README.md

@@ -1,10 +1,10 @@
-# Patroni Dockerfile
+# Dockerfile and Dockerfile.citus
-You can run Patroni in a docker container using this Dockerfile
+You can run Patroni in a docker container using these Dockerfiles
 
-This Dockerfile is meant in aiding development of Patroni and quick testing of features. It is not a production-worthy
+They are meant in aiding development of Patroni and quick testing of features and not a production-worthy!
-Dockerfile
 
     docker build -t patroni .
+    docker build -f Dockerfile.citus -t patroni-citus .
 
 # Examples
 
@@ -12,7 +12,10 @@ Dockerfile
 
     docker run -d patroni
 
-## Three-node Patroni cluster with three-node etcd cluster and one haproxy container using docker-compose
+## Three-node Patroni cluster
+
+In addition to three Patroni containers the stack starts three containers with etcd (forming a three-node cluster), and one container with haproxy.
+The haproxy listens on ports 5000 (connects to the primary) and 5001 (does load-balancing between healthy standbys).
 
 Example session:
 
@@ -92,7 +95,8 @@ Example session:
     b2e169fcb8a34028: name=etcd1 peerURLs=http://etcd1:2380 clientURLs=http://etcd1:2379 isLeader=false
     postgres@patroni1:~$ exit
 
-    $ psql -h localhost -p 5000 -U postgres -W
+    $ docker exec -ti demo-haproxy bash
+    postgres@haproxy:~$ psql -h localhost -p 5000 -U postgres -W
     Password: postgres
     psql (11.2 (Ubuntu 11.2-1.pgdg18.04+1), server 10.7 (Debian 10.7-1.pgdg90+1))
     Type "help" for help.
@@ -105,7 +109,7 @@ Example session:
 
     localhost/postgres=# \q
 
-    $ psql -h localhost -p 5001 -U postgres -W
+    $postgres@haproxy:~ psql -h localhost -p 5001 -U postgres -W
     Password: postgres
     psql (11.2 (Ubuntu 11.2-1.pgdg18.04+1), server 10.7 (Debian 10.7-1.pgdg90+1))
     Type "help" for help.
@@ -115,3 +119,188 @@ Example session:
     ───────────────────
      t
     (1 row)
+
+## Citus cluster
+
+The stack starts three containers with etcd (forming a three-node etcd cluster), seven containers with Patroni+PostgreSQL+Citus (three coordinator nodes, and two worker clusters with two nodes each), and one container with haproxy.
+The haproxy listens on ports 5000 (connects to the coordinator primary) and 5001 (does load-balancing between worker primary nodes).
+
+Example session:
+
+    $ docker-compose -f docker-compose-citus.yml up -d
+    Creating demo-work2-1 ... done
+    Creating demo-work1-1 ... done
+    Creating demo-etcd2   ... done
+    Creating demo-etcd1   ... done
+    Creating demo-coord3  ... done
+    Creating demo-etcd3   ... done
+    Creating demo-coord1  ... done
+    Creating demo-haproxy ... done
+    Creating demo-work2-2 ... done
+    Creating demo-coord2  ... done
+    Creating demo-work1-2 ... done
+
+    $ docker ps
+    CONTAINER ID   IMAGE                  COMMAND                  CREATED         STATUS         PORTS                              NAMES
+    852d8885a612   patroni-citus          "/bin/sh /entrypoint…"   6 seconds ago   Up 3 seconds                                      demo-coord3
+    cdd692f947ab   patroni-citus          "/bin/sh /entrypoint…"   6 seconds ago   Up 3 seconds                                      demo-work1-2
+    9f4e340b36da   patroni-citus          "/bin/sh /entrypoint…"   6 seconds ago   Up 3 seconds                                      demo-etcd3
+    d69c129a960a   patroni-citus          "/bin/sh /entrypoint…"   6 seconds ago   Up 4 seconds                                      demo-etcd1
+    c5849689b8cd   patroni-citus          "/bin/sh /entrypoint…"   6 seconds ago   Up 4 seconds                                      demo-coord1
+    c9d72bd6217d   patroni-citus          "/bin/sh /entrypoint…"   6 seconds ago   Up 3 seconds                                      demo-work2-1
+    24b1b43efa05   patroni-citus          "/bin/sh /entrypoint…"   6 seconds ago   Up 4 seconds                                      demo-coord2
+    cb0cc2b4ca0a   patroni-citus          "/bin/sh /entrypoint…"   6 seconds ago   Up 3 seconds                                      demo-work2-2
+    9796c6b8aad5   patroni-citus          "/bin/sh /entrypoint…"   6 seconds ago   Up 5 seconds                                      demo-work1-1
+    8baccd74dcae   patroni-citus          "/bin/sh /entrypoint…"   6 seconds ago   Up 4 seconds                                      demo-etcd2
+    353ec62a0187   patroni-citus          "/bin/sh /entrypoint…"   6 seconds ago   Up 4 seconds   0.0.0.0:5000-5001->5000-5001/tcp   demo-haproxy
+
+    $ docker logs demo-coord1
+    2023-01-05 15:09:31,295 INFO: Selected new etcd server http://172.27.0.4:2379
+    2023-01-05 15:09:31,388 INFO: Lock owner: None; I am coord1
+    2023-01-05 15:09:31,501 INFO: trying to bootstrap a new cluster
+    ...
+    2023-01-05 15:09:45,096 INFO: postmaster pid=39
+    localhost:5432 - no response
+    2023-01-05 15:09:45.137 UTC [39] LOG:  starting PostgreSQL 15.1 (Debian 15.1-1.pgdg110+1) on x86_64-pc-linux-gnu, compiled by gcc (Debian 10.2.1-6) 10.2.1 20210110, 64-bit
+    2023-01-05 15:09:45.137 UTC [39] LOG:  listening on IPv4 address "0.0.0.0", port 5432
+    2023-01-05 15:09:45.152 UTC [39] LOG:  listening on Unix socket "/var/run/postgresql/.s.PGSQL.5432"
+    2023-01-05 15:09:45.177 UTC [43] LOG:  database system was shut down at 2023-01-05 15:09:32 UTC
+    2023-01-05 15:09:45.193 UTC [39] LOG:  database system is ready to accept connections
+    localhost:5432 - accepting connections
+    localhost:5432 - accepting connections
+    2023-01-05 15:09:46,139 INFO: establishing a new patroni connection to the postgres cluster
+    2023-01-05 15:09:46,208 INFO: running post_bootstrap
+    2023-01-05 15:09:47.209 UTC [55] LOG:  starting maintenance daemon on database 16386 user 10
+    2023-01-05 15:09:47.209 UTC [55] CONTEXT:  Citus maintenance daemon for database 16386 user 10
+    2023-01-05 15:09:47,215 WARNING: Could not activate Linux watchdog device: "Can't open watchdog device: [Errno 2] No such file or directory: '/dev/watchdog'"
+    2023-01-05 15:09:47.446 UTC [41] LOG:  checkpoint starting: immediate force wait
+    2023-01-05 15:09:47,466 INFO: initialized a new cluster
+    2023-01-05 15:09:47,594 DEBUG: query(SELECT nodeid, groupid, nodename, nodeport, noderole FROM pg_catalog.pg_dist_node WHERE noderole = 'primary', ())
+    2023-01-05 15:09:47,594 INFO: establishing a new patroni connection to the postgres cluster
+    2023-01-05 15:09:47,467 INFO: Lock owner: coord1; I am coord1
+    2023-01-05 15:09:47,613 DEBUG: query(SELECT pg_catalog.citus_set_coordinator_host(%s, %s, 'primary', 'default'), ('172.27.0.6', 5432))
+    2023-01-05 15:09:47,924 INFO: no action. I am (coord1), the leader with the lock
+    2023-01-05 15:09:51.282 UTC [41] LOG:  checkpoint complete: wrote 1086 buffers (53.0%); 0 WAL file(s) added, 0 removed, 0 recycled; write=0.029 s, sync=3.746 s, total=3.837 s; sync files=280, longest=0.028 s, average=0.014 s; distance=8965 kB, estimate=8965 kB
+    2023-01-05 15:09:51.283 UTC [41] LOG:  checkpoint starting: immediate force wait
+    2023-01-05 15:09:51.495 UTC [41] LOG:  checkpoint complete: wrote 18 buffers (0.9%); 0 WAL file(s) added, 0 removed, 0 recycled; write=0.044 s, sync=0.091 s, total=0.212 s; sync files=15, longest=0.015 s, average=0.007 s; distance=67 kB, estimate=8076 kB
+    2023-01-05 15:09:57,467 INFO: Lock owner: coord1; I am coord1
+    2023-01-05 15:09:57,569 INFO: Assigning synchronous standby status to ['coord3']
+    server signaled
+    2023-01-05 15:09:57.574 UTC [39] LOG:  received SIGHUP, reloading configuration files
+    2023-01-05 15:09:57.580 UTC [39] LOG:  parameter "synchronous_standby_names" changed to "coord3"
+    2023-01-05 15:09:59,637 INFO: Synchronous standby status assigned to ['coord3']
+    2023-01-05 15:09:59,638 DEBUG: query(SELECT pg_catalog.citus_add_node(%s, %s, %s, 'primary', 'default'), ('172.27.0.2', 5432, 1))
+    2023-01-05 15:09:59.690 UTC [67] LOG:  standby "coord3" is now a synchronous standby with priority 1
+    2023-01-05 15:09:59.690 UTC [67] STATEMENT:  START_REPLICATION SLOT "coord3" 0/3000000 TIMELINE 1
+    2023-01-05 15:09:59,694 INFO: no action. I am (coord1), the leader with the lock
+    2023-01-05 15:09:59,704 DEBUG: query(SELECT pg_catalog.citus_add_node(%s, %s, %s, 'primary', 'default'), ('172.27.0.8', 5432, 2))
+    2023-01-05 15:10:07,625 INFO: no action. I am (coord1), the leader with the lock
+    2023-01-05 15:10:17,579 INFO: no action. I am (coord1), the leader with the lock
+
+    $ docker exec -ti demo-haproxy bash
+    postgres@haproxy:~$ etcdctl member list
+    1bab629f01fa9065, started, etcd3, http://etcd3:2380, http://172.27.0.10:2379
+    8ecb6af518d241cc, started, etcd2, http://etcd2:2380, http://172.27.0.4:2379
+    b2e169fcb8a34028, started, etcd1, http://etcd1:2380, http://172.27.0.7:2379
+
+    postgres@haproxy:~$ etcdctl get --keys-only --prefix /service/demo
+    /service/demo/0/config
+    /service/demo/0/initialize
+    /service/demo/0/leader
+    /service/demo/0/members/coord1
+    /service/demo/0/members/coord2
+    /service/demo/0/members/coord3
+    /service/demo/0/status
+    /service/demo/0/sync
+    /service/demo/1/config
+    /service/demo/1/initialize
+    /service/demo/1/leader
+    /service/demo/1/members/work1-1
+    /service/demo/1/members/work1-2
+    /service/demo/1/status
+    /service/demo/1/sync
+    /service/demo/2/config
+    /service/demo/2/initialize
+    /service/demo/2/leader
+    /service/demo/2/members/work2-1
+    /service/demo/2/members/work2-2
+    /service/demo/2/status
+    /service/demo/2/sync
+
+    postgres@haproxy:~$ psql -h localhost -p 5000 -U postgres -d citus
+    Password for user postgres: postgres
+    psql (15.1 (Debian 15.1-1.pgdg110+1))
+    SSL connection (protocol: TLSv1.3, cipher: TLS_AES_256_GCM_SHA384, compression: off)
+    Type "help" for help.
+
+    citus=# select pg_is_in_recovery();
+     pg_is_in_recovery
+    -------------------
+     f
+    (1 row)
+
+    citus=# table pg_dist_node;
+     nodeid | groupid |  nodename  | nodeport | noderack | hasmetadata | isactive | noderole | nodecluster | metadatasynced | shouldhaveshards
+    --------+---------+------------+----------+----------+-------------+----------+----------+-------------+----------------+------------------
+          1 |       0 | 172.27.0.6 |     5432 | default  | t           | t        | primary  | default     | t              | f
+          2 |       1 | 172.27.0.2 |     5432 | default  | t           | t        | primary  | default     | t              | t
+          3 |       2 | 172.27.0.8 |     5432 | default  | t           | t        | primary  | default     | t              | t
+    (3 rows)
+
+    citus=# \q
+
+    postgres@haproxy:~$ patronictl list
+    + Citus cluster: demo ----------+--------------+---------+----+-----------+
+    | Group | Member  | Host        | Role         | State   | TL | Lag in MB |
+    +-------+---------+-------------+--------------+---------+----+-----------+
+    |     0 | coord1  | 172.27.0.6  | Leader       | running |  1 |           |
+    |     0 | coord2  | 172.27.0.5  | Replica      | running |  1 |         0 |
+    |     0 | coord3  | 172.27.0.9  | Sync Standby | running |  1 |         0 |
+    |     1 | work1-1 | 172.27.0.2  | Leader       | running |  1 |           |
+    |     1 | work1-2 | 172.27.0.12 | Sync Standby | running |  1 |         0 |
+    |     2 | work2-1 | 172.27.0.11 | Sync Standby | running |  1 |         0 |
+    |     2 | work2-2 | 172.27.0.8  | Leader       | running |  1 |           |
+    +-------+---------+-------------+--------------+---------+----+-----------+
+
+    postgres@haproxy:~$ patronictl switchover --group 2 --force
+    Current cluster topology
+    + Citus cluster: demo (group: 2, 7185185529556963355) +-----------+
+    | Member  | Host        | Role         | State   | TL | Lag in MB |
+    +---------+-------------+--------------+---------+----+-----------+
+    | work2-1 | 172.27.0.11 | Sync Standby | running |  1 |         0 |
+    | work2-2 | 172.27.0.8  | Leader       | running |  1 |           |
+    +---------+-------------+--------------+---------+----+-----------+
+    2023-01-05 15:29:29.54204 Successfully switched over to "work2-1"
+    + Citus cluster: demo (group: 2, 7185185529556963355) -------+
+    | Member  | Host        | Role    | State   | TL | Lag in MB |
+    +---------+-------------+---------+---------+----+-----------+
+    | work2-1 | 172.27.0.11 | Leader  | running |  1 |           |
+    | work2-2 | 172.27.0.8  | Replica | stopped |    |   unknown |
+    +---------+-------------+---------+---------+----+-----------+
+
+    postgres@haproxy:~$ patronictl list
+    + Citus cluster: demo ----------+--------------+---------+----+-----------+
+    | Group | Member  | Host        | Role         | State   | TL | Lag in MB |
+    +-------+---------+-------------+--------------+---------+----+-----------+
+    |     0 | coord1  | 172.27.0.6  | Leader       | running |  1 |           |
+    |     0 | coord2  | 172.27.0.5  | Replica      | running |  1 |         0 |
+    |     0 | coord3  | 172.27.0.9  | Sync Standby | running |  1 |         0 |
+    |     1 | work1-1 | 172.27.0.2  | Leader       | running |  1 |           |
+    |     1 | work1-2 | 172.27.0.12 | Sync Standby | running |  1 |         0 |
+    |     2 | work2-1 | 172.27.0.11 | Leader       | running |  2 |           |
+    |     2 | work2-2 | 172.27.0.8  | Sync Standby | running |  2 |         0 |
+    +-------+---------+-------------+--------------+---------+----+-----------+
+
+    postgres@haproxy:~$ psql -h localhost -p 5000 -U postgres -d citus
+    Password for user postgres: postgres
+    psql (15.1 (Debian 15.1-1.pgdg110+1))
+    SSL connection (protocol: TLSv1.3, cipher: TLS_AES_256_GCM_SHA384, compression: off)
+    Type "help" for help.
+
+    citus=# table pg_dist_node;
+     nodeid | groupid |  nodename   | nodeport | noderack | hasmetadata | isactive | noderole | nodecluster | metadatasynced | shouldhaveshards
+    --------+---------+-------------+----------+----------+-------------+----------+----------+-------------+----------------+------------------
+          1 |       0 | 172.27.0.6  |     5432 | default  | t           | t        | primary  | default     | t              | f
+          3 |       2 | 172.27.0.11 |     5432 | default  | t           | t        | primary  | default     | t              | t
+          2 |       1 | 172.27.0.2  |     5432 | default  | t           | t        | primary  | default     | t              | t
+    (3 rows)

docker/entrypoint.sh

@@ -7,29 +7,36 @@ if [ -f /a.tar.xz ]; then
     sudo ln -snf dash /bin/sh
 fi
 
-readonly PATRONI_SCOPE=${PATRONI_SCOPE:-batman}
+readonly PATRONI_SCOPE="${PATRONI_SCOPE:-batman}"
-PATRONI_NAMESPACE=${PATRONI_NAMESPACE:-/service}
+PATRONI_NAMESPACE="${PATRONI_NAMESPACE:-/service}"
-readonly PATRONI_NAMESPACE=${PATRONI_NAMESPACE%/}
+readonly PATRONI_NAMESPACE="${PATRONI_NAMESPACE%/}"
-readonly DOCKER_IP=$(hostname --ip-address)
+DOCKER_IP=$(hostname --ip-address)
+readonly DOCKER_IP
 
 case "$1" in
     haproxy)
         haproxy -f /etc/haproxy/haproxy.cfg -p /var/run/haproxy.pid -D
-        CONFD="confd -prefix=$PATRONI_NAMESPACE/$PATRONI_SCOPE -interval=10 -backend"
+        set -- confd "-prefix=$PATRONI_NAMESPACE/$PATRONI_SCOPE" -interval=10 -backend
-        if [ ! -z "$PATRONI_ZOOKEEPER_HOSTS" ]; then
+        if [ -n "$PATRONI_ZOOKEEPER_HOSTS" ]; then
-            while ! /usr/share/zookeeper/bin/zkCli.sh -server $PATRONI_ZOOKEEPER_HOSTS ls /; do
+            while ! /usr/share/zookeeper/bin/zkCli.sh -server "$PATRONI_ZOOKEEPER_HOSTS" ls /; do
                 sleep 1
             done
-            exec dumb-init $CONFD zookeeper -node $PATRONI_ZOOKEEPER_HOSTS
+            set -- "$@" zookeeper -node "$PATRONI_ZOOKEEPER_HOSTS"
         else
-            while ! etcdctl cluster-health 2> /dev/null; do
+            while ! etcdctl member list 2> /dev/null; do
                 sleep 1
             done
-            exec dumb-init $CONFD etcd -node $(echo $ETCDCTL_ENDPOINTS | sed 's/,/ -node /g')
+            set -- "$@" etcdv3
+            while IFS='' read -r line; do
+                set -- "$@" -node "$line"
+            done <<-EOT
+$(echo "$ETCDCTL_ENDPOINTS" | sed 's/,/\n/g')
+EOT
         fi
+        exec dumb-init "$@"
         ;;
     etcd)
-        exec "$@" -advertise-client-urls http://$DOCKER_IP:2379
+        exec "$@" -advertise-client-urls "http://$DOCKER_IP:2379"
         ;;
     zookeeper)
         exec /usr/share/zookeeper/bin/zkServer.sh start-foreground
@@ -37,7 +44,7 @@ case "$1" in
 esac
 
 ## We start an etcd
-if [ -z "$PATRONI_ETCD_HOSTS" ] && [ -z "$PATRONI_ZOOKEEPER_HOSTS" ]; then
+if [ -z "$PATRONI_ETCD3_HOSTS" ] && [ -z "$PATRONI_ZOOKEEPER_HOSTS" ]; then
     export PATRONI_ETCD_URL="http://127.0.0.1:2379"
     etcd --data-dir /tmp/etcd.data -advertise-client-urls=$PATRONI_ETCD_URL -listen-client-urls=http://0.0.0.0:2379 > /var/log/etcd.log 2> /var/log/etcd.err &
 fi
@@ -56,5 +63,13 @@ export PATRONI_REPLICATION_USERNAME="${PATRONI_REPLICATION_USERNAME:-replicator}
 export PATRONI_REPLICATION_PASSWORD="${PATRONI_REPLICATION_PASSWORD:-replicate}"
 export PATRONI_SUPERUSER_USERNAME="${PATRONI_SUPERUSER_USERNAME:-postgres}"
 export PATRONI_SUPERUSER_PASSWORD="${PATRONI_SUPERUSER_PASSWORD:-postgres}"
+export PATRONI_REPLICATION_SSLMODE="${PATRONI_REPLICATION_SSLMODE:-$PGSSLMODE}"
+export PATRONI_REPLICATION_SSLKEY="${PATRONI_REPLICATION_SSLKEY:-$PGSSLKEY}"
+export PATRONI_REPLICATION_SSLCERT="${PATRONI_REPLICATION_SSLCERT:-$PGSSLCERT}"
+export PATRONI_REPLICATION_SSLROOTCERT="${PATRONI_REPLICATION_SSLROOTCERT:-$PGSSLROOTCERT}"
+export PATRONI_SUPERUSER_SSLMODE="${PATRONI_SUPERUSER_SSLMODE:-$PGSSLMODE}"
+export PATRONI_SUPERUSER_SSLKEY="${PATRONI_SUPERUSER_SSLKEY:-$PGSSLKEY}"
+export PATRONI_SUPERUSER_SSLCERT="${PATRONI_SUPERUSER_SSLCERT:-$PGSSLCERT}"
+export PATRONI_SUPERUSER_SSLROOTCERT="${PATRONI_SUPERUSER_SSLROOTCERT:-$PGSSLROOTCERT}"
 
 exec python3 /patroni.py postgres0.yml

docker/etcd.env

@@ -1,5 +0,0 @@
-ETCD_LISTEN_PEER_URLS=http://0.0.0.0:2380
-ETCD_LISTEN_CLIENT_URLS=http://0.0.0.0:2379
-ETCD_INITIAL_CLUSTER=etcd1=http://etcd1:2380,etcd2=http://etcd2:2380,etcd3=http://etcd3:2380
-ETCD_INITIAL_CLUSTER_STATE=new
-ETCD_INITIAL_CLUSTER_TOKEN=tutorial

docker/patroni.env

@@ -1,6 +1,3 @@
-PATRONI_SCOPE=demo
-PATRONI_ETCD_HOSTS='etcd1:2379','etcd2:2379','etcd3:2379'
-
 PATRONI_RESTAPI_USERNAME=admin
 PATRONI_RESTAPI_PASSWORD=admin
 PATRONI_SUPERUSER_USERNAME=postgres
@@ -9,6 +6,3 @@ PATRONI_REPLICATION_USERNAME=replicator
 PATRONI_REPLICATION_PASSWORD=replicate
 PATRONI_admin_PASSWORD=admin
 PATRONI_admin_OPTIONS=createdb,createrole
-
-# for etcdctl
-ETCDCTL_ENDPOINTS=http://etcd1:2379,http://etcd2:2379,http://etcd3:2379

docs/CONTRIBUTING.rst

@@ -8,7 +8,40 @@ Wanna contribute to Patroni? Yay - here is how!
 Chatting
 --------
 
-Just want to chat with other Patroni users?  Looking for interactive troubleshooting help? Join us on channel #patroni in the `PostgreSQL Slack <https://postgres-slack.herokuapp.com/>`__.
+Just want to chat with other Patroni users?  Looking for interactive troubleshooting help? Join us on channel `#patroni <https://postgresteam.slack.com/archives/C9XPYG92A>`__ in the `PostgreSQL Slack <https://postgresteam.slack.com/>`__.
+
+Running tests
+-------------
+
+Requirements for running behave tests:
+
+1. PostgreSQL packages need to be installed.
+2. PostgreSQL binaries must be available in your `PATH`. You may need to add them to the path with something like `PATH=/usr/lib/postgresql/11/bin:$PATH python -m behave`.
+3. If you'd like to test with external DCSs (e.g., Etcd, Consul, and Zookeeper) you'll need the packages installed and respective services running and accepting unencrypted/unprotected connections on localhost and default port. In the case of Etcd or Consul, the behave test suite could start them up if binaries are available in the `PATH`.
+
+ Install dependencies:
+
+.. code-block:: bash
+
+    # You may want to use Virtualenv or specify pip3.
+    pip install -r requirements.txt
+    pip install -r requirements.dev.txt
+
+After you have all dependencies installed, you can run the various test suites:
+
+.. code-block:: bash
+
+    # You may want to use Virtualenv or specify python3.
+
+    # Run flake8 to check syntax and formatting:
+    python setup.py flake8
+
+    # Run the pytest suite in tests/:
+    python setup.py test
+
+    # Run the behave (https://behave.readthedocs.io/en/latest/) test suite in features/;
+    # modify DCS as desired (raft has no dependencies so is the easiest to start with):
+    DCS=raft python -m behave
 
 Reporting issues
 ----------------

docs/ENVIRONMENT.rst

@@ -11,7 +11,11 @@ Global/Universal
 -  **PATRONI\_NAME**: name of the node where the current instance of Patroni is running. Must be unique for the cluster.
 -  **PATRONI\_NAMESPACE**: path within the configuration store where Patroni will keep information about the cluster. Default value: "/service"
 -  **PATRONI\_SCOPE**: cluster name
+
+Log
+---
 -  **PATRONI\_LOG\_LEVEL**: sets the general logging level. Default value is **INFO** (see `the docs for Python logging <https://docs.python.org/3.6/library/logging.html#levels>`_)
+-  **PATRONI\_LOG\_TRACEBACK\_LEVEL**: sets the level where tracebacks will be visible. Default value is **ERROR**. Set it to **DEBUG** if you want to see tracebacks only if you enable **PATRONI\_LOG\_LEVEL=DEBUG**.
 -  **PATRONI\_LOG\_FORMAT**: sets the log formatting string. Default value is **%(asctime)s %(levelname)s: %(message)s** (see `the LogRecord attributes <https://docs.python.org/3.6/library/logging.html#logrecord-attributes>`_)
 -  **PATRONI\_LOG\_DATEFORMAT**: sets the datetime formatting string. (see the `formatTime() documentation <https://docs.python.org/3.6/library/logging.html#logging.Formatter.formatTime>`_)
 -  **PATRONI\_LOG\_MAX\_QUEUE\_SIZE**: Patroni is using two-step logging. Log records are written into the in-memory queue and there is a separate thread which pulls them from the queue and writes to stderr or file. The maximum size of the internal queue is limited by default by **1000** records, which is enough to keep logs for the past 1h20m.
@@ -19,8 +23,6 @@ Global/Universal
 -  **PATRONI\_LOG\_FILE\_NUM**: The number of application logs to retain.
 -  **PATRONI\_LOG\_FILE\_SIZE**: Size of patroni.log file (in bytes) that triggers a log rolling.
 -  **PATRONI\_LOG\_LOGGERS**: Redefine logging level per python module. Example ``PATRONI_LOG_LOGGERS="{patroni.postmaster: WARNING, urllib3: DEBUG}"``
--  **PATRONI\_DEBUG\_MODE**: When set to a non-empty value, makes Patroni run in the special debug mode that enables stepping with a debugger through some execution paths within Patroni `
-Example ``PATRONI_DEBUG_MODE="on"``
 
 Bootstrap configuration
 -----------------------
@@ -31,10 +33,17 @@ It is possible to create new database users right after the successful initializ
 
 Example: defining ``PATRONI_admin_PASSWORD=strongpasswd`` and ``PATRONI_admin_OPTIONS='createrole,createdb'`` will cause creation of the user **admin** with the password **strongpasswd** that is allowed to create other users and databases.
 
+Citus
+-----
+Enables integration Patroni with `Citus <https://docs.citusdata.com>`__. If configured, Patroni will take care of registering Citus worker nodes on the coordinator. You can find more information about Citus support :ref:`here <citus>`.
+
+-  **PATRONI\_CITUS\_GROUP**: the Citus group id, integer. Use ``0`` for coordinator and ``1``, ``2``, etc... for workers
+-  **PATRONI\_CITUS\_DATABASE**: the database where ``citus`` extension should be created. Must be the same on the coordinator and all workers. Currently only one database is supported.
+
 Consul
 ------
--  **PATRONI\_CONSUL\_HOST**: the host:port for the Consul endpoint.
+-  **PATRONI\_CONSUL\_HOST**: the host:port for the Consul local agent.
--  **PATRONI\_CONSUL\_URL**: url for the Consul, in format: http(s)://host:port
+-  **PATRONI\_CONSUL\_URL**: url for the Consul local agent, in format: http(s)://host:port
 -  **PATRONI\_CONSUL\_PORT**: (optional) Consul port
 -  **PATRONI\_CONSUL\_SCHEME**: (optional) **http** or **https**, defaults to **http**
 -  **PATRONI\_CONSUL\_TOKEN**: (optional) ACL token
@@ -44,9 +53,11 @@ Consul
 -  **PATRONI\_CONSUL\_KEY**: (optional) File with the client key. Can be empty if the key is part of certificate.
 -  **PATRONI\_CONSUL\_DC**: (optional) Datacenter to communicate with. By default the datacenter of the host is used.
 -  **PATRONI\_CONSUL\_CONSISTENCY**: (optional) Select consul consistency mode. Possible values are ``default``, ``consistent``, or ``stale`` (more details in `consul API reference <https://www.consul.io/api/features/consistency.html/>`__)
--  **PATRONI\_CONSUL\_CHECKS**: (optional) list of Consul health checks used for the session. If not specified Consul will use "serfHealth" in additional to the TTL based check created by Patroni. Additional checks, in particular the "serfHealth", may cause the leader lock to expire faster than in `ttl` seconds when the leader instance becomes unavailable.
+-  **PATRONI\_CONSUL\_CHECKS**: (optional) list of Consul health checks used for the session. By default an empty list is used.
--  **PATRONI\_CONSUL\_REGISTER\_SERVICE**: (optional) whether or not to register a service with the name defined by the scope parameter and the tag master, replica or standby-leader depending on the node's role. Defaults to **false**
+-  **PATRONI\_CONSUL\_REGISTER\_SERVICE**: (optional) whether or not to register a service with the name defined by the scope parameter and the tag master, primary, replica, or standby-leader depending on the node's role. Defaults to **false**
+-  **PATRONI\_CONSUL\_SERVICE\_TAGS**: (optional) additional static tags to add to the Consul service apart from the role (``master``/``primary``/``replica``/``standby-leader``). By default an empty list is used.
 -  **PATRONI\_CONSUL\_SERVICE\_CHECK\_INTERVAL**: (optional) how often to perform health check against registered url
+-  **PATRONI\_CONSUL\_SERVICE\_CHECK\_TLS\_SERVER\_NAME**: (optional) overide SNI host when connecting via TLS, see also `consul agent check API reference <https://www.consul.io/api-docs/agent/check#tlsservername>`__.
 
 Etcd
 ----
@@ -57,13 +68,37 @@ Etcd
 -  **PATRONI\_ETCD\_USE\_PROXIES**: If this parameter is set to true, Patroni will consider **hosts** as a list of proxies and will not perform a topology discovery of etcd cluster but stick to a fixed list of **hosts**.
 -  **PATRONI\_ETCD\_PROTOCOL**: http or https, if not specified http is used. If the **url** or **proxy** is specified - will take protocol from them.
 -  **PATRONI\_ETCD\_HOST**: the host:port for the etcd endpoint.
--  **PATRONI\_ETCD\_SRV**: Domain to search the SRV record(s) for cluster autodiscovery.
+-  **PATRONI\_ETCD\_SRV**: Domain to search the SRV record(s) for cluster autodiscovery. Patroni will try to query these SRV service names for specified domain (in that order until first success): ``_etcd-client-ssl``, ``_etcd-client``, ``_etcd-ssl``, ``_etcd``, ``_etcd-server-ssl``, ``_etcd-server``. If SRV records for ``_etcd-server-ssl`` or ``_etcd-server`` are retrieved then ETCD peer protocol is used do query ETCD for available members. Otherwise hosts from SRV records will be used.
+-  **PATRONI\_ETCD\_SRV\_SUFFIX**: Configures a suffix to the SRV name that is queried during discovery. Use this flag to differentiate between multiple etcd clusters under the same domain. Works only with conjunction with **PATRONI\_ETCD\_SRV**. For example, if ``PATRONI_ETCD_SRV_SUFFIX=foo`` and ``PATRONI_ETCD_SRV=example.org`` are set, the following DNS SRV query is made:``_etcd-client-ssl-foo._tcp.example.com`` (and so on for every possible ETCD SRV service name).
 -  **PATRONI\_ETCD\_USERNAME**: username for etcd authentication.
 -  **PATRONI\_ETCD\_PASSWORD**: password for etcd authentication.
 -  **PATRONI\_ETCD\_CACERT**: The ca certificate. If present it will enable validation.
 -  **PATRONI\_ETCD\_CERT**: File with the client certificate.
 -  **PATRONI\_ETCD\_KEY**: File with the client key. Can be empty if the key is part of certificate.
 
+Etcdv3
+------
+Environment names for Etcdv3 are similar as for Etcd, you just need to use ``ETCD3`` instead of ``ETCD`` in the variable name. Example: ``PATRONI_ETCD3_HOST``, ``PATRONI_ETCD3_CACERT``, and so on.
+
+.. warning::
+    Keys created with protocol version 2 are not visible with protocol version 3 and the other way around, therefore it is not possible to switch from Etcd to Etcdv3 just by updating Patroni configuration.
+
+
+ZooKeeper
+---------
+-  **PATRONI\_ZOOKEEPER\_HOSTS**: Comma separated list of ZooKeeper cluster members: "'host1:port1','host2:port2','etc...'". It is important to quote every single entity!
+-  **PATRONI\_ZOOKEEPER\_USE\_SSL**: (optional) Whether SSL is used or not. Defaults to ``false``. If set to ``false``, all SSL specific parameters are ignored.
+-  **PATRONI\_ZOOKEEPER\_CACERT**: (optional) The CA certificate. If present it will enable validation.
+-  **PATRONI\_ZOOKEEPER\_CERT**: (optional) File with the client certificate.
+-  **PATRONI\_ZOOKEEPER\_KEY**: (optional) File with the client key.
+-  **PATRONI\_ZOOKEEPER\_KEY\_PASSWORD**: (optional) The client key password.
+-  **PATRONI\_ZOOKEEPER\_VERIFY**: (optional) Whether to verify certificate or not. Defaults to ``true``.
+-  **PATRONI\_ZOOKEEPER\_SET\_ACLS**: (optional) If set, configure Kazoo to apply a default ACL to each ZNode that it creates. ACLs will assume 'x509' schema and should be specified as a dictionary with the principal as the key and one or more permissions as a list in the value.  Permissions may be one of ``CREATE``, ``READ``, ``WRITE``, ``DELETE`` or ``ADMIN``.  For example, ``set_acls: {CN=principal1: [CREATE, READ], CN=principal2: [ALL]}``.
+
+.. note::
+    It is required to install ``kazoo>=2.6.0`` to support SSL.
+
+
 Exhibitor
 ---------
 -  **PATRONI\_EXHIBITOR\_HOSTS**: initial list of Exhibitor (ZooKeeper) nodes in format: 'host1,host2,etc...'. This list updates automatically whenever the Exhibitor (ZooKeeper) cluster topology changes.
@@ -73,28 +108,67 @@ Exhibitor
 
 Kubernetes
 ----------
+-  **PATRONI\_KUBERNETES\_BYPASS\_API\_SERVICE**: (optional) When communicating with the Kubernetes API, Patroni is usually relying on the `kubernetes` service, the address of which is exposed in the pods via the `KUBERNETES_SERVICE_HOST` environment variable. If `PATRONI_KUBERNETES_BYPASS_API_SERVICE` is set to ``true``, Patroni will resolve the list of API nodes behind the service and connect directly to them.
 -  **PATRONI\_KUBERNETES\_NAMESPACE**: (optional) Kubernetes namespace where the Patroni pod is running. Default value is `default`.
 -  **PATRONI\_KUBERNETES\_LABELS**: Labels in format ``{label1: value1, label2: value2}``. These labels will be used to find existing objects (Pods and either Endpoints or ConfigMaps) associated with the current cluster. Also Patroni will set them on every object (Endpoint or ConfigMap) it creates.
 -  **PATRONI\_KUBERNETES\_SCOPE\_LABEL**: (optional) name of the label containing cluster name. Default value is `cluster-name`.
 -  **PATRONI\_KUBERNETES\_ROLE\_LABEL**: (optional) name of the label containing Postgres role (`master` or `replica`). Patroni will set this label on the pod it is running in. Default value is `role`.
 -  **PATRONI\_KUBERNETES\_USE\_ENDPOINTS**: (optional) if set to true, Patroni will use Endpoints instead of ConfigMaps to run leader elections and keep cluster state.
 -  **PATRONI\_KUBERNETES\_POD\_IP**: (optional) IP address of the pod Patroni is running in. This value is required when `PATRONI_KUBERNETES_USE_ENDPOINTS` is enabled and is used to populate the leader endpoint subsets when the pod's PostgreSQL is promoted.
--  **PATRONI\_KUBERNETES\_PORTS**: (optional) if the Service object has the name for the port, the same name must appear in the Endpoint object, otherwise service won't work. For example, if your service is defined as ``{Kind: Service, spec: {ports: [{name: postgresql, port: 5432, targetPort: 5432}]}}``, then you have to set ``PATRONI_KUBERNETES_PORTS='{[{"name": "postgresql", "port": 5432}]}'`` and Patroni will use it for updating subsets of the leader Endpoint. This parameter is used only if `PATRONI_KUBERNETES_USE_ENDPOINTS` is set.
+-  **PATRONI\_KUBERNETES\_PORTS**: (optional) if the Service object has the name for the port, the same name must appear in the Endpoint object, otherwise service won't work. For example, if your service is defined as ``{Kind: Service, spec: {ports: [{name: postgresql, port: 5432, targetPort: 5432}]}}``, then you have to set ``PATRONI_KUBERNETES_PORTS='[{"name": "postgresql", "port": 5432}]'`` and Patroni will use it for updating subsets of the leader Endpoint. This parameter is used only if `PATRONI_KUBERNETES_USE_ENDPOINTS` is set.
+-  **PATRONI\_KUBERNETES\_CACERT**: (optional) Specifies the file with the CA_BUNDLE file with certificates of trusted CAs to use while verifying Kubernetes API SSL certs. If not provided, patroni will use the value provided by the ServiceAccount secret.
+
+Raft (deprecated)
+-----------------
+
+-  **PATRONI\_RAFT\_SELF\_ADDR**: ``ip:port`` to listen on for Raft connections. The ``self_addr`` must be accessible from other nodes of the cluster. If not set, the node will not participate in consensus.
+-  **PATRONI\_RAFT\_BIND\_ADDR**: (optional) ``ip:port`` to listen on for Raft connections. If not specified the ``self_addr`` will be used.
+-  **PATRONI\_RAFT\_PARTNER\_ADDRS**: list of other Patroni nodes in the cluster in format ``"'ip1:port1','ip2:port2'"``. It is important to quote every single entity!
+-  **PATRONI\_RAFT\_DATA\_DIR**: directory where to store Raft log and snapshot. If not specified the current working directory is used.
+-  **PATRONI\_RAFT\_PASSWORD**: (optional) Encrypt Raft traffic with a specified password, requires ``cryptography`` python module.
 
 PostgreSQL
 ----------
 -  **PATRONI\_POSTGRESQL\_LISTEN**: IP address + port that Postgres listens to. Multiple comma-separated addresses are permitted, as long as the port component is appended after to the last one with a colon, i.e. ``listen: 127.0.0.1,127.0.0.2:5432``. Patroni will use the first address from this list to establish local connections to the PostgreSQL node.
 -  **PATRONI\_POSTGRESQL\_CONNECT\_ADDRESS**: IP address + port through which Postgres is accessible from other nodes and applications.
+-  **PATRONI\_POSTGRESQL\_PROXY\_ADDRESS**: IP address + port through which a connection pool (e.g. pgbouncer) running next to Postgres is accessible. The value is written to the member key in DCS as ``proxy_url`` and could be used/useful for service discovery.
 -  **PATRONI\_POSTGRESQL\_DATA\_DIR**: The location of the Postgres data directory, either existing or to be initialized by Patroni.
 -  **PATRONI\_POSTGRESQL\_CONFIG\_DIR**: The location of the Postgres configuration directory, defaults to the data directory. Must be writable by Patroni.
 -  **PATRONI\_POSTGRESQL\_BIN_DIR**: Path to PostgreSQL binaries. (pg_ctl, pg_rewind, pg_basebackup, postgres) The  default value is an empty string meaning that PATH environment variable will be used to find the executables.
 -  **PATRONI\_POSTGRESQL\_PGPASS**: path to the `.pgpass <https://www.postgresql.org/docs/current/static/libpq-pgpass.html>`__ password file. Patroni creates this file before executing pg\_basebackup and under some other circumstances. The location must be writable by Patroni.
--  **PATRONI\_REPLICATION\_USERNAME**: replication username; the user will be created during initialization. Replicas will use this user to access master via streaming replication
+-  **PATRONI\_REPLICATION\_USERNAME**: replication username; the user will be created during initialization. Replicas will use this user to access the replication source via streaming replication
 -  **PATRONI\_REPLICATION\_PASSWORD**: replication password; the user will be created during initialization.
+-  **PATRONI\_REPLICATION\_SSLMODE**: (optional) maps to the `sslmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLMODE>`__ connection parameter, which allows a client to specify the type of TLS negotiation mode with the server. For more information on how each mode works, please visit the `PostgreSQL documentation <https://www.postgresql.org/docs/current/libpq-ssl.html#LIBPQ-SSL-SSLMODE-STATEMENTS>`__. The default mode is ``prefer``.
+-  **PATRONI\_REPLICATION\_SSLKEY**: (optional) maps to the `sslkey <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLKEY>`__ connection parameter, which specifies the location of the secret key used with the client's certificate.
+-  **PATRONI\_REPLICATION\_SSLPASSWORD**: (optional) maps to the `sslpassword <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLPASSWORD>`__ connection parameter, which specifies the password for the secret key specified in ``PATRONI_REPLICATION_SSLKEY``.
+-  **PATRONI\_REPLICATION\_SSLCERT**: (optional) maps to the `sslcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCERT>`__ connection parameter, which specifies the location of the client certificate.
+-  **PATRONI\_REPLICATION\_SSLROOTCERT**: (optional) maps to the `sslrootcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLROOTCERT>`__ connection parameter, which specifies the location of a file containing one ore more certificate authorities (CA) certificates that the client will use to verify a server's certificate.
+-  **PATRONI\_REPLICATION\_SSLCRL**: (optional) maps to the `sslcrl <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRL>`__ connection parameter, which specifies the location of a file containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
+-  **PATRONI\_REPLICATION\_SSLCRLDIR**: (optional) maps to the `sslcrldir <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRLDIR>`__ connection parameter, which specifies the location of a directory with files containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
+-  **PATRONI\_REPLICATION\_GSSENCMODE**: (optional) maps to the `gssencmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-GSSENCMODE>`__ connection parameter, which determines whether or with what priority a secure GSS TCP/IP connection will be negotiated with the server
+-  **PATRONI\_REPLICATION\_CHANNEL\_BINDING**: (optional) maps to the `channel_binding <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-CHANNEL-BINDING>`__ connection parameter, which controls the client's use of channel binding.
 -  **PATRONI\_SUPERUSER\_USERNAME**: name for the superuser, set during initialization (initdb) and later used by Patroni to connect to the postgres. Also this user is used by pg_rewind.
 -  **PATRONI\_SUPERUSER\_PASSWORD**: password for the superuser, set during initialization (initdb).
+-  **PATRONI\_SUPERUSER\_SSLMODE**: (optional) maps to the `sslmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLMODE>`__ connection parameter, which allows a client to specify the type of TLS negotiation mode with the server. For more information on how each mode works, please visit the `PostgreSQL documentation <https://www.postgresql.org/docs/current/libpq-ssl.html#LIBPQ-SSL-SSLMODE-STATEMENTS>`__. The default mode is ``prefer``.
+-  **PATRONI\_SUPERUSER\_SSLKEY**: (optional) maps to the `sslkey <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLKEY>`__ connection parameter, which specifies the location of the secret key used with the client's certificate.
+-  **PATRONI\_SUPERUSER\_SSLPASSWORD**: (optional) maps to the `sslpassword <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLPASSWORD>`__ connection parameter, which specifies the password for the secret key specified in ``PATRONI_SUPERUSER_SSLKEY``.
+-  **PATRONI\_SUPERUSER\_SSLCERT**: (optional) maps to the `sslcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCERT>`__ connection parameter, which specifies the location of the client certificate.
+-  **PATRONI\_SUPERUSER\_SSLROOTCERT**: (optional) maps to the `sslrootcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLROOTCERT>`__ connection parameter, which specifies the location of a file containing one ore more certificate authorities (CA) certificates that the client will use to verify a server's certificate.
+-  **PATRONI\_SUPERUSER\_SSLCRL**: (optional) maps to the `sslcrl <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRL>`__ connection parameter, which specifies the location of a file containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
+-  **PATRONI\_SUPERUSER\_SSLCRLDIR**: (optional) maps to the `sslcrldir <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRLDIR>`__ connection parameter, which specifies the location of a directory with files containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
+-  **PATRONI\_SUPERUSER\_GSSENCMODE**: (optional) maps to the `gssencmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-GSSENCMODE>`__ connection parameter, which determines whether or with what priority a secure GSS TCP/IP connection will be negotiated with the server
+-  **PATRONI\_SUPERUSER\_CHANNEL\_BINDING**: (optional) maps to the `channel_binding <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-CHANNEL-BINDING>`__ connection parameter, which controls the client's use of channel binding.
 -  **PATRONI\_REWIND\_USERNAME**: name for the user for ``pg_rewind``; the user will be created during initialization of postgres 11+ and all necessary `permissions <https://www.postgresql.org/docs/11/app-pgrewind.html#id-1.9.5.8.8>`__ will be granted.
 -  **PATRONI\_REWIND\_PASSWORD**: password for the user for ``pg_rewind``; the user will be created during initialization.
+-  **PATRONI\_REWIND\_SSLMODE**: (optional) maps to the `sslmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLMODE>`__ connection parameter, which allows a client to specify the type of TLS negotiation mode with the server. For more information on how each mode works, please visit the `PostgreSQL documentation <https://www.postgresql.org/docs/current/libpq-ssl.html#LIBPQ-SSL-SSLMODE-STATEMENTS>`__. The default mode is ``prefer``.
+-  **PATRONI\_REWIND\_SSLKEY**: (optional) maps to the `sslkey <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLKEY>`__ connection parameter, which specifies the location of the secret key used with the client's certificate.
+-  **PATRONI\_REWIND\_SSLPASSWORD**: (optional) maps to the `sslpassword <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLPASSWORD>`__ connection parameter, which specifies the password for the secret key specified in ``PATRONI_REWIND_SSLKEY``.
+-  **PATRONI\_REWIND\_SSLCERT**: (optional) maps to the `sslcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCERT>`__ connection parameter, which specifies the location of the client certificate.
+-  **PATRONI\_REWIND\_SSLROOTCERT**: (optional) maps to the `sslrootcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLROOTCERT>`__ connection parameter, which specifies the location of a file containing one ore more certificate authorities (CA) certificates that the client will use to verify a server's certificate.
+-  **PATRONI\_REWIND\_SSLCRL**: (optional) maps to the `sslcrl <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRL>`__ connection parameter, which specifies the location of a file containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
+-  **PATRONI\_REWIND\_SSLCRLDIR**: (optional) maps to the `sslcrldir <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRLDIR>`__ connection parameter, which specifies the location of a directory with files containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
+-  **PATRONI\_REWIND\_GSSENCMODE**: (optional) maps to the `gssencmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-GSSENCMODE>`__ connection parameter, which determines whether or with what priority a secure GSS TCP/IP connection will be negotiated with the server
+-  **PATRONI\_REWIND\_CHANNEL\_BINDING**: (optional) maps to the `channel_binding <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-CHANNEL-BINDING>`__ connection parameter, which controls the client's use of channel binding.
 
 REST API
 --------
@@ -104,7 +178,19 @@ REST API
 -  **PATRONI\_RESTAPI\_PASSWORD**: Basic-auth password to protect unsafe REST API endpoints.
 -  **PATRONI\_RESTAPI\_CERTFILE**: Specifies the file with the certificate in the PEM format. If the certfile is not specified or is left empty, the API server will work without SSL.
 -  **PATRONI\_RESTAPI\_KEYFILE**: Specifies the file with the secret key in the PEM format.
+-  **PATRONI\_RESTAPI\_KEYFILE\_PASSWORD**: Specifies a password for decrypting the keyfile.
+-  **PATRONI\_RESTAPI\_CAFILE**: Specifies the file with the CA_BUNDLE with certificates of trusted CAs to use while verifying client certs.
+-  **PATRONI\_RESTAPI\_CIPHERS**: (optional) Specifies the permitted cipher suites (e.g. "ECDHE-RSA-AES256-GCM-SHA384:DHE-RSA-AES256-GCM-SHA384:ECDHE-RSA-AES128-GCM-SHA256:DHE-RSA-AES128-GCM-SHA256:!SSLv1:!SSLv2:!SSLv3:!TLSv1:!TLSv1.1")
+-  **PATRONI\_RESTAPI\_VERIFY\_CLIENT**: ``none`` (default), ``optional`` or ``required``. When ``none`` REST API will not check client certificates. When ``required`` client certificates are required for all REST API calls. When ``optional`` client certificates are required for all unsafe REST API endpoints. When ``required`` is used, then client authentication succeeds, if the certificate signature verification succeeds. For ``optional`` the client cert will only be checked for ``PUT``, ``POST``, ``PATCH``, and ``DELETE`` requests.
+-  **PATRONI\_RESTAPI\_ALLOWLIST**: (optional): Specifies the set of hosts that are allowed to call unsafe REST API endpoints. The single element could be a host name, an IP address or a network address using CIDR notation. By default ``allow all`` is used. In case if ``allowlist`` or ``allowlist_include_members`` are set, anything that is not included is rejected.
+-  **PATRONI\_RESTAPI\_ALLOWLIST\_INCLUDE\_MEMBERS**: (optional): If set to ``true`` it allows accessing unsafe REST API endpoints from other cluster members registered in DCS (IP address or hostname is taken from the members ``api_url``). Be careful, it might happen that OS will use a different IP for outgoing connections.
+-  **PATRONI\_RESTAPI\_HTTP\_EXTRA\_HEADERS**: (optional) HTTP headers let the REST API server pass additional information with an HTTP response.
+-  **PATRONI\_RESTAPI\_HTTPS\_EXTRA\_HEADERS**: (optional) HTTPS headers let the REST API server pass additional information with an HTTP response when TLS is enabled. This will also pass additional information set in ``http_extra_headers``.
 
-ZooKeeper
+CTL
----------
+---
--  **PATRONI\_ZOOKEEPER\_HOSTS**: comma separated list of ZooKeeper cluster members: "'host1:port1','host2:port2','etc...'". It is important to quote every single entity!
+-  **PATRONICTL\_CONFIG\_FILE**: location of the configuration file.
+-  **PATRONI\_CTL\_INSECURE**: Allow connections to REST API without verifying SSL certs.
+-  **PATRONI\_CTL\_CACERT**: Specifies the file with the CA_BUNDLE file or directory with certificates of trusted CAs to use while verifying REST API SSL certs. If not provided patronictl will use the value provided for REST API "cafile" parameter.
+-  **PATRONI\_CTL\_CERTFILE**: Specifies the file with the client certificate in the PEM format. If not provided patronictl will use the value provided for REST API "certfile" parameter.
+-  **PATRONI\_CTL\_KEYFILE**: Specifies the file with the client secret key in the PEM format. If not provided patronictl will use the value provided for REST API "keyfile" parameter.

docs/README.rst

@@ -35,7 +35,7 @@ To install requirements on a Mac, run the following:
 
 .. _psycopg2_install_options:
 
-**Psycopg2**
+**Psycopg**
 
 Starting from `psycopg2-2.8 <http://initd.org/psycopg/articles/2019/04/04/psycopg-28-released/>`__ the binary version of psycopg2 will no longer be installed by default. Installing it from the source code requires C compiler and postgres+python dev packages.
 Since in the python world it is not possible to specify dependency as ``psycopg2 OR psycopg2-binary`` you will have to decide how to install it.
@@ -62,6 +62,12 @@ There are a few options available:
 
     pip install psycopg2>=2.5.4
 
+4. Use psycopg 3.0 instead of psycopg2
+
+::
+
+    pip install psycopg[binary]>=3.0.0
+
 **General installation for pip**
 
 Patroni can be installed with pip:
@@ -72,8 +78,8 @@ Patroni can be installed with pip:
 
 where dependencies can be either empty, or consist of one or more of the following:
 
-etcd
+etcd or etcd3
-    `python-etcd` module in order to use Etcd as DCS
+    `python-etcd` module in order to use Etcd as Distributed Configuration Store (DCS)
 consul
     `python-consul` module in order to use Consul as DCS
 zookeeper
@@ -82,6 +88,8 @@ exhibitor
     `kazoo` module in order to use Exhibitor as DCS (same dependencies as for Zookeeper)
 kubernetes
     `kubernetes` module in order to use Kubernetes as DCS in Patroni
+raft
+    `pysyncobj` module in order to use python Raft implementation as DCS
 aws
     `boto` in order to use AWS callbacks
 
@@ -97,6 +105,13 @@ independently of Patroni.
 
 .. _running_configuring:
 
+Planning the Number of PostgreSQL Nodes
+---------------------------------------
+
+Patroni/PostgreSQL nodes are decoupled from DCS nodes (except when Patroni implements RAFT on its own) and therefore
+there is no requirement on the minimal number of nodes. Running a cluster consisting of one primary and one standby is
+perfectly fine. You can add more standby nodes later.
+
 Running and Configuring
 -----------------------
 
@@ -107,7 +122,7 @@ obtain those files from the git repository and replace `./patroni.py` below with
 To get started, do the following from different terminals:
 ::
 
-    > etcd --data-dir=data/etcd
+    > etcd --data-dir=data/etcd --enable-v2=true
     > ./patroni.py postgres0.yml
     > ./patroni.py postgres1.yml
 
@@ -154,3 +169,20 @@ When connecting from an application, always use a non-superuser. Patroni require
    :target: https://travis-ci.org/zalando/patroni
 .. |Coverage Status| image:: https://coveralls.io/repos/zalando/patroni/badge.svg?branch=master
    :target: https://coveralls.io/r/zalando/patroni?branch=master
+
+Testing Your HA Solution
+--------------------------------------
+Testing an HA solution is a time consuming process, with many variables. This is particularly true considering a cross-platform application. You need a trained system administrator or a consultant to do this work. It is not something we can cover in depth in the documentation.
+
+That said, here are some pieces of your infrastructure you should be sure to test:
+
+* Network (the network in front of your system as well as the NICs [physical or virtual] themselves)
+* Disk IO
+* file limits (nofile in Linux)
+* RAM. Even if you have oomkiller turned off as suggested, the unavailability of RAM could cause issues.
+* CPU
+* Virtualization Contention (overcommitting the hypervisor)
+* Any cgroup limitation (likely to be related to the above)
+* ``kill -9`` of any postgres process (except postmaster!). This is a decent simulation of a segfault.
+
+One thing that you should not do is run ``kill -9`` on a postmaster process. This is because doing so does not mimic any real life scenario. If you are concerned your infrastructure is insecure and an attacker could run ``kill -9``, no amount of HA process is going to fix that. The attacker will simply kill the process again, or cause chaos in another way.

docs/SETTINGS.rst

@@ -4,6 +4,69 @@
 YAML Configuration Settings
 ===========================
 
+.. _dynamic_configuration_settings:
+
+Dynamic configuration settings
+------------------------------
+
+Dynamic configuration is stored in the DCS (Distributed Configuration Store) and applied on all cluster nodes. Some parameters, like **loop_wait**, **ttl**, **postgresql.parameters.max_connections**, **postgresql.parameters.max_worker_processes** and so on could be set only in the dynamic configuration. Some other parameters like **postgresql.listen**, **postgresql.data_dir** could be set only locally, i.e. in the Patroni config file or via :ref:`configuration <environment>` variable. In most cases the local configuration will override the dynamic configuration. In order to change the dynamic configuration you can use either ``patronictl edit-config`` tool or Patroni :ref:`REST API <rest_api>`.
+
+-  **loop\_wait**: the number of seconds the loop will sleep. Default value: 10
+-  **ttl**: the TTL to acquire the leader lock (in seconds). Think of it as the length of time before initiation of the automatic failover process. Default value: 30
+-  **retry\_timeout**: timeout for DCS and PostgreSQL operation retries (in seconds). DCS or network issues shorter than this will not cause Patroni to demote the leader. Default value: 10
+-  **maximum\_lag\_on\_failover**: the maximum bytes a follower may lag to be able to participate in leader election.
+-  **maximum\_lag\_on\_syncnode**: the maximum bytes a synchronous follower may lag before it is considered as an unhealthy candidate and swapped by healthy asynchronous follower. Patroni utilize the max replica lsn if there is more than one follower, otherwise it will use leader's current wal lsn. Default is -1, Patroni will not take action to swap synchronous unhealthy follower when the value is set to 0 or below. Please set the value high enough so Patroni won't swap synchrounous follower fequently during high transaction volume.
+-  **max\_timelines\_history**: maximum number of timeline history items kept in DCS.  Default value: 0. When set to 0, it keeps the full history in DCS.
+-  **primary\_start\_timeout**: the amount of time a primary is allowed to recover from failures before failover is triggered (in seconds). Default is 300 seconds. When set to 0 failover is done immediately after a crash is detected if possible. When using asynchronous replication a failover can cause lost transactions. Worst case failover time for primary failure is: loop\_wait + primary\_start\_timeout + loop\_wait, unless primary\_start\_timeout is zero, in which case it's just loop\_wait. Set the value according to your durability/availability tradeoff.
+- **primary\_stop\_timeout**: The number of seconds Patroni is allowed to wait when stopping Postgres and effective only when synchronous_mode is enabled. When set to > 0 and the synchronous_mode is enabled, Patroni sends SIGKILL to the postmaster if the stop operation is running for more than the value set by primary\_stop\_timeout. Set the value according to your durability/availability tradeoff. If the parameter is not set or set <= 0, primary\_stop\_timeout does not apply.
+-  **synchronous\_mode**: turns on synchronous replication mode. In this mode a replica will be chosen as synchronous and only the latest leader and synchronous replica are able to participate in leader election. Synchronous mode makes sure that successfully committed transactions will not be lost at failover, at the cost of losing availability for writes when Patroni cannot ensure transaction durability. See :ref:`replication modes documentation <replication_modes>` for details.
+-  **synchronous\_mode\_strict**: prevents disabling synchronous replication if no synchronous replicas are available, blocking all client writes to the primary. See :ref:`replication modes documentation <replication_modes>` for details.
+-  **failsafe\_mode**: Enables :ref:`DCS Failsafe Mode <dcs_failsafe_mode>`. Defaults to `false`.
+-  **postgresql**:
+    -  **use\_pg\_rewind**: whether or not to use pg_rewind. Defaults to `false`.
+    -  **use\_slots**: whether or not to use replication slots. Defaults to `true` on PostgreSQL 9.4+.
+    -  **recovery\_conf**: additional configuration settings written to recovery.conf when configuring follower. There is no recovery.conf anymore in PostgreSQL 12, but you may continue using this section, because Patroni handles it transparently.
+    -  **parameters**: list of configuration settings for Postgres.
+-  **standby\_cluster**: if this section is defined, we want to bootstrap a standby cluster.
+    -  **host**: an address of remote node
+    -  **port**: a port of remote node
+    -  **primary\_slot\_name**: which slot on the remote node to use for replication. This parameter is optional, the default value is derived from the instance name (see function `slot_name_from_member_name`).
+    -  **create\_replica\_methods**: an ordered list of methods that can be used to bootstrap standby leader from the remote primary, can be different from the list defined in :ref:`postgresql_settings`
+    -  **restore\_command**: command to restore WAL records from the remote primary to nodes in a standby cluster, can be different from the list defined in :ref:`postgresql_settings`
+    -  **archive\_cleanup\_command**: cleanup command for standby leader
+    -  **recovery\_min\_apply\_delay**: how long to wait before actually apply WAL records on a standby leader
+-  **slots**: define permanent replication slots. These slots will be preserved during switchover/failover. The logical slots are copied from the primary to a standby with restart, and after that their position advanced every **loop_wait** seconds (if necessary). Copying logical slot files performed via ``libpq`` connection and using either rewind or superuser credentials (see **postgresql.authentication** section). There is always a chance that the logical slot position on the replica is a bit behind the former primary, therefore application should be prepared that some messages could be received the second time after the failover. The easiest way of doing so - tracking ``confirmed_flush_lsn``. Enabling permanent logical replication slots requires **postgresql.use_slots** to be set and will also automatically enable the ``hot_standby_feedback``. Since the failover of logical replication slots is unsafe on PostgreSQL 9.6 and older and PostgreSQL version 10 is missing some important functions, the feature only works with PostgreSQL 11+.
+    -  **my_slot_name**: the name of replication slot. If the permanent slot name matches with the name of the current primary it will not be created. Everything else is the responsibility of the operator to make sure that there are no clashes in names between replication slots automatically created by Patroni for members and permanent replication slots.
+        -  **type**: slot type. Could be ``physical`` or ``logical``. If the slot is logical, you have to additionally define ``database`` and ``plugin``.
+        -  **database**: the database name where logical slots should be created.
+        -  **plugin**: the plugin name for the logical slot.
+-  **ignore_slots**: list of sets of replication slot properties for which Patroni should ignore matching slots. This configuration/feature/etc. is useful when some replication slots are managed outside of Patroni. Any subset of matching properties will cause a slot to be ignored.
+    -  **name**: the name of the replication slot.
+    -  **type**: slot type. Can be ``physical`` or ``logical``. If the slot is logical, you may additionally define ``database`` and/or ``plugin``.
+    -  **database**: the database name (when matching a ``logical`` slot).
+    -  **plugin**: the logical decoding plugin (when matching a ``logical`` slot).
+
+Note: **slots** is a hashmap while **ignore_slots** is an array. For example:
+
+.. code:: YAML
+
+        slots:
+          permanent_logical_slot_name:
+            type: logical
+            database: my_db
+            plugin: test_decoding
+          permanent_physical_slot_name:
+            type: physical
+          ...
+        ignore_slots:
+          - name: ignored_logical_slot_name
+            type: logical
+            database: my_db
+            plugin: test_decoding
+          - name: ignored_physical_slot_name
+            type: physical
+          ...
+
 Global/Universal
 ----------------
 -  **name**: the name of the host. Must be unique for the cluster.
@@ -13,6 +76,7 @@ Global/Universal
 Log
 ---
 -  **level**: sets the general logging level. Default value is **INFO** (see `the docs for Python logging <https://docs.python.org/3.6/library/logging.html#levels>`_)
+-  **traceback\_level**: sets the level where tracebacks will be visible. Default value is **ERROR**. Set it to **DEBUG** if you want to see tracebacks only if you enable **log.level=DEBUG**.
 -  **format**: sets the log formatting string. Default value is **%(asctime)s %(levelname)s: %(message)s** (see `the LogRecord attributes <https://docs.python.org/3.6/library/logging.html#logrecord-attributes>`_)
 -  **dateformat**: sets the datetime formatting string. (see the `formatTime() documentation <https://docs.python.org/3.6/library/logging.html#logging.Formatter.formatTime>`_)
 -  **max\_queue\_size**: Patroni is using two-step logging. Log records are written into the in-memory queue and there is a separate thread which pulls them from the queue and writes to stderr or file. The maximum size of the internal queue is limited by default by **1000** records, which is enough to keep logs for the past 1h20m.
@@ -27,50 +91,35 @@ Log
 
 Bootstrap configuration
 -----------------------
--  **dcs**: This section will be written into `/<namespace>/<scope>/config` of a given configuration store after initializing of new cluster. This is the global configuration for the cluster. If you want to change some parameters for all cluster nodes - just do it in DCS (or via Patroni API) and all nodes will apply this configuration.
+-  **bootstrap**:
-    -  **loop\_wait**: the number of seconds the loop will sleep. Default value: 10
+    -  **dcs**: This section will be written into `/<namespace>/<scope>/config` of the given configuration store after initializing of new cluster. The global dynamic configuration for the cluster. Under the ``bootstrap.dcs`` you can put any of the parameters described in the :ref:`Dynamic Configuration settings <dynamic_configuration_settings>` and after Patroni initialized (bootstrapped) the new cluster, it will write this section into `/<namespace>/<scope>/config` of the configuration store. All later changes of ``bootstrap.dcs`` will not take any effect! If you want to change them please use either ``patronictl edit-config`` or Patroni :ref:`REST API <rest_api>`.
-    -  **ttl**: the TTL to acquire the leader lock. Think of it as the length of time before initiation of the automatic failover process. Default value: 30
+    -  **method**: custom script to use for bootstrapping this cluster.
-    -  **retry\_timeout**: timeout for DCS and PostgreSQL operation retries. DCS or network issues shorter than this will not cause Patroni to demote the leader. Default value: 10
+       See :ref:`custom bootstrap methods documentation <custom_bootstrap>` for details.
-    -  **maximum\_lag\_on\_failover**: the maximum bytes a follower may lag to be able to participate in leader election.
+       When ``initdb`` is specified revert to the default ``initdb`` command. ``initdb`` is also triggered when no ``method``
-    -  **master\_start\_timeout**: the amount of time a master is allowed to recover from failures before failover is triggered. Default is 300 seconds. When set to 0 failover is done immediately after a crash is detected if possible. When using asynchronous replication a failover can cause lost transactions. Best worst case failover time for master failure is: loop\_wait + master\_start\_timeout + loop\_wait, unless master\_start\_timeout is zero, in which case it's just loop\_wait. Set the value according to your durability/availability tradeoff.
+       parameter is present in the configuration file.
-    -  **synchronous\_mode**: turns on synchronous replication mode. In this mode a replica will be chosen as synchronous and only the latest leader and synchronous replica are able to participate in leader election. Synchronous mode makes sure that successfully committed transactions will not be lost at failover, at the cost of losing availability for writes when Patroni cannot ensure transaction durability. See :ref:`replication modes documentation <replication_modes>` for details.
+    -  **initdb**: List options to be passed on to initdb.
-    -  **synchronous\_mode\_strict**: prevents disabling synchronous replication if no synchronous replicas are available, blocking all client writes to the master. See :ref:`replication modes documentation <replication_modes>` for details.
+            -  **- data-checksums**: Must be enabled when pg_rewind is needed on 9.3.
-    -  **postgresql**:
+            -  **- encoding: UTF8**: default encoding for new databases.
-        -  **use\_pg\_rewind**: whether or not to use pg_rewind
+            -  **- locale: UTF8**: default locale for new databases.
-        -  **use\_slots**: whether or not to use replication_slots. Must be False for PostgreSQL 9.3. You should comment out max_replication_slots before it becomes ineligible for leader status.
+    -  **pg\_hba**: list of lines that you should add to pg\_hba.conf.
-        -  **recovery\_conf**: additional configuration settings written to recovery.conf when configuring follower.
+            -  **- host all all 0.0.0.0/0 md5**.
-        -  **parameters**: list of configuration settings for Postgres. Many of these are required for replication to work.
+            -  **- host replication replicator 127.0.0.1/32 md5**: A line like this is required for replication.
-    -  **standby\_cluster**: if this section is defined, we want to bootstrap a standby cluster.
+    -  **users**: Some additional users which need to be created after initializing new cluster
-        -  **host**: an address of remote master
+        -  **admin**: the name of user
-        -  **port**: a port of remote master
+            -  **password: zalando**:
-        -  **primary\_slot\_name**: which slot on the remote master to use for replication. This parameter is optional, the default value is derived from the instance name (see function `slot_name_from_member_name`).
+            -  **options**: list of options for CREATE USER statement
-        -  **create\_replica\_methods**: an ordered list of methods that can be used to bootstrap standby leader from the remote master, can be different from the list defined in :ref:`postgresql_settings`
+                -  **- createrole**
-        -  **restore\_command**: command to restore WAL records from the remote master to standby leader, can be different from the list defined in :ref:`postgresql_settings`
+                -  **- createdb**
-        -  **archive\_cleanup\_command**: cleanup command for standby leader
+    -  **post\_bootstrap** or **post\_init**: An additional script that will be executed after initializing the cluster. The script receives a connection string URL (with the cluster superuser as a user name). The PGPASSFILE variable is set to the location of pgpass file.
-        -  **recovery\_min\_apply\_delay**: how long to wait before actually apply WAL records on a standby leader
+
-    -  **slots**: define permanent replication slots. These slots will be preserved during switchover/failover. Patroni will try to create slots before opening connections to the cluster.
+.. _citus_settings:
-        -  **my_slot_name**: the name of replication slot. It is the responsibility of the operator to make sure that there are no clashes in names between replication slots automatically created by Patroni for members and permanent replication slots.
+
-            -  **type**: slot type. Could be ``physical`` or ``logical``. If the slot is logical, you have to additionally define ``database`` and ``plugin``.
+Citus
-            -  **database**: the database name where logical slots should be created.
+-----
-            -  **plugin**: the plugin name for the logical slot.
+Enables integration Patroni with `Citus <https://docs.citusdata.com>`__. If configured, Patroni will take care of registering Citus worker nodes on the coordinator. You can find more information about Citus support :ref:`here <citus>`.
--  **method**: custom script to use for bootstrapping this cluster.
+
-   See :ref:`custom bootstrap methods documentation <custom_bootstrap>` for details.
+-  **group**: the Citus group id, integer. Use ``0`` for coordinator and ``1``, ``2``, etc... for workers
-   When ``initdb`` is specified revert to the default ``initdb`` command. ``initdb`` is also triggered when no ``method``
+-  **database**: the database where ``citus`` extension should be created. Must be the same on the coordinator and all workers. Currently only one database is supported.
-   parameter is present in the configuration file.
--  **initdb**: List options to be passed on to initdb.
-        -  **- data-checksums**: Must be enabled when pg_rewind is needed on 9.3.
-        -  **- encoding: UTF8**: default encoding for new databases.
-        -  **- locale: UTF8**: default locale for new databases.
--  **pg\_hba**: list of lines that you should add to pg\_hba.conf.
-        -  **- host all all 0.0.0.0/0 md5**.
-        -  **- host replication replicator 127.0.0.1/32 md5**: A line like this is required for replication.
--  **users**: Some additional users which need to be created after initializing new cluster
-    -  **admin**: the name of user
-        -  **password: zalando**:
-        -  **options**: list of options for CREATE USER statement
-            -  **- createrole**
-            -  **- createdb**
--  **post\_bootstrap** or **post\_init**: An additional script that will be executed after initializing the cluster. The script receives a connection string URL (with the cluster superuser as a user name). The PGPASSFILE variable is set to the location of pgpass file.
 
 .. _consul_settings:
 
@@ -78,20 +127,36 @@ Consul
 ------
 Most of the parameters are optional, but you have to specify one of the **host** or **url**
 
--  **host**: the host:port for the Consul endpoint, in format: http(s)://host:port
+-  **host**: the host:port for the Consul local agent.
--  **url**: url for the Consul endpoint
+-  **url**: url for the Consul local agent, in format: http(s)://host:port.
--  **port**: (optional) Consul port
+-  **port**: (optional) Consul port.
--  **scheme**: (optional) **http** or **https**, defaults to **http**
+-  **scheme**: (optional) **http** or **https**, defaults to **http**.
--  **token**: (optional) ACL token
+-  **token**: (optional) ACL token.
--  **verify**: (optional) whether to verify the SSL certificate for HTTPS requests
+-  **verify**: (optional) whether to verify the SSL certificate for HTTPS requests.
 -  **cacert**: (optional) The ca certificate. If present it will enable validation.
--  **cert**: (optional) file with the client certificate
+-  **cert**: (optional) file with the client certificate.
 -  **key**: (optional) file with the client key. Can be empty if the key is part of **cert**.
 -  **dc**: (optional) Datacenter to communicate with. By default the datacenter of the host is used.
 -  **consistency**: (optional) Select consul consistency mode. Possible values are ``default``, ``consistent``, or ``stale`` (more details in `consul API reference <https://www.consul.io/api/features/consistency.html/>`__)
--  **checks**: (optional) list of Consul health checks used for the session. If not specified Consul will use "serfHealth" in additional to the TTL based check created by Patroni. Additional checks, in particular the "serfHealth", may cause the leader lock to expire faster than in `ttl` seconds when the leader instance becomes unavailable
+-  **checks**: (optional) list of Consul health checks used for the session. By default an empty list is used.
--  **register\_service**: (optional) whether or not to register a service with the name defined by the scope parameter and the tag master, replica or standby-leader depending on the node's role. Defaults to **false**
+-  **register\_service**: (optional) whether or not to register a service with the name defined by the scope parameter and the tag master, primary, replica, or standby-leader depending on the node's role. Defaults to **false**.
--  **service\_check\_interval**: (optional) how often to perform health check against registered url
+-  **service\_tags**: (optional) additional static tags to add to the Consul service apart from the role (``master``/``primary``/``replica``/``standby-leader``). By default an empty list is used.
+-  **service\_check\_interval**: (optional) how often to perform health check against registered url. Defaults to '5s'.
+-  **service\_check\_tls\_server\_name**: (optional) overide SNI host when connecting via TLS, see also `consul agent check API reference <https://www.consul.io/api-docs/agent/check#tlsservername>`__.
+
+The ``token`` needs to have the following ACL permissions:
+
+::
+
+    service_prefix "${scope}" {
+        policy = "write"
+    }
+    key_prefix "${namespace}/${scope}" {
+        policy = "write"
+    }
+    session_prefix "" {
+        policy = "write"
+    }
 
 Etcd
 ----
@@ -100,9 +165,10 @@ Most of the parameters are optional, but you have to specify one of the **host**
 -  **host**: the host:port for the etcd endpoint.
 -  **hosts**: list of etcd endpoint in format host1:port1,host2:port2,etc... Could be a comma separated string or an actual yaml list.
 -  **use\_proxies**: If this parameter is set to true, Patroni will consider **hosts** as a list of proxies and will not perform a topology discovery of etcd cluster.
--  **url**: url for the etcd
+-  **url**: url for the etcd.
--  **proxy**: proxy url for the etcd. If you are connecting to the etcd using proxy, use this parameter instead of **url**
+-  **proxy**: proxy url for the etcd. If you are connecting to the etcd using proxy, use this parameter instead of **url**.
--  **srv**: Domain to search the SRV record(s) for cluster autodiscovery.
+-  **srv**: Domain to search the SRV record(s) for cluster autodiscovery. Patroni will try to query these SRV service names for specified domain (in that order until first success): ``_etcd-client-ssl``, ``_etcd-client``, ``_etcd-ssl``, ``_etcd``, ``_etcd-server-ssl``, ``_etcd-server``. If SRV records for ``_etcd-server-ssl`` or ``_etcd-server`` are retrieved then ETCD peer protocol is used do query ETCD for available members. Otherwise hosts from SRV records will be used.
+-  **srv\_suffix**: Configures a suffix to the SRV name that is queried during discovery. Use this flag to differentiate between multiple etcd clusters under the same domain. Works only with conjunction with **srv**. For example, if ``srv_suffix: foo`` and ``srv: example.org`` are set, the following DNS SRV query is made:``_etcd-client-ssl-foo._tcp.example.com`` (and so on for every possible ETCD SRV service name).
 -  **protocol**: (optional) http or https, if not specified http is used. If the **url** or **proxy** is specified - will take protocol from them.
 -  **username**: (optional) username for etcd authentication.
 -  **password**: (optional) password for etcd authentication.
@@ -110,98 +176,234 @@ Most of the parameters are optional, but you have to specify one of the **host**
 -  **cert**: (optional) file with the client certificate.
 -  **key**: (optional) file with the client key. Can be empty if the key is part of **cert**.
 
+Etcdv3
+------
+If you want that Patroni works with Etcd cluster via protocol version 3, you need to use the ``etcd3`` section in the Patroni configuration file. All configuration parameters are the same as for ``etcd``.
+
+.. warning::
+    Keys created with protocol version 2 are not visible with protocol version 3 and the other way around, therefore it is not possible to switch from ``etcd`` to ``etcd3`` just by updating Patroni config file.
+
+
+ZooKeeper
+----------
+-  **hosts**: List of ZooKeeper cluster members in format: ['host1:port1', 'host2:port2', 'etc...'].
+-  **use_ssl**: (optional) Whether SSL is used or not. Defaults to ``false``. If set to ``false``, all SSL specific parameters are ignored.
+-  **cacert**: (optional) The CA certificate. If present it will enable validation.
+-  **cert**: (optional) File with the client certificate.
+-  **key**: (optional) File with the client key.
+-  **key_password**: (optional) The client key password.
+-  **verify**: (optional) Whether to verify certificate or not. Defaults to ``true``.
+-  **set_acls**: (optional) If set, configure Kazoo to apply a default ACL to each ZNode that it creates. ACLs will assume 'x509' schema and should be specified as a dictionary with the principal as the key and one or more permissions as a list in the value.  Permissions may be one of ``CREATE``, ``READ``, ``WRITE``, ``DELETE`` or ``ADMIN``.  For example, ``set_acls: {CN=principal1: [CREATE, READ], CN=principal2: [ALL]}``.
+
+.. note::
+    It is required to install ``kazoo>=2.6.0`` to support SSL.
+
+
 Exhibitor
 ---------
 -  **hosts**: initial list of Exhibitor (ZooKeeper) nodes in format: 'host1,host2,etc...'. This list updates automatically whenever the Exhibitor (ZooKeeper) cluster topology changes.
--  **poll\_interval**: how often the list of ZooKeeper and Exhibitor nodes should be updated from Exhibitor
+-  **poll\_interval**: how often the list of ZooKeeper and Exhibitor nodes should be updated from Exhibitor.
 -  **port**: Exhibitor port.
 
 .. _kubernetes_settings:
 
 Kubernetes
 ----------
+-  **bypass\_api\_service**: (optional) When communicating with the Kubernetes API, Patroni is usually relying on the `kubernetes` service, the address of which is exposed in the pods via the `KUBERNETES_SERVICE_HOST` environment variable. If `bypass_api_service` is set to ``true``, Patroni will resolve the list of API nodes behind the service and connect directly to them.
 -  **namespace**: (optional) Kubernetes namespace where Patroni pod is running. Default value is `default`.
 -  **labels**: Labels in format ``{label1: value1, label2: value2}``. These labels will be used to find existing objects (Pods and either Endpoints or ConfigMaps) associated with the current cluster. Also Patroni will set them on every object (Endpoint or ConfigMap) it creates.
 -  **scope\_label**: (optional) name of the label containing cluster name. Default value is `cluster-name`.
 -  **role\_label**: (optional) name of the label containing role (master or replica). Patroni will set this label on the pod it runs in. Default value is ``role``.
 -  **use\_endpoints**: (optional) if set to true, Patroni will use Endpoints instead of ConfigMaps to run leader elections and keep cluster state.
 -  **pod\_ip**: (optional) IP address of the pod Patroni is running in. This value is required when `use_endpoints` is enabled and is used to populate the leader endpoint subsets when the pod's PostgreSQL is promoted.
--  **ports**: (optional) if the Service object has the name for the port, the same name must appear in the Endpoint object, otherwise service won't work. For example, if your service is defined as ``{Kind: Service, spec: {ports: [{name: postgresql, port: 5432, targetPort: 5432}]}}``, then you have to set ``kubernetes.ports: {[{"name": "postgresql", "port": 5432}]}`` and Patroni will use it for updating subsets of the leader Endpoint. This parameter is used only if `kubernetes.use_endpoints` is set.
+-  **ports**: (optional) if the Service object has the name for the port, the same name must appear in the Endpoint object, otherwise service won't work. For example, if your service is defined as ``{Kind: Service, spec: {ports: [{name: postgresql, port: 5432, targetPort: 5432}]}}``, then you have to set ``kubernetes.ports: [{"name": "postgresql", "port": 5432}]`` and Patroni will use it for updating subsets of the leader Endpoint. This parameter is used only if `kubernetes.use_endpoints` is set.
+-  **cacert**: (optional) Specifies the file with the CA_BUNDLE file with certificates of trusted CAs to use while verifying Kubernetes API SSL certs. If not provided, patroni will use the value provided by the ServiceAccount secret.
+
+
+.. _raft_settings:
+
+Raft (deprecated)
+-----------------
+-  **self\_addr**: ``ip:port`` to listen on for Raft connections. The ``self_addr`` must be accessible from other nodes of the cluster. If not set, the node will not participate in consensus.
+-  **bind\_addr**: (optional) ``ip:port`` to listen on for Raft connections. If not specified the ``self_addr`` will be used.
+-  **partner\_addrs**: list of other Patroni nodes in the cluster in format: ['ip1:port', 'ip2:port', 'etc...']
+-  **data\_dir**: directory where to store Raft log and snapshot. If not specified the current working directory is used.
+-  **password**: (optional) Encrypt Raft traffic with a specified password, requires ``cryptography`` python module.
+
+  Short FAQ about Raft implementation
+
+  - Q: How to list all the nodes providing consensus?
+
+    A: ``syncobj_admin -conn host:port -status`` where the host:port is the address of one of the cluster nodes
+
+  - Q: Node that was a part of consensus and has gone and I can't reuse the same IP for other node. How to remove this node from the consensus?
+
+    A: ``syncobj_admin -conn host:port -remove host2:port2`` where the ``host2:port2`` is the address of the node you want to remove from consensus.
+
+  - Q: Where to get the ``syncobj_admin`` utility?
+
+    A: It is installed together with ``pysyncobj`` module (python RAFT implementation), which is Patroni dependency.
+
+  - Q: it is possible to run Patroni node without adding in to the consensus?
+
+    A: Yes, just comment out or remove ``raft.self_addr`` from Patroni configuration.
+
+  - Q: It is possible to run Patroni and PostgreSQL only on two nodes?
+
+    A: Yes, on the third node you can run ``patroni_raft_controller`` (without Patroni and PostgreSQL). In such a setup, one can temporarily lose one node without affecting the primary.
+
 
 .. _postgresql_settings:
 
 PostgreSQL
 ----------
--  **authentication**:
+-  **postgresql**:
-    -  **superuser**:
+    -  **authentication**:
-        -  **username**: name for the superuser, set during initialization (initdb) and later used by Patroni to connect to the postgres.
+        -  **superuser**:
-        -  **password**: password for the superuser, set during initialization (initdb).
+            -  **username**: name for the superuser, set during initialization (initdb) and later used by Patroni to connect to the postgres.
-    -  **replication**:
+            -  **password**: password for the superuser, set during initialization (initdb).
-        -  **username**: replication username; the user will be created during initialization. Replicas will use this user to access master via streaming replication
+            -  **sslmode**: (optional) maps to the `sslmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLMODE>`__ connection parameter, which allows a client to specify the type of TLS negotiation mode with the server. For more information on how each mode works, please visit the `PostgreSQL documentation <https://www.postgresql.org/docs/current/libpq-ssl.html#LIBPQ-SSL-SSLMODE-STATEMENTS>`__. The default mode is ``prefer``.
-        -  **password**: replication password; the user will be created during initialization.
+            -  **sslkey**: (optional) maps to the `sslkey <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLKEY>`__ connection parameter, which specifies the location of the secret key used with the client's certificate.
-    -  **rewind**:
+            -  **sslpassword**: (optional) maps to the `sslpassword <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLPASSWORD>`__ connection parameter, which specifies the password for the secret key specified in ``sslkey``.
-        -  **username**: name for the user for ``pg_rewind``; the user will be created during initialization of postgres 11+ and all necessary `permissions <https://www.postgresql.org/docs/11/app-pgrewind.html#id-1.9.5.8.8>`__ will be granted.
+            -  **sslcert**: (optional) maps to the `sslcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCERT>`__ connection parameter, which specifies the location of the client certificate.
-        -  **password**: password for the user for ``pg_rewind``; the user will be created during initialization.
+            -  **sslrootcert**: (optional) maps to the `sslrootcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLROOTCERT>`__ connection parameter, which specifies the location of a file containing one ore more certificate authorities (CA) certificates that the client will use to verify a server's certificate.
--  **callbacks**: callback scripts to run on certain actions. Patroni will pass the action, role and cluster name. (See scripts/aws.py as an example of how to write them.)
+            -  **sslcrl**: (optional) maps to the `sslcrl <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRL>`__ connection parameter, which specifies the location of a file containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
-        -  **on\_reload**: run this script when configuration reload is triggered.
+            -  **sslcrldir**: (optional) maps to the `sslcrldir <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRLDIR>`__ connection parameter, which specifies the location of a directory with files containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
-        -  **on\_restart**: run this script when the postgres restarts (without changing role).
+            -  **gssencmode**: (optional) maps to the `gssencmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-GSSENCMODE>`__ connection parameter, which determines whether or with what priority a secure GSS TCP/IP connection will be negotiated with the server
-        -  **on\_role\_change**: run this script when the postgres is being promoted or demoted.
+            -  **channel_binding**: (optional) maps to the `channel_binding <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-CHANNEL-BINDING>`__ connection parameter, which controls the client's use of channel binding.
-        -  **on\_start**: run this script when the postgres starts.
+        -  **replication**:
-        -  **on\_stop**: run this script when the postgres stops.
+            -  **username**: replication username; the user will be created during initialization. Replicas will use this user to access the replication source via streaming replication
--  **connect\_address**: IP address + port through which Postgres is accessible from other nodes and applications.
+            -  **password**: replication password; the user will be created during initialization.
--  **create\_replica\_methods**: an ordered list of the create methods for turning a Patroni node into a new replica.
+            -  **sslmode**: (optional) maps to the `sslmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLMODE>`__ connection parameter, which allows a client to specify the type of TLS negotiation mode with the server. For more information on how each mode works, please visit the `PostgreSQL documentation <https://www.postgresql.org/docs/current/libpq-ssl.html#LIBPQ-SSL-SSLMODE-STATEMENTS>`__. The default mode is ``prefer``.
-   "basebackup" is the default method; other methods are assumed to refer to scripts, each of which is configured as its
+            -  **sslkey**: (optional) maps to the `sslkey <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLKEY>`__ connection parameter, which specifies the location of the secret key used with the client's certificate.
-   own config item. See :ref:`custom replica creation methods documentation <custom_replica_creation>` for further explanation.
+            -  **sslpassword**: (optional) maps to the `sslpassword <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLPASSWORD>`__ connection parameter, which specifies the password for the secret key specified in ``sslkey``.
--  **data\_dir**: The location of the Postgres data directory, either :ref:`existing <existing_data>` or to be initialized by Patroni.
+            -  **sslcert**: (optional) maps to the `sslcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCERT>`__ connection parameter, which specifies the location of the client certificate.
--  **config\_dir**: The location of the Postgres configuration directory, defaults to the data directory. Must be writable by Patroni.
+            -  **sslrootcert**: (optional) maps to the `sslrootcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLROOTCERT>`__ connection parameter, which specifies the location of a file containing one ore more certificate authorities (CA) certificates that the client will use to verify a server's certificate.
--  **bin\_dir**: Path to PostgreSQL binaries (pg_ctl, pg_rewind, pg_basebackup, postgres). The default value is an empty string meaning that PATH environment variable will be used to find the executables.
+            -  **sslcrl**: (optional) maps to the `sslcrl <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRL>`__ connection parameter, which specifies the location of a file containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
--  **listen**: IP address + port that Postgres listens to; must be accessible from other nodes in the cluster, if you're using streaming replication. Multiple comma-separated addresses are permitted, as long as the port component is appended after to the last one with a colon, i.e. ``listen: 127.0.0.1,127.0.0.2:5432``. Patroni will use the first address from this list to establish local connections to the PostgreSQL node.
+            -  **sslcrldir**: (optional) maps to the `sslcrldir <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRLDIR>`__ connection parameter, which specifies the location of a directory with files containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
--  **use\_unix\_socket**: specifies that Patroni should prefer to use unix sockets to connect to the cluster. Default value is ``false``. If ``unix_socket_directories`` is defined, Patroni will use the first suitable value from it to connect to the cluster and fallback to tcp if nothing is suitable. If ``unix_socket_directories`` is not specified in ``postgresql.parameters``, Patroni will assume that the default value should be used and omit ``host`` from the connection parameters.
+            -  **gssencmode**: (optional) maps to the `gssencmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-GSSENCMODE>`__ connection parameter, which determines whether or with what priority a secure GSS TCP/IP connection will be negotiated with the server
--  **pgpass**: path to the `.pgpass <https://www.postgresql.org/docs/current/static/libpq-pgpass.html>`__ password file. Patroni creates this file before executing pg\_basebackup, the post_init script and under some other circumstances. The location must be writable by Patroni.
+            -  **channel_binding**: (optional) maps to the `channel_binding <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-CHANNEL-BINDING>`__ connection parameter, which controls the client's use of channel binding.
--  **recovery\_conf**: additional configuration settings written to recovery.conf when configuring follower.
+        -  **rewind**:
--  **custom\_conf** : path to an optional custom ``postgresql.conf`` file, that will be used in place of ``postgresql.base.conf``. The file must exist on all cluster nodes, be readable by PostgreSQL and will be included from its location on the real ``postgresql.conf``. Note that Patroni will not monitor this file for changes, nor backup it. However, its settings can still be overridden by Patroni's own configuration facilities - see :ref:`dynamic configuration <dynamic_configuration>` for details.
+            -  **username**: name for the user for ``pg_rewind``; the user will be created during initialization of postgres 11+ and all necessary `permissions <https://www.postgresql.org/docs/11/app-pgrewind.html#id-1.9.5.8.8>`__ will be granted.
--  **parameters**: list of configuration settings for Postgres. Many of these are required for replication to work.
+            -  **password**: password for the user for ``pg_rewind``; the user will be created during initialization.
--  **pg\_hba**: list of lines that Patroni will use to generate ``pg_hba.conf``. This parameter has higher priority than ``bootstrap.pg_hba``. Together with :ref:`dynamic configuration <dynamic_configuration>` it simplifies management of ``pg_hba.conf``.
+            -  **sslmode**: (optional) maps to the `sslmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLMODE>`__ connection parameter, which allows a client to specify the type of TLS negotiation mode with the server. For more information on how each mode works, please visit the `PostgreSQL documentation <https://www.postgresql.org/docs/current/libpq-ssl.html#LIBPQ-SSL-SSLMODE-STATEMENTS>`__. The default mode is ``prefer``.
-        -  **- host all all 0.0.0.0/0 md5**.
+            -  **sslkey**: (optional) maps to the `sslkey <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLKEY>`__ connection parameter, which specifies the location of the secret key used with the client's certificate.
-        -  **- host replication replicator 127.0.0.1/32 md5**: A line like this is required for replication.
+            -  **sslpassword**: (optional) maps to the `sslpassword <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLPASSWORD>`__ connection parameter, which specifies the password for the secret key specified in ``sslkey``.
--  **pg\_ident**: list of lines that Patroni will use to generate ``pg_ident.conf``. Together with :ref:`dynamic configuration <dynamic_configuration>` it simplifies management of ``pg_ident.conf``.
+            -  **sslcert**: (optional) maps to the `sslcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCERT>`__ connection parameter, which specifies the location of the client certificate.
-        -  **- mapname1 systemname1 pguser1**.
+            -  **sslrootcert**: (optional) maps to the `sslrootcert <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLROOTCERT>`__ connection parameter, which specifies the location of a file containing one ore more certificate authorities (CA) certificates that the client will use to verify a server's certificate.
-        -  **- mapname1 systemname2 pguser2**.
+            -  **sslcrl**: (optional) maps to the `sslcrl <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRL>`__ connection parameter, which specifies the location of a file containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
--  **pg\_ctl\_timeout**: How long should pg_ctl wait when doing ``start``, ``stop`` or ``restart``. Default value is 60 seconds.
+            -  **sslcrldir**: (optional) maps to the `sslcrldir <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-SSLCRLDIR>`__ connection parameter, which specifies the location of a directory with files containing a certificate revocation list. A client will reject connecting to any server that has a certificate present in this list.
--  **use\_pg\_rewind**: try to use pg\_rewind on the former leader when it joins cluster as a replica.
+            -  **gssencmode**: (optional) maps to the `gssencmode <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-GSSENCMODE>`__ connection parameter, which determines whether or with what priority a secure GSS TCP/IP connection will be negotiated with the server
--  **remove\_data\_directory\_on\_rewind\_failure**: If this option is enabled, Patroni will remove the PostgreSQL data directory and recreate the replica. Otherwise it will try to follow the new leader. Default value is **false**.
+            -  **channel_binding**: (optional) maps to the `channel_binding <https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNECT-CHANNEL-BINDING>`__ connection parameter, which controls the client's use of channel binding.
--  **remove\_data\_directory\_on\_diverged\_timelines**: Patroni will remove the PostgreSQL data directory and recreate the replica if it notices that timelines are diverging and the former master can not start streaming from the new master. This option is useful when ``pg_rewind`` can not be used. Default value is **false**.
+    -  **callbacks**: callback scripts to run on certain actions. Patroni will pass the action, role and cluster name. (See scripts/aws.py as an example of how to write them.)
--  **replica\_method**: for each create_replica_methods other than basebackup, you would add a configuration section of the same name. At a minimum, this should include "command" with a full path to the actual script to be executed. Other configuration parameters will be passed along to the script in the form "parameter=value".
+            -  **on\_reload**: run this script when configuration reload is triggered.
+            -  **on\_restart**: run this script when the postgres restarts (without changing role).
+            -  **on\_role\_change**: run this script when the postgres is being promoted or demoted.
+            -  **on\_start**: run this script when the postgres starts.
+            -  **on\_stop**: run this script when the postgres stops.
+    -  **connect\_address**: IP address + port through which Postgres is accessible from other nodes and applications.
+    -  **proxy\_address**: IP address + port through which a connection pool (e.g. pgbouncer) running next to Postgres is accessible. The value is written to the member key in DCS as ``proxy_url`` and could be used/useful for service discovery.
+    -  **create\_replica\_methods**: an ordered list of the create methods for turning a Patroni node into a new replica.
+       "basebackup" is the default method; other methods are assumed to refer to scripts, each of which is configured as its
+       own config item. See :ref:`custom replica creation methods documentation <custom_replica_creation>` for further explanation.
+    -  **data\_dir**: The location of the Postgres data directory, either :ref:`existing <existing_data>` or to be initialized by Patroni.
+    -  **config\_dir**: The location of the Postgres configuration directory, defaults to the data directory. Must be writable by Patroni.
+    -  **bin\_dir**: Path to PostgreSQL binaries (pg_ctl, pg_rewind, pg_basebackup, postgres). The default value is an empty string meaning that PATH environment variable will be used to find the executables.
+    -  **listen**: IP address + port that Postgres listens to; must be accessible from other nodes in the cluster, if you're using streaming replication. Multiple comma-separated addresses are permitted, as long as the port component is appended after to the last one with a colon, i.e. ``listen: 127.0.0.1,127.0.0.2:5432``. Patroni will use the first address from this list to establish local connections to the PostgreSQL node.
+    -  **use\_unix\_socket**: specifies that Patroni should prefer to use unix sockets to connect to the cluster. Default value is ``false``. If ``unix_socket_directories`` is defined, Patroni will use the first suitable value from it to connect to the cluster and fallback to tcp if nothing is suitable. If ``unix_socket_directories`` is not specified in ``postgresql.parameters``, Patroni will assume that the default value should be used and omit ``host`` from the connection parameters.
+    -  **use\_unix\_socket\_repl**: specifies that Patroni should prefer to use unix sockets for replication user cluster connection. Default value is ``false``. If ``unix_socket_directories`` is defined, Patroni will use the first suitable value from it to connect to the cluster and fallback to tcp if nothing is suitable. If ``unix_socket_directories`` is not specified in ``postgresql.parameters``, Patroni will assume that the default value should be used and omit ``host`` from the connection parameters.
+    -  **pgpass**: path to the `.pgpass <https://www.postgresql.org/docs/current/static/libpq-pgpass.html>`__ password file. Patroni creates this file before executing pg\_basebackup, the post_init script and under some other circumstances. The location must be writable by Patroni.
+    -  **recovery\_conf**: additional configuration settings written to recovery.conf when configuring follower.
+    -  **custom\_conf** : path to an optional custom ``postgresql.conf`` file, that will be used in place of ``postgresql.base.conf``. The file must exist on all cluster nodes, be readable by PostgreSQL and will be included from its location on the real ``postgresql.conf``. Note that Patroni will not monitor this file for changes, nor backup it. However, its settings can still be overridden by Patroni's own configuration facilities - see :ref:`dynamic configuration <dynamic_configuration>` for details.
+    -  **parameters**: list of configuration settings for Postgres. Many of these are required for replication to work.
+    -  **pg\_hba**: list of lines that Patroni will use to generate ``pg_hba.conf``. This parameter has higher priority than ``bootstrap.pg_hba``. Together with :ref:`dynamic configuration <dynamic_configuration>` it simplifies management of ``pg_hba.conf``.
+            -  **- host all all 0.0.0.0/0 md5**.
+            -  **- host replication replicator 127.0.0.1/32 md5**: A line like this is required for replication.
+    -  **pg\_ident**: list of lines that Patroni will use to generate ``pg_ident.conf``. Together with :ref:`dynamic configuration <dynamic_configuration>` it simplifies management of ``pg_ident.conf``.
+            -  **- mapname1 systemname1 pguser1**.
+            -  **- mapname1 systemname2 pguser2**.
+    -  **pg\_ctl\_timeout**: How long should pg_ctl wait when doing ``start``, ``stop`` or ``restart``. Default value is 60 seconds.
+    -  **use\_pg\_rewind**: try to use pg\_rewind on the former leader when it joins cluster as a replica.
+    -  **remove\_data\_directory\_on\_rewind\_failure**: If this option is enabled, Patroni will remove the PostgreSQL data directory and recreate the replica. Otherwise it will try to follow the new leader. Default value is **false**.
+    -  **remove\_data\_directory\_on\_diverged\_timelines**: Patroni will remove the PostgreSQL data directory and recreate the replica if it notices that timelines are diverging and the former primary can not start streaming from the new primary. This option is useful when ``pg_rewind`` can not be used. While performing timelines divergence check on PostgreSQL v10 and older Patroni will try to connect with replication credential to the "postgres" database. Hence, such access should be allowed in the  pg_hba.conf. Default value is **false**.
+    -  **replica\_method**: for each create_replica_methods other than basebackup, you would add a configuration section of the same name. At a minimum, this should include "command" with a full path to the actual script to be executed. Other configuration parameters will be passed along to the script in the form "parameter=value".
+    -  **pre\_promote**: a fencing script that executes during a failover after acquiring the leader lock but before promoting the replica. If the script exits with a non-zero code, Patroni does not promote the replica and removes the leader key from DCS.
+
+.. _restapi_settings:
 
 REST API
 --------
--  **connect\_address**: IP address (or hostname) and port, to access the Patroni's REST API. All the members of the cluster must be able to connect to this address, so unless the Patroni setup is intended for a demo inside the localhost, this address must be a non "localhost" or loopback addres (ie: "localhost" or "127.0.0.1"). It can serve as a endpoint for HTTP health checks (read below about the "listen" REST API parameter), and also for user queries (either directly or via the REST API), as well as for the health checks done by the cluster members during leader elections (for example, to determine whether the master is still running, or if there is a node which has a WAL position that is ahead of the one doing the query; etc.) The connect_address is put in the member key in DCS, making it possible to translate the member name into the address to connect to its REST API.
+-  **restapi**:
+        -  **connect\_address**: IP address (or hostname) and port, to access the Patroni's :ref:`REST API <rest_api>`. All the members of the cluster must be able to connect to this address, so unless the Patroni setup is intended for a demo inside the localhost, this address must be a non "localhost" or loopback address (ie: "localhost" or "127.0.0.1"). It can serve as an endpoint for HTTP health checks (read below about the "listen" REST API parameter), and also for user queries (either directly or via the REST API), as well as for the health checks done by the cluster members during leader elections (for example, to determine whether the leader is still running, or if there is a node which has a WAL position that is ahead of the one doing the query; etc.) The connect_address is put in the member key in DCS, making it possible to translate the member name into the address to connect to its REST API.
 
--  **listen**: IP address (or hostname) and port that Patroni will listen to for the REST API - to provide also the same health checks and cluster messaging between the participating nodes, as described above. to provide health-check information for HAProxy (or any other load balancer capable of doing a HTTP "OPTION" or "GET" checks).
+        -  **listen**: IP address (or hostname) and port that Patroni will listen to for the REST API - to provide also the same health checks and cluster messaging between the participating nodes, as described above. to provide health-check information for HAProxy (or any other load balancer capable of doing a HTTP "OPTION" or "GET" checks).
 
--  **Optional**:
+        -  **authentication**: (optional)
-        -  **authentication**:
             -  **username**: Basic-auth username to protect unsafe REST API endpoints.
             -  **password**: Basic-auth password to protect unsafe REST API endpoints.
+        -  **certfile**: (optional): Specifies the file with the certificate in the PEM format. If the certfile is not specified or is left empty, the API server will work without SSL.
+        -  **keyfile**: (optional): Specifies the file with the secret key in the PEM format.
+        -  **keyfile\_password**: (optional): Specifies a password for decrypting the keyfile.
+        -  **cafile**: (optional): Specifies the file with the CA_BUNDLE with certificates of trusted CAs to use while verifying client certs.
+        -  **ciphers**: (optional): Specifies the permitted cipher suites (e.g. "ECDHE-RSA-AES256-GCM-SHA384:DHE-RSA-AES256-GCM-SHA384:ECDHE-RSA-AES128-GCM-SHA256:DHE-RSA-AES128-GCM-SHA256:!SSLv1:!SSLv2:!SSLv3:!TLSv1:!TLSv1.1")
+        -  **verify\_client**: (optional): ``none`` (default), ``optional`` or ``required``. When ``none`` REST API will not check client certificates. When ``required`` client certificates are required for all REST API calls. When ``optional`` client certificates are required for all unsafe REST API endpoints. When ``required`` is used, then client authentication succeeds, if the certificate signature verification succeeds.  For ``optional`` the client cert will only be checked for ``PUT``, ``POST``, ``PATCH``, and ``DELETE`` requests.
+        -  **allowlist**: (optional): Specifies the set of hosts that are allowed to call unsafe REST API endpoints. The single element could be a host name, an IP address or a network address using CIDR notation. By default ``allow all`` is used. In case if ``allowlist`` or ``allowlist_include_members`` are set, anything that is not included is rejected.
+        -  **allowlist\_include\_members**: (optional): If set to ``true`` it allows accessing unsafe REST API endpoints from other cluster members registered in DCS (IP address or hostname is taken from the members ``api_url``). Be careful, it might happen that OS will use a different IP for outgoing connections.
+        -  **http\_extra\_headers**: (optional): HTTP headers let the REST API server pass additional information with an HTTP response.
+        -  **https\_extra\_headers**: (optional): HTTPS headers let the REST API server pass additional information with an HTTP response when TLS is enabled. This will also pass additional information set in ``http_extra_headers``.
 
-        -  **certfile**: Specifies the file with the certificate in the PEM format. If the certfile is not specified or is left empty, the API server will work without SSL.
+Here is an example of both **http_extra_headers** and **https_extra_headers**:
-        -  **keyfile**: Specifies the file with the secret key in the PEM format.
+
+.. code:: YAML
+
+        restapi:
+          listen: <listen>
+          connect_address: <connect_address>
+          authentication:
+            username: <username>
+            password: <password>
+          http_extra_headers:
+            'X-Frame-Options': 'SAMEORIGIN'
+            'X-XSS-Protection': '1; mode=block'
+            'X-Content-Type-Options': 'nosniff'
+          cafile: <ca file>
+          certfile: <cert>
+          keyfile: <key>
+          https_extra_headers:
+            'Strict-Transport-Security': 'max-age=31536000; includeSubDomains'
 
 .. _patronictl_settings:
 
 CTL
 ---
--  **Optional**:
+- **ctl**: (optional)
     -  **insecure**: Allow connections to REST API without verifying SSL certs.
-    -  **cacert**: Specifices the file with the CA_BUNDLE file or directory with certificates of trusted CAs to use while verifying REST API SSL certs.
+    -  **cacert**: Specifies the file with the CA_BUNDLE file or directory with certificates of trusted CAs to use while verifying REST API SSL certs. If not provided patronictl will use the value provided for REST API "cafile" parameter.
-    -  **certfile**: Specifies the file with the certificate in the PEM format to use while verifying REST API SSL certs. If not provided patronictl will use the value provided for REST API "certfile" parameter.
+    -  **certfile**: Specifies the file with the client certificate in the PEM format. If not provided patronictl will use the value provided for REST API "certfile" parameter.
-
+    -  **keyfile**: Specifies the file with the client secret key in the PEM format. If not provided patronictl will use the value provided for REST API "keyfile" parameter.
-ZooKeeper
+    -  **keyfile\_password**: Specifies a password for decrypting the keyfile. If not provided patronictl will use the value provided for REST API "keyfile\_password" parameter.
-----------
--  **hosts**: list of ZooKeeper cluster members in format: ['host1:port1', 'host2:port2', 'etc...'].
 
 Watchdog
 --------
 - **mode**: ``off``, ``automatic`` or ``required``. When ``off`` watchdog is disabled. When ``automatic`` watchdog will be used if available, but ignored if it is not. When ``required`` the node will not become a leader unless watchdog can be successfully enabled.
 - **device**: Path to watchdog device. Defaults to ``/dev/watchdog``.
 - **safety_margin**: Number of seconds of safety margin between watchdog triggering and leader key expiration.
+
+.. _tags_settings:
+
+Tags
+----
+- **nofailover**: ``true`` or ``false``, controls whether this node is allowed to participate in the leader race and become a leader. Defaults to ``false``
+- **clonefrom**: ``true`` or ``false``. If set to ``true`` other nodes might prefer to use this node for bootstrap (take ``pg_basebackup`` from). If there are several nodes with ``clonefrom`` tag set to ``true`` the node to bootstrap from will be chosen randomly. The default value is ``false``.
+- **noloadbalance**: ``true`` or ``false``. If set to ``true`` the node will return HTTP Status Code 503 for the ``GET /replica`` REST API health-check and therefore will be excluded from the load-balancing. Defaults to ``false``.
+- **replicatefrom**: The IP address/hostname of another replica. Used to support cascading replication.
+- **nosync**: ``true`` or ``false``. If set to ``true`` the node will never be selected as a synchronous replica.
+
+In addition to these predefined tags, you can also add your own ones:
+
+- **key1**: ``true``
+- **key2**: ``false``
+- **key3**: ``1.4``
+- **key4**: ``"RandomString"``
+
+Tags are visible in the :ref:`REST API <rest_api>` and ``patronictl list`` You can also check for an instance health using these tags. If the tag isn't defined for an instance, or if the respective value doesn't match the querying value, it will return HTTP Status Code 503.

docs/citus.rst

@@ -0,0 +1,352 @@
+.. _citus:
+
+Citus support
+=============
+
+Patroni makes it extremely simple to deploy `Multi-Node Citus`__ clusters.
+
+__ https://docs.citusdata.com/en/stable/installation/multi_node.html
+
+TL;DR
+-----
+
+There are only a few simple rules you need to follow:
+
+1. Citus extension must be available on all nodes. Absolute minimum supported
+   Citus version is 10.0, but, to take all benefits from transparent
+   switchovers and restarts of workers we recommend using at least Citus 11.2.
+2. Cluster name (``scope``) must be the same for all Citus nodes!
+3. Superuser credentials must be the same on coordinator and all worker
+   nodes, and ``pg_hba.conf`` should allow superuser access between all nodes.
+4. :ref:`REST API <restapi_settings>` access should be allowed from worker
+   nodes to the coordinator. E.g., credentials should be the same and if
+   configured, client certificates from worker nodes must be accepted by the
+   coordinator.
+5. Add the following section to the ``patroni.yaml``:
+
+.. code:: YAML
+
+        citus:
+          group: X  # 0 for coordinator and 1, 2, 3, etc for workers
+          database: citus  # must be the same on all nodes
+
+
+After that you just need to start Patroni and it will handle the rest:
+
+1. ``citus`` extension will be automatically added to ``shared_preload_libraries``.
+2. If ``max_prepared_transactions`` isn't explicitly set in the global
+   :ref:`dynamic configuration <dynamic_configuration>` Patroni will
+   automatically set it to ``2*max_connections``.
+3. The ``citus.database`` will be automatically created followed by ``CREATE EXTENSION citus``.
+4. Current superuser :ref:`credentials <postgresql_settings>` will be added to the ``pg_dist_authinfo``
+   table to allow cross-node communication. Don't forget to update them if
+   later you decide to change superuser username/password/sslcert/sslkey!
+5. The coordinator primary node will automatically discover worker primary
+   nodes and add them to the ``pg_dist_node`` table using the
+   ``citus_add_node()`` function.
+6. Patroni will also maintain ``pg_dist_node`` in case failover/switchover
+   on the coordinator or worker clusters occurs.
+
+patronictl
+----------
+
+Coordinator and worker clusters are physically different PostgreSQL/Patroni
+clusters that are just logically groupped together using Citus. Therefore in
+most cases it is not possible to manage them as a single entity.
+
+It results in two major differences in ``patronictl`` behaviour when
+``patroni.yaml`` has the ``citus`` section comparing with the usual:
+
+1. The ``list`` and the ``topology`` by default output all members of the Citus
+   formation (coordinators and workers). The new column ``Group`` indicates
+   which Citus group they belong to.
+2. For all ``patronictl`` commands the new option is introduced, named
+   ``--group``. For some commands the default value for the group might be
+   taken from the ``patroni.yaml``. For example, ``patronictl pause`` will
+   enable the maintenance mode by default for the ``group`` that is set in the
+   ``citus`` section, but for example for ``patronictl  switchover`` or
+   ``patronictl remove`` the group must be explicitly specified.
+
+An example of ``patronictl list`` output for the Citus cluster::
+
+    postgres@coord1:~$ patronictl list demo
+    + Citus cluster: demo ----------+--------------+---------+----+-----------+
+    | Group | Member  | Host        | Role         | State   | TL | Lag in MB |
+    +-------+---------+-------------+--------------+---------+----+-----------+
+    |     0 | coord1  | 172.27.0.10 | Replica      | running |  1 |         0 |
+    |     0 | coord2  | 172.27.0.6  | Sync Standby | running |  1 |         0 |
+    |     0 | coord3  | 172.27.0.4  | Leader       | running |  1 |           |
+    |     1 | work1-1 | 172.27.0.8  | Sync Standby | running |  1 |         0 |
+    |     1 | work1-2 | 172.27.0.2  | Leader       | running |  1 |           |
+    |     2 | work2-1 | 172.27.0.5  | Sync Standby | running |  1 |         0 |
+    |     2 | work2-2 | 172.27.0.7  | Leader       | running |  1 |           |
+    +-------+---------+-------------+--------------+---------+----+-----------+
+
+If we add the ``--group`` option, the output will change to::
+
+    postgres@coord1:~$ patronictl list demo --group 0
+    + Citus cluster: demo (group: 0, 7179854923829112860) -----------+
+    | Member | Host        | Role         | State   | TL | Lag in MB |
+    +--------+-------------+--------------+---------+----+-----------+
+    | coord1 | 172.27.0.10 | Replica      | running |  1 |         0 |
+    | coord2 | 172.27.0.6  | Sync Standby | running |  1 |         0 |
+    | coord3 | 172.27.0.4  | Leader       | running |  1 |           |
+    +--------+-------------+--------------+---------+----+-----------+
+
+    postgres@coord1:~$ patronictl list demo --group 1
+    + Citus cluster: demo (group: 1, 7179854923881963547) -----------+
+    | Member  | Host       | Role         | State   | TL | Lag in MB |
+    +---------+------------+--------------+---------+----+-----------+
+    | work1-1 | 172.27.0.8 | Sync Standby | running |  1 |         0 |
+    | work1-2 | 172.27.0.2 | Leader       | running |  1 |           |
+    +---------+------------+--------------+---------+----+-----------+
+
+Citus worker switchover
+-----------------------
+
+When a switchover is orchestrated for a Citus worker node, Citus offers the
+opportunity to make the switchover close to transparent for an application.
+Because the application connects to the coordinator, which in turn connects to
+the worker nodes, then it is possible with Citus to `pause` the SQL traffic on
+the coordinator for the shards hosted on a worker node. The switchover then
+happens while the traffic is kept on the coordinator, and resumes as soon as a
+new primary worker node is ready to accept read-write queries.
+
+An example of ``patronictl switchover`` on the worker cluster::
+
+    postgres@coord1:~$ patronictl switchover demo
+    + Citus cluster: demo ----------+--------------+---------+----+-----------+
+    | Group | Member  | Host        | Role         | State   | TL | Lag in MB |
+    +-------+---------+-------------+--------------+---------+----+-----------+
+    |     0 | coord1  | 172.27.0.10 | Replica      | running |  1 |         0 |
+    |     0 | coord2  | 172.27.0.6  | Sync Standby | running |  1 |         0 |
+    |     0 | coord3  | 172.27.0.4  | Leader       | running |  1 |           |
+    |     1 | work1-1 | 172.27.0.8  | Leader       | running |  1 |           |
+    |     1 | work1-2 | 172.27.0.2  | Sync Standby | running |  1 |         0 |
+    |     2 | work2-1 | 172.27.0.5  | Sync Standby | running |  1 |         0 |
+    |     2 | work2-2 | 172.27.0.7  | Leader       | running |  1 |           |
+    +-------+---------+-------------+--------------+---------+----+-----------+
+    Citus group: 2
+    Primary [work2-2]:
+    Candidate ['work2-1'] []:
+    When should the switchover take place (e.g. 2022-12-22T08:02 )  [now]:
+    Current cluster topology
+    + Citus cluster: demo (group: 2, 7179854924063375386) -----------+
+    | Member  | Host       | Role         | State   | TL | Lag in MB |
+    +---------+------------+--------------+---------+----+-----------+
+    | work2-1 | 172.27.0.5 | Sync Standby | running |  1 |         0 |
+    | work2-2 | 172.27.0.7 | Leader       | running |  1 |           |
+    +---------+------------+--------------+---------+----+-----------+
+    Are you sure you want to switchover cluster demo, demoting current primary work2-2? [y/N]: y
+    2022-12-22 07:02:40.33003 Successfully switched over to "work2-1"
+    + Citus cluster: demo (group: 2, 7179854924063375386) ------+
+    | Member  | Host       | Role    | State   | TL | Lag in MB |
+    +---------+------------+---------+---------+----+-----------+
+    | work2-1 | 172.27.0.5 | Leader  | running |  1 |           |
+    | work2-2 | 172.27.0.7 | Replica | stopped |    |   unknown |
+    +---------+------------+---------+---------+----+-----------+
+
+    postgres@coord1:~$ patronictl list demo
+    + Citus cluster: demo ----------+--------------+---------+----+-----------+
+    | Group | Member  | Host        | Role         | State   | TL | Lag in MB |
+    +-------+---------+-------------+--------------+---------+----+-----------+
+    |     0 | coord1  | 172.27.0.10 | Replica      | running |  1 |         0 |
+    |     0 | coord2  | 172.27.0.6  | Sync Standby | running |  1 |         0 |
+    |     0 | coord3  | 172.27.0.4  | Leader       | running |  1 |           |
+    |     1 | work1-1 | 172.27.0.8  | Leader       | running |  1 |           |
+    |     1 | work1-2 | 172.27.0.2  | Sync Standby | running |  1 |         0 |
+    |     2 | work2-1 | 172.27.0.5  | Leader       | running |  2 |           |
+    |     2 | work2-2 | 172.27.0.7  | Sync Standby | running |  2 |         0 |
+    +-------+---------+-------------+--------------+---------+----+-----------+
+
+And this is how it looks on the coordinator side::
+
+    # The worker primary notifies the coordinator that it is going to execute "pg_ctl stop".
+    2022-12-22 07:02:38,636 DEBUG: query("BEGIN")
+    2022-12-22 07:02:38,636 DEBUG: query("SELECT pg_catalog.citus_update_node(3, '172.27.0.7-demoted', 5432, true, 10000)")
+    # From this moment all application traffic on the coordinator to the worker group 2 is paused.
+
+    # The future worker primary notifies the coordinator that it acquired the leader lock in DCS and about to run "pg_ctl promote".
+    2022-12-22 07:02:40,085 DEBUG: query("SELECT pg_catalog.citus_update_node(3, '172.27.0.5', 5432)")
+
+    # The new worker primary just finished promote and notifies coordinator that it is ready to accept read-write traffic.
+    2022-12-22 07:02:41,485 DEBUG: query("COMMIT")
+    # From this moment the application traffic on the coordinator to the worker group 2 is unblocked.
+
+Peek into DCS
+-------------
+
+The Citus cluster (coordinator and workers) are stored in DCS as a fleet of
+Patroni clusters logically grouped together::
+
+    /service/batman/              # scope=batman
+    /service/batman/0/            # citus.group=0, coordinator
+    /service/batman/0/initialize
+    /service/batman/0/leader
+    /service/batman/0/members/
+    /service/batman/0/members/m1
+    /service/batman/0/members/m2
+    /service/batman/1/            # citus.group=1, worker
+    /service/batman/1/initialize
+    /service/batman/1/leader
+    /service/batman/1/members/
+    /service/batman/1/members/m3
+    /service/batman/1/members/m4
+    ...
+
+Such an approach was chosen because for most DCS it becomes possible to fetch
+the entire Citus cluster with a single recursive read request. Only Citus
+coordinator nodes are reading the whole tree, because they have to discover
+worker nodes. Worker nodes are reading only the subtree for their own group and
+in some cases they could read the subtree of the coordinator group.
+
+Citus on Kubernetes
+-------------------
+
+Since Kubernetes doesn't support hierarchical structures we had to include the
+citus group to all K8s objects Patroni creates::
+
+    batman-0-leader  # the leader config map for the coordinator
+    batman-0-config  # the config map holding initialize, config, and history "keys"
+    ...
+    batman-1-leader  # the leader config map for worker group 1
+    batman-1-config
+    ...
+
+I.e., the naming pattern is: ``${scope}-${citus.group}-${type}``.
+
+All Kubernetes objects are discovered by Patroni using the `label selector`__,
+therefore all Pods with Patroni&Citus and Endpoints/ConfigMaps must have
+similar labels, and Patroni must be configured to use them using Kubernetes
+:ref:`settings <kubernetes_settings>` or :ref:`environment variables
+<kubernetes_environment>`.
+
+__ https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/#label-selectors
+
+A couple of examples of Patroni configuration using Pods environment variables:
+
+1. for the coordinator cluster
+
+.. code:: YAML
+
+        apiVersion: v1
+        kind: Pod
+        metadata:
+          labels:
+            application: patroni
+            citus-group: "0"
+            citus-type: coordinator
+            cluster-name: citusdemo
+          name: citusdemo-0-0
+          namespace: default
+        spec:
+          containers:
+          - env:
+            - name: PATRONI_SCOPE
+              value: citusdemo
+            - name: PATRONI_NAME
+              valueFrom:
+                fieldRef:
+                  apiVersion: v1
+                  fieldPath: metadata.name
+            - name: PATRONI_KUBERNETES_POD_IP
+              valueFrom:
+                fieldRef:
+                  apiVersion: v1
+                  fieldPath: status.podIP
+            - name: PATRONI_KUBERNETES_NAMESPACE
+              valueFrom:
+                fieldRef:
+                  apiVersion: v1
+                  fieldPath: metadata.namespace
+            - name: PATRONI_KUBERNETES_LABELS
+              value: '{application: patroni}'
+            - name: PATRONI_CITUS_DATABASE
+              value: citus
+            - name: PATRONI_CITUS_GROUP
+              value: "0"
+
+2. for the worker cluster from the group 2
+
+.. code:: YAML
+
+        apiVersion: v1
+        kind: Pod
+        metadata:
+          labels:
+            application: patroni
+            citus-group: "2"
+            citus-type: worker
+            cluster-name: citusdemo
+          name: citusdemo-2-0
+          namespace: default
+        spec:
+          containers:
+          - env:
+            - name: PATRONI_SCOPE
+              value: citusdemo
+            - name: PATRONI_NAME
+              valueFrom:
+                fieldRef:
+                  apiVersion: v1
+                  fieldPath: metadata.name
+            - name: PATRONI_KUBERNETES_POD_IP
+              valueFrom:
+                fieldRef:
+                  apiVersion: v1
+                  fieldPath: status.podIP
+            - name: PATRONI_KUBERNETES_NAMESPACE
+              valueFrom:
+                fieldRef:
+                  apiVersion: v1
+                  fieldPath: metadata.namespace
+            - name: PATRONI_KUBERNETES_LABELS
+              value: '{application: patroni}'
+            - name: PATRONI_CITUS_DATABASE
+              value: citus
+            - name: PATRONI_CITUS_GROUP
+              value: "2"
+
+As you may noticed, both examples have ``citus-group`` label set. This label
+allows Patroni to identify object as belonging to a certain Citus group. In
+addition to that, there is also ``PATRONI_CITUS_GROUP`` environment variable,
+which has the same value as the ``citus-group`` label. When Patroni creates
+new Kubernetes objects ConfigMaps or Endpoints, it automatically puts the
+``citus-group: ${env.PATRONI_CITUS_GROUP}`` label on them:
+
+.. code:: YAML
+
+        apiVersion: v1
+        kind: ConfigMap
+        metadata:
+          name: citusdemo-0-leader  # Is generated as ${env.PATRONI_SCOPE}-${env.PATRONI_CITUS_GROUP}-leader
+          labels:
+            application: patroni    # Is set from the ${env.PATRONI_KUBERNETES_LABELS}
+            cluster-name: citusdemo # Is automatically set from the ${env.PATRONI_SCOPE}
+            citus-group: '0'        # Is automatically set from the ${env.PATRONI_CITUS_GROUP}
+
+You can find a complete example of Patroni deployment on Kubernetes with Citus
+support in the `kubernetes`__ folder of the Patroni repository.
+
+__ https://github.com/zalando/patroni/tree/master/kubernetes
+
+There are two important files for you:
+
+1. Dockerfile.citus
+2. citus_k8s.yaml
+
+Citus upgrades and PostgreSQL major upgrades
+--------------------------------------------
+
+First, please read about upgrading Citus version in the `documentation`__.
+There is one minor change in the process. When executing upgrade, you have to
+use ``patronictl restart`` instead of ``systemctl restart`` to restart
+PostgreSQL.
+
+__ https://docs.citusdata.com/en/latest/admin_guide/upgrading_citus.html
+
+The PostgreSQL major upgrade with Citus is a bit more complex. You will have to
+combine techniques used in the Citus documentation about major upgrades and
+Patroni documentation about :ref:`PostgreSQL major upgrade<major_upgrade>`.
+Please keep in mind that Citus cluster consists of many Patroni clusters
+(coordinator and workers) and they all have to be upgraded independently.

docs/conf.py

@@ -194,4 +194,7 @@ intersphinx_mapping = {'https://docs.python.org/': None}
 # A possibility to have an own stylesheet, to add new rules or override existing ones
 # For the latter case, the CSS specificity of the rules should be higher than the default ones
 def setup(app):
-   app.add_stylesheet("custom.css")
+    if hasattr(app, 'add_css_file'):
+        app.add_css_file('custom.css')
+    else:
+        app.add_stylesheet('custom.css')

docs/dcs_failsafe_mode.rst

@@ -0,0 +1,63 @@
+.. _dcs_failsafe_mode:
+
+DCS Failsafe Mode
+=================
+
+The problem
+-----------
+
+Patroni is heavily relying on Distributed Configuration Store (DCS) to solve the task of leader elections and detect network partitioning. That is, the node is allowed to run Postgres as the primary only if it can update the leader lock in DCS. In case the update of the leader lock fails, Postgres is immediately demoted and started as read-only. Depending on which DCS is used, the chances of hitting the "problem" differ. For example, with Etcd which is only used for Patroni, chances are close to zero, while with K8s API (backed by Etcd) it could be observed more frequently.
+
+
+Reasons for the current implementation
+---------------------------------------
+
+The leader lock update failure could be caused by two main reasons:
+
+1. Network partitioning
+2. DCS being down
+
+In general, it is impossible to distinguish between these two from a single node, and therefore Patroni assumes the worst case - network partitioning. In the case of a partitioned network, other nodes of the Patroni cluster may successfully grab the leader lock and promote Postgres to primary. In order to avoid a split-brain, the old primary is demoted before the leader lock expires.
+
+
+DCS Failsafe Mode
+-----------------
+
+We introduce a new special option, the ``failsafe_mode``. It could be enabled only via global configuration stored in the DCS ``/config`` key. If the failsafe mode is enabled and the leader lock update in DCS failed due to reasons different from the version/value/index mismatch, Postgres may continue to run as a primary if it can access all known members of the cluster via Patroni REST API.
+
+
+Low-level implementation details
+--------------------------------
+
+- We introduce a new, permanent key in DCS, named ``/failsafe``.
+- The ``/failsafe`` key contains all known members of the given Patroni cluster at a given time.
+- The current leader maintains the ``/failsafe`` key.
+- The member is allowed to participate in the leader race and become the new leader only if it is present in the ``/failsafe`` key.
+- If the cluster consists of a single node the ``/failsafe`` key will contain a single member.
+- In the case of DCS "outage" the existing primary connects to all members presented in the ``/failsafe`` key via the ``POST /failsafe`` REST API and may continue to run as the primary if all replicas acknowledge it.
+- If one of the members doesn't respond, the primary is demoted.
+- Replicas are using incoming ``POST /failsafe`` REST API requests as an indicator that the primary is still alive. This information is cached for ``ttl`` seconds.
+
+
+F.A.Q.
+------
+
+- Why MUST the current primary see ALL other members? Can’t we rely on quorum here?
+
+  This is a great question! The problem is that the view on the quorum might be different from the perspective of DCS and Patroni. While DCS nodes must be evenly distributed across availability zones, there is no such rule for Patroni, and more importantly, there is no mechanism for introducing and enforcing such a rule. If the majority of Patroni nodes ends up in the losing part of the partitioned network (including primary) while minority nodes are in the winning part, the primary must be demoted. Only checking ALL other members allows detecting such a situation.
+
+- What if node/pod gets terminated while DCS is down?
+
+  If DCS isn’t accessible, the check “are ALL other cluster members accessible?” is executed every cycle of the heartbeat loop (every ``loop_wait`` seconds). If pod/node is terminated, the check will fail and Postgres will be demoted to a read-only and will not recover until DCS is restored.
+
+- What if all members of the Patroni cluster are lost while DCS is down?
+
+  Patroni could be configured to create the new replica from the backup even when the cluster doesn't have a leader. But, if the new member isn't present in the ``/failsafe`` key, it will not be able to grab the leader lock and promote.
+  
+- What will happen if the primary lost access to DCS while replicas didn't?
+
+  The primary will execute the failsafe code and contact all known replicas. These replicas will use this information as an indicator that the primary is alive and will not start the leader race even if the leader lock in DCS has expired.
+
+- How to enable the Failsafe Mode?
+
+  Before enabling the ``failsafe_mode`` please make sure that Patroni version on all members is up-to-date. After that, you can use either the ``PATCH /config`` :ref:`REST API <rest_api>` or ``patronictl edit-config -s failsafe_mode=true``

docs/dynamic_configuration.rst

@@ -14,13 +14,15 @@ Patroni configuration is stored in the DCS (Distributed Configuration Store). Th
 
 - Local :ref:`configuration <settings>` (patroni.yml).
 	These options are defined in the configuration file and take precedence over dynamic configuration.
-	patroni.yml could be changed and reload in runtime (without restart of Patroni) by sending SIGHUP to the Patroni process or by performing ``POST /reload`` REST-API request.
+	patroni.yml could be changed and reloaded in runtime (without restart of Patroni) by sending SIGHUP to the Patroni process, performing ``POST /reload`` REST-API request or executing ``patronictl reload``.
 
 - Environment :ref:`configuration <environment>`.
 	It is possible to set/override some of the "Local" configuration parameters with environment variables.
 	Environment configuration is very useful when you are running in a dynamic environment and you don't know some of the parameters in advance (for example it's not possible to know your external IP address when you are running inside ``docker``).
 
-Some of the PostgreSQL parameters must hold the same values on the master and the replicas. For those, values set either in the local patroni configuration files or via the environment variables take no effect. To alter or set their values one must change the shared configuration in the DCS. Below is the actual list of such parameters together with the default values:
+The local configuration can be either a single YAML file or a directory. When it is a directory, all YAML files in that directory are loaded one by one in sorted order. In case a key is defined in multiple files, the occurrence in the last file takes precedence.
+
+Some of the PostgreSQL parameters must hold the same values on the primary and the replicas. For those, values set either in the local patroni configuration files or via the environment variables take no effect. To alter or set their values one must change the shared configuration in the DCS. Below is the actual list of such parameters together with the default values:
 
 - max_connections: 100
 - max_locks_per_transaction: 64
@@ -30,11 +32,12 @@ Some of the PostgreSQL parameters must hold the same values on the master and th
 - wal_log_hints: on
 - track_commit_timestamp: off
 
-For the parameters below, PostgreSQL does not require equal values among the master and all the replicas. However, considering the possibility of a replica to become the master at any time, it doesn't really make sense to set them differently; therefore, Patroni restricts setting their values to the Dynamic configuration
+For the parameters below, PostgreSQL does not require equal values among the primary and all the replicas. However, considering the possibility of a replica to become the primary at any time, it doesn't really make sense to set them differently; therefore, Patroni restricts setting their values to the Dynamic configuration
 
 - max_wal_senders: 5
 - max_replication_slots: 5
 - wal_keep_segments: 8
+- wal_keep_size: 128MB
 
 These parameters are validated to ensure they are sane, or meet a minimum value.
 
@@ -76,154 +79,11 @@ Also, the following Patroni configuration options can be changed only dynamicall
 - loop_wait: 10
 - retry_timeouts: 10
 - maximum_lag_on_failover: 1048576
+- max_timelines_history: 0
 - check_timeline: false
 - postgresql.use_slots: true
 
 Upon changing these options, Patroni will read the relevant section of the configuration stored in DCS and change its
 run-time values.
 
-Patroni nodes are dumping the state of the DCS options to disk upon for every change of the configuration into the file ``patroni.dynamic.json`` located in the Postgres data directory. Only the master is allowed to restore these options from the on-disk dump if these are completely absent from the DCS or if they are invalid.
+Patroni nodes are dumping the state of the DCS options to disk upon for every change of the configuration into the file ``patroni.dynamic.json`` located in the Postgres data directory. Only the leader is allowed to restore these options from the on-disk dump if these are completely absent from the DCS or if they are invalid.
-
-REST API
-========
-
-We provide a REST API endpoint for working with dynamic configuration.
-
-GET /config
------------
-Get current version of dynamic configuration.
-
-.. code-block:: bash
-
-	$ curl -s localhost:8008/config | jq .
-	{
-	  "ttl": 30,
-	  "loop_wait": 10,
-	  "retry_timeout": 10,
-	  "maximum_lag_on_failover": 1048576,
-	  "postgresql": {
-	    "use_slots": true,
-	    "use_pg_rewind": true,
-	    "parameters": {
-	      "hot_standby": "on",
-	      "wal_log_hints": "on",
-	      "wal_keep_segments": 8,
-	      "wal_level": "hot_standby",
-	      "max_wal_senders": 5,
-	      "max_replication_slots": 5,
-	      "max_connections": "100"
-	    }
-	  }
-	}
-
-PATCH /config
--------------
-Change existing configuration.
-
-.. code-block:: bash
-
-	$ curl -s -XPATCH -d \
-		'{"loop_wait":5,"ttl":20,"postgresql":{"parameters":{"max_connections":"101"}}}' \
-		http://localhost:8008/config | jq .
-	{
-	  "ttl": 20,
-	  "loop_wait": 5,
-	  "maximum_lag_on_failover": 1048576,
-	  "retry_timeout": 10,
-	  "postgresql": {
-	    "use_slots": true,
-	    "use_pg_rewind": true,
-	    "parameters": {
-	      "hot_standby": "on",
-	      "wal_log_hints": "on",
-	      "wal_keep_segments": 8,
-	      "wal_level": "hot_standby",
-	      "max_wal_senders": 5,
-	      "max_replication_slots": 5,
-	      "max_connections": "101"
-	    }
-	  }
-	}
-
-The above REST API call patches the existing configuration and returns the new configuration.
-
-Let's check that the node processed this configuration. First of all it should start printing log lines every 5 seconds (loop_wait=5). The change of "max_connections" requires a restart, so the "restart_pending" flag should be exposed:
-
-.. code-block:: bash
-
-	$ curl -s http://localhost:8008/patroni | jq .
-	{
-	  "pending_restart": true,
-	  "database_system_identifier": "6287881213849985952",
-	  "postmaster_start_time": "2016-06-13 13:13:05.211 CEST",
-	  "xlog": {
-	    "location": 2197818976
-	  },
-	  "patroni": {
-	    "scope": "batman",
-	    "version": "1.0"
-	  },
-	  "state": "running",
-	  "role": "master",
-	  "server_version": 90503
-	}
-
-Removing parameters:
-
-If you want to remove (reset) some setting just patch it with ``null``:
-
-.. code-block:: bash
-
-	$ curl -s -XPATCH -d \
-		'{"postgresql":{"parameters":{"max_connections":null}}}' \
-		http://localhost:8008/config | jq .
-	{
-	  "ttl": 20,
-	  "loop_wait": 5,
-	  "retry_timeout": 10,
-	  "maximum_lag_on_failover": 1048576,
-	  "postgresql": {
-	    "use_slots": true,
-	    "use_pg_rewind": true,
-	    "parameters": {
-	      "hot_standby": "on",
-	      "unix_socket_directories": ".",
-	      "wal_keep_segments": 8,
-	      "wal_level": "hot_standby",
-	      "wal_log_hints": "on",
-	      "max_wal_senders": 5,
-	      "max_replication_slots": 5
-	    }
-	  }
-	}
-
-Above call removes ``postgresql.parameters.max_connections`` from the dynamic configuration.
-
-PUT /config
------------
-
-It's also possible to perform the full rewrite of an existing dynamic configuration unconditionally:
-
-.. code-block:: bash
-
-	$ curl -s -XPUT -d \
-		'{"maximum_lag_on_failover":1048576,"retry_timeout":10,"postgresql":{"use_slots":true,"use_pg_rewind":true,"parameters":{"hot_standby":"on","wal_log_hints":"on","wal_keep_segments":8,"wal_level":"hot_standby","unix_socket_directories":".","max_wal_senders":5}},"loop_wait":3,"ttl":20}' \
-		http://localhost:8008/config | jq .
-	{
-	  "ttl": 20,
-	  "maximum_lag_on_failover": 1048576,
-	  "retry_timeout": 10,
-	  "postgresql": {
-	    "use_slots": true,
-	    "parameters": {
-	      "hot_standby": "on",
-	      "unix_socket_directories": ".",
-	      "wal_keep_segments": 8,
-	      "wal_level": "hot_standby",
-	      "wal_log_hints": "on",
-	      "max_wal_senders": 5
-	    },
-	    "use_pg_rewind": true
-	  },
-	  "loop_wait": 3
-	}

docs/existing_data.rst

@@ -23,6 +23,23 @@ A Patroni cluster can be started with a data directory from a single-node Postgr
 3. Start Patroni (e.g. ``patroni /etc/patroni/patroni.yml``). It automatically detects that PostgreSQL daemon is already running but its configuration might be out-of-date.
 4. Ask Patroni to restart the node with ``patronictl restart cluster-name node-name``. This step is only required if PostgreSQL configuration is out-of-date.
 
+.. _major_upgrade:
+
+Major Upgrade of PostgreSQL Version
+===================================
+
+The only possible way to do a major upgrade currently is:
+
+1. Stop Patroni
+2. Upgrade PostgreSQL binaries and perform `pg_upgrade <https://www.postgresql.org/docs/current/pgupgrade.html>`_ on the primary node
+3. Update patroni.yml
+4. Remove the initialize key from DCS or wipe complete cluster state from DCS. The second one could be achieved by running ``patronictl remove <cluster-name>``. It is necessary because pg_upgrade runs initdb which actually creates a new database with a new PostgreSQL system identifier.
+5. If you wiped the cluster state in the previous step, you may wish to copy patroni.dynamic.json from old data dir to the new one.  It will help you to retain some PostgreSQL parameters you had set before.
+6. Start Patroni on the primary node.
+7. Upgrade PostgreSQL binaries, update patroni.yml and wipe the data_dir on standby nodes.
+8. Start Patroni on the standby nodes and wait for the replication to complete.
+
+Running pg_upgrade on standby nodes is not supported by PostgreSQL. If you know what you are doing, you can try the rsync procedure described in https://www.postgresql.org/docs/current/pgupgrade.html instead of wiping data_dir on standby nodes. The safest way is however to let Patroni replicate the data for you.
 
 FAQ
 ---

docs/index.rst

@@ -10,6 +10,10 @@ Patroni is a template for you to create your own customized, high-availability s
 
 We call Patroni a "template" because it is far from being a one-size-fits-all or plug-and-play replication system. It will have its own caveats. Use wisely. There are many ways to run high availability with PostgreSQL; for a list, see the `PostgreSQL Documentation <https://wiki.postgresql.org/wiki/Replication,_Clustering,_and_Connection_Pooling>`__.
 
+Currently supported PostgreSQL versions: 9.3 to 15.
+
+**Note to Citus users**: Starting from 3.0 Patroni nicely integrates with `Citus <https://www.citusdata.com>`__. Please check :ref:`Citus support <citus>` page for more information.
+
 **Note to Kubernetes users**: Patroni can run natively on top of Kubernetes. Take a look at the :ref:`Kubernetes <kubernetes>` chapter of the Patroni documentation.
 
 
@@ -18,9 +22,14 @@ We call Patroni a "template" because it is far from being a one-size-fits-all or
    :caption: Contents:
 
    README
+   citus
    dynamic_configuration
+   dcs_failsafe_mode
+   rest_api
+   existing_data
    ENVIRONMENT
    SETTINGS
+   security
    replica_bootstrap
    replication_modes
    pause

docs/kubernetes.rst

@@ -23,10 +23,7 @@ Use ConfigMaps
 In this mode, Patroni will create ConfigMaps instead of Endpoints and store keys inside meta-data of those ConfigMaps.
 Changing the leader takes at least two updates, one to the leader ConfigMap and another to the respective Endpoint.
 
-There are two ways to direct the traffic to the Postgres master:
+To direct the traffic to the Postgres leader you need to configure the Kubernetes Postgres service to use the label selector with the `role_label` (configured in patroni configuration).
-
-- use the `callback script <https://github.com/zalando/patroni/blob/master/kubernetes/callback.py>`_ provided by Patroni
-- configure the Kubernetes Postgres service to use the label selector with the `role_label` (configured in patroni configuration).
 
 Note that in some cases, for instance, when running on OpenShift, there is no alternative to using ConfigMaps.
 
@@ -39,7 +36,7 @@ Examples
 --------
 
 - The `kubernetes <https://github.com/zalando/patroni/tree/master/kubernetes>`__ folder of the Patroni repository contains
-  examples of the Docker image, the Kubernetes manifest and the callback script in order to test Patroni Kubernetes setup.
+  examples of the Docker image, and the Kubernetes manifest to test Patroni Kubernetes setup.
   Note that in the current state it will not be able to use PersistentVolumes because of permission issues.
 
 - You can find the full-featured Docker image that can use Persistent Volumes in the

docs/pause.rst

@@ -6,7 +6,7 @@ Pause/Resume mode for the cluster
 The goal
 --------
 
-Under certain circumstances Patroni needs to temporary step down from managing the cluster, while still retaining the cluster state in DCS. Possible use cases are uncommon activities on the cluster, such as major version upgrades or corruption recovery. During those activities nodes are often started and stopped for the reason unknown to Patroni, some nodes can be even temporary promoted, violating the assumption of running only one master. Therefore, Patroni needs to be able to "detach" from the running cluster, implementing an equivalent of the maintenance mode in Pacemaker.
+Under certain circumstances Patroni needs to temporarily step down from managing the cluster, while still retaining the cluster state in DCS. Possible use cases are uncommon activities on the cluster, such as major version upgrades or corruption recovery. During those activities nodes are often started and stopped for reasons unknown to Patroni, some nodes can be even temporarily promoted, violating the assumption of running only one primary. Therefore, Patroni needs to be able to "detach" from the running cluster, implementing an equivalent of the maintenance mode in Pacemaker.
 
 
 
@@ -17,16 +17,18 @@ When Patroni runs in a paused mode, it does not change the state of PostgreSQL,
 
 - For each node, the member key in DCS is updated with the current information about the cluster. This causes Patroni to run read-only queries on a member node if the member is running.
 
-- For the Postgres master with the leader lock Patroni updates the lock. If the node with the leader lock stops being the master (i.e. is demoted manually), Patroni will release the lock instead of promoting the node back.
+- For the Postgres primary with the leader lock Patroni updates the lock. If the node with the leader lock stops being the primary (i.e. is demoted manually), Patroni will release the lock instead of promoting the node back.
 
-- Manual unscheduled restart, reinitialize and manual failover are allowed. Manual failover is only allowed if the node to failover to is specified. In the paused mode, manual failover does not require a running master node.
+- Manual unscheduled restart, reinitialize and manual failover are allowed. Manual failover is only allowed if the node to failover to is specified. In the paused mode, manual failover does not require a running primary node.
 
-- If 'parallel' masters are detected by Patroni, it emits a warning, but does not demote the masters without the leader lock.
+- If 'parallel' primaries are detected by Patroni, it emits a warning, but does not demote the primary without the leader lock.
 
-- If there is no leader lock in the cluster, the running master acquires the lock. If there is more than one master node, then the first master to acquire the lock wins. If there are no masters altogether, Patroni does not try to promote any replicas. There is an exception in this rule: if there is no leader lock because the old master has demoted itself due to the manual promotion, then only the candidate node mentioned in the promotion request may take the leader lock. When the new leader lock is granted (i.e. after promoting a replica manually), Patroni makes sure the replicas that were streaming from the previous leader will switch to the new one.
+- If there is no leader lock in the cluster, the running primary acquires the lock. If there is more than one primary node, then the first primary to acquire the lock wins. If there are no primary altogether, Patroni does not try to promote any replicas. There is an exception in this rule: if there is no leader lock because the old primary has demoted itself due to the manual promotion, then only the candidate node mentioned in the promotion request may take the leader lock. When the new leader lock is granted (i.e. after promoting a replica manually), Patroni makes sure the replicas that were streaming from the previous leader will switch to the new one.
 
 - When Postgres is stopped, Patroni does not try to start it. When Patroni is stopped, it does not try to stop the Postgres instance it is managing.
 
+- Patroni will not try to remove replication slots that don't represent the other cluster member or are not listed in the configuration of the permanent slots.
+
 User guide
 ----------
 

docs/replica_bootstrap.rst

@@ -24,6 +24,7 @@ arguments to them, i.e. the name of the cluster and the path to the data directo
         <custom_bootstrap_method_name>:
             command: <path_to_custom_bootstrap_script> [param1 [, ...]]
             keep_existing_recovery_conf: False
+            no_params: False
             recovery_conf:
                 recovery_target_action: promote
                 recovery_target_timeline: latest
@@ -40,6 +41,8 @@ in the configuration files, Patroni supplies two cluster-specific ones:
 --datadir
     Path to the data directory of the cluster instance to be bootstrapped
 
+Passing these two additional flags can be disabled by setting a special ``no_params`` parameter to ``True``.
+
 If the bootstrap script returns 0, Patroni tries to configure and start the PostgreSQL instance produced by it. If any
 of the intermediate steps fail, or the script returns a non-zero value, Patroni assumes that the bootstrap has failed,
 cleans up after itself and releases the initialize lock to give another node the opportunity to bootstrap.
@@ -60,7 +63,7 @@ Building replicas
 -----------------
 
 Patroni uses tried and proven ``pg_basebackup`` in order to create new replicas. One downside of it is that it requires
-a running master node. Another one is the lack of 'on-the-fly' compression for the backup data and no built-in cleanup
+a running leader node. Another one is the lack of 'on-the-fly' compression for the backup data and no built-in cleanup
 for outdated backup files. Some people prefer other backup solutions, such as ``WAL-E``, ``pgBackRest``, ``Barman`` and
 others, or simply roll their own scripts. In order to accommodate all those use-cases Patroni supports running custom
 scripts to clone a new replica. Those are configured in the ``postgresql`` configuration block:
@@ -74,7 +77,7 @@ scripts to clone a new replica. Those are configured in the ``postgresql`` confi
             command: <command name>
             keep_data: True
             no_params: True
-            no_master: 1
+            no_leader: 1
 
 example: wal_e
 
@@ -86,7 +89,7 @@ example: wal_e
             - basebackup
         wal_e:
             command: patroni_wale_restore
-            no_master: 1
+            no_leader: 1
             envdir: {{WALE_ENV_DIR}}
             use_iam: 1
         basebackup:
@@ -120,26 +123,30 @@ to execute and any custom parameters that should be passed to that command. All
 --role
     Always 'replica'
 --connstring
-    Connection string to connect to the cluster member to clone from (master or other replica). The user in the
+    Connection string to connect to the cluster member to clone from (primary or other replica). The user in the
     connection string can execute SQL and replication protocol commands.
 
-A special ``no_master`` parameter, if defined, allows Patroni to call the replica creation method even if there is no
+A special ``no_leader`` parameter, if defined, allows Patroni to call the replica creation method even if there is no
-running master or replicas. In that case, an empty string will be passed in a connection string. This is useful for
+running leader or replicas. In that case, an empty string will be passed in a connection string. This is useful for
 restoring the formerly running cluster from the binary backup.
 
-A special ``keep_data`` parameter, if defined, will instuct Patroni to  not clean PGDATA folder before calling restore.
+A special ``keep_data`` parameter, if defined, will instruct Patroni to  not clean PGDATA folder before calling restore.
 
 A special ``no_params`` parameter, if defined, restricts passing parameters to custom command.
 
 A ``basebackup`` method is a special case: it will be used if
 ``create_replica_methods`` is empty, although it is possible
 to list it explicitly among the ``create_replica_methods`` methods. This method initializes a new replica with the
-``pg_basebackup``, the base backup is taken from the master unless there are replicas with ``clonefrom`` tag, in which case one
+``pg_basebackup``, the base backup is taken from the leader unless there are replicas with ``clonefrom`` tag, in which case one
 of such replicas will be used as the origin for pg_basebackup. It works without any configuration; however, it is
 possible to specify a ``basebackup`` configuration section. Same rules as with the other method configuration apply,
 namely, only long (with --) options should be specified there. Not all parameters make sense, if you override a connection
 string or provide an option to created tar-ed or compressed base backups, patroni won't be able to make a replica out
 of it. There is no validation performed on the names or values of the parameters passed to the ``basebackup`` section.
+Also note that in case symlinks are used for the WAL folder it is up to the user to specify the correct ``--waldir``
+path as an option, so that after replica buildup or re-initialization the symlink would persist. This option is supported
+only since v10 though.
+
 You can specify basebackup parameters as either a map (key-value pairs) or a list of elements, where each element
 could be either a key-value pair or a single key (for options that does not receive any values, for instance, ``--verbose``).
 Consider those 2 examples:
@@ -159,6 +166,7 @@ and
         basebackup:
             - verbose
             - max-rate: '100M'
+            - waldir: /pg-wal-mount/external-waldir
 
 If all replica creation methods fail, Patroni will try again all methods in order during the next event loop cycle.
 
@@ -168,10 +176,10 @@ Standby cluster
 ---------------
 
 Another available option is to run a "standby cluster", that contains only of
-standby nodes replicating from some remote master. This type of clusters has:
+standby nodes replicating from some remote node. This type of clusters has:
 
 * "standby leader", that behaves pretty much like a regular cluster leader,
-  except it replicates from a remote master.
+  except it replicates from a remote node.
 
 * cascade replicas, that are replicating from standby leader.
 
@@ -179,6 +187,13 @@ Standby leader holds and updates a leader lock in DCS. If the leader lock
 expires, cascade replicas will perform an election to choose another leader
 from the standbys.
 
+There is no further relationship between the standby cluster and the primary
+cluster it replicates from, in particular, they must not share the same DCS
+scope if they use the same DCS. They do not know anything else from each other
+apart from replication information. Also, the standby cluster is not being
+displayed in ``patronictl list`` or ``patronictl topology`` output on the
+primary cluster.
+
 For the sake of flexibility, you can specify methods of creating a replica and
 recovery WAL records when a cluster is in the "standby mode" by providing
 `create_replica_methods` key in `standby_cluster` section. It is distinct from
@@ -204,4 +219,9 @@ in a patroni configuration:
 Note, that these options will be applied only once during cluster bootstrap,
 and the only way to change them afterwards is through DCS.
 
-If you use replication slots on the standby cluster, you must also create the corresponding replication slot on the primary cluster.  It will not be done automatically by the standby cluster implementation.  You can use Patroni's permenant replication slots feature on the primary cluster to maintain a replication slot with the same name as ``primary_slot_name``, or its default value if ``primary_slot_name`` is not provided.
+Patroni expects to find `postgresql.conf` or `postgresql.conf.backup` in PGDATA
+of the remote primary and will not start if it does not find it after a
+basebackup. If the remote primary keeps its `postgresql.conf` elsewhere, it is
+your responsibility to copy it to PGDATA.
+
+If you use replication slots on the standby cluster, you must also create the corresponding replication slot on the primary cluster.  It will not be done automatically by the standby cluster implementation.  You can use Patroni's permanent replication slots feature on the primary cluster to maintain a replication slot with the same name as ``primary_slot_name``, or its default value if ``primary_slot_name`` is not provided.

docs/replication_modes.rst

@@ -13,7 +13,7 @@ In asynchronous mode the cluster is allowed to lose some committed transactions
 
 The amount of transactions that can be lost is controlled via ``maximum_lag_on_failover`` parameter. Because the primary transaction log position is not sampled in real time, in reality the amount of lost data on failover is worst case bounded by  ``maximum_lag_on_failover`` bytes of transaction log plus the amount that is written in the last ``ttl`` seconds (``loop_wait``/2 seconds in the average case). However typical steady state replication delay is well under a second.
 
-By default, when running leader elections, Patroni does not take into account the current timeline of replicas, what in some cases could be undesirable behavior. You can prevent the node not having the same timeline as a former master become the new leader by changing the value of ``check_timeline`` parameter to ``true``.
+By default, when running leader elections, Patroni does not take into account the current timeline of replicas, what in some cases could be undesirable behavior. You can prevent the node not having the same timeline as a former primary become the new leader by changing the value of ``check_timeline`` parameter to ``true``.
 
 PostgreSQL synchronous replication
 ----------------------------------
@@ -46,7 +46,7 @@ When ``synchronous_mode`` is on and a standby crashes, commits will block until
 
 When it is absolutely necessary to guarantee that each write is stored durably
 on at least two nodes, enable ``synchronous_mode_strict`` in addition to the
-``synchronous_node``. This parameter prevents Patroni from switching off the
+``synchronous_mode``. This parameter prevents Patroni from switching off the
 synchronous replication on the primary when no synchronous standby candidates
 are available. As a downside, the primary is not be available for writes
 (unless the Postgres transaction explicitly turns of ``synchronous_mode``),
@@ -57,11 +57,16 @@ You can ensure that a standby never becomes the synchronous standby by setting `
 
 Synchronous mode can be switched on and off via Patroni REST interface. See :ref:`dynamic configuration <dynamic_configuration>` for instructions.
 
+Note: Because of the way synchronous replication is implemented in PostgreSQL it is still possible to lose transactions even when using ``synchronous_mode_strict``. If the PostgreSQL backend is cancelled while waiting to acknowledge replication (as a result of packet cancellation due to client timeout or backend failure) transaction changes become visible for other backends. Such changes are not yet replicated and may be lost in case of standby promotion.
+
+Synchronous Replication Factor
+------------------------------
+The parameter ``synchronous_node_count`` is used by Patroni to manage number of synchronous standby databases. It is set to 1 by default. It has no effect when ``synchronous_mode`` is set to off. When enabled, Patroni manages precise number of synchronous standby databases based on parameter ``synchronous_node_count`` and adjusts the state in DCS & synchronous_standby_names as members join and leave.
 
 Synchronous mode implementation
 -------------------------------
 
-When in synchronous mode Patroni maintains synchronization state in the DCS, containing the latest primary and current synchronous standby. This state is updated with strict ordering constraints to ensure the following invariants:
+When in synchronous mode Patroni maintains synchronization state in the DCS, containing the latest primary and current synchronous standby databases. This state is updated with strict ordering constraints to ensure the following invariants:
 
 - A node must be marked as the latest leader whenever it can accept write transactions. Patroni crashing or PostgreSQL not shutting down can cause violations of this invariant.
 
@@ -69,9 +74,9 @@ When in synchronous mode Patroni maintains synchronization state in the DCS, con
 
 - A node that is not the leader or current synchronous standby is not allowed to promote itself automatically.
 
-Patroni will only ever assign one standby to ``synchronous_standby_names`` because with multiple candidates it is not possible to know which node was acting as synchronous during the failure.
+Patroni will only assign one or more synchronous standby nodes based on ``synchronous_node_count`` parameter to ``synchronous_standby_names``.
 
-On each HA loop iteration Patroni re-evaluates synchronous standby choice. If the current synchronous standby is connected and has not requested its synchronous status to be removed it remains picked. Otherwise the cluster member available for sync that is furthest ahead in replication is picked.
+On each HA loop iteration Patroni re-evaluates synchronous standby nodes choice. If the current list of synchronous standby nodes are connected and has not requested its synchronous status to be removed it remains picked. Otherwise the cluster member available for sync that is furthest ahead in replication is picked.
 
 
 .. [1] The data is still there, but recovering it requires a manual recovery effort by data recovery specialists. When Patroni is allowed to rewind with ``use_pg_rewind`` the forked timeline will be automatically erased to rejoin the failed primary with the cluster.

docs/rest_api.rst

@@ -0,0 +1,392 @@
+.. _rest_api:
+
+Patroni REST API
+================
+
+Patroni has a rich REST API, which is used by Patroni itself during the leader race, by the ``patronictl`` tool in order to perform failovers/switchovers/reinitialize/restarts/reloads, by HAProxy or any other kind of load balancer to perform HTTP health checks, and of course could also be used for monitoring. Below you will find the list of Patroni REST API endpoints.
+
+Health check endpoints
+----------------------
+For all health check ``GET`` requests Patroni returns a JSON document with the status of the node, along with the HTTP status code. If you don't want or don't need the JSON document, you might consider using the ``HEAD`` or ``OPTIONS`` method instead of ``GET``.
+
+- The following requests to Patroni REST API will return HTTP status code **200** only when the Patroni node is running as the primary with leader lock:
+
+  - ``GET /``
+  - ``GET /primary``
+  - ``GET /read-write``
+
+- ``GET /standby-leader``: returns HTTP status code **200** only when the Patroni node is running as the leader in a :ref:`standby cluster <standby_cluster>`.
+
+- ``GET /leader``: returns HTTP status code **200** when the Patroni node has the leader lock. The major difference from the two previous endpoints is that it doesn't take into account whether PostgreSQL is running as the ``primary`` or the ``standby_leader``.
+
+- ``GET /replica``: replica health check endpoint. It returns HTTP status code **200** only when the Patroni node is in the state ``running``, the role is ``replica`` and ``noloadbalance`` tag is not set.
+
+- ``GET /replica?lag=<max-lag>``: replica check endpoint. In addition to checks from ``replica``, it also checks replication latency and returns status code **200** only when it is below specified value. The key cluster.last_leader_operation from DCS is used for Leader wal position and compute latency on replica for performance reasons. max-lag can be specified in bytes (integer) or in human readable values, for e.g. 16kB, 64MB, 1GB.
+
+  - ``GET /replica?lag=1048576``
+  - ``GET /replica?lag=1024kB``
+  - ``GET /replica?lag=10MB``
+  - ``GET /replica?lag=1GB``
+
+- ``GET /replica?tag_key1=value1&tag_key2=value2``: replica check endpoint. In addition, It will also check for user defined tags ``key1`` and ``key2`` and their respective values in the **tags** section of the yaml configuration management. If the tag isn't defined for an instance, or if the value in the yaml configuration doesn't match the querying value, it will return HTTP Status Code 503.
+
+ In the following requests, since we are checking for the leader or standby-leader status, Patroni doesn't apply any of the user defined tags and they will be ignored.
+  - ``GET /?tag_key1=value1&tag_key2=value2``
+  - ``GET /leader?tag_key1=value1&tag_key2=value2``
+  - ``GET /primary?tag_key1=value1&tag_key2=value2``
+  - ``GET /read-write?tag_key1=value1&tag_key2=value2``
+  - ``GET /standby_leader?tag_key1=value1&tag_key2=value2``
+  - ``GET /standby-leader?tag_key1=value1&tag_key2=value2``
+
+- ``GET /read-only``: like the above endpoint, but also includes the primary.
+
+- ``GET /synchronous`` or ``GET /sync``: returns HTTP status code **200** only when the Patroni node is running as a synchronous standby.
+
+- ``GET /read-only-sync``: like the above endpoint, but also includes the primary.
+
+- ``GET /asynchronous`` or ``GET /async``: returns HTTP status code **200** only when the Patroni node is running as an asynchronous standby.
+
+
+- ``GET /asynchronous?lag=<max-lag>`` or ``GET /async?lag=<max-lag>``: asynchronous standby check endpoint. In addition to checks from ``asynchronous`` or ``async``, it also checks replication latency and returns status code **200** only when it is below specified value. The key cluster.last_leader_operation from DCS is used for Leader wal position and compute latency on replica for performance reasons. max-lag can be specified in bytes (integer) or in human readable values, for e.g. 16kB, 64MB, 1GB.
+
+  - ``GET /async?lag=1048576``
+  - ``GET /async?lag=1024kB``
+  - ``GET /async?lag=10MB``
+  - ``GET /async?lag=1GB``
+
+- ``GET /health``: returns HTTP status code **200** only when PostgreSQL is up and running.
+
+- ``GET /liveness``: returns HTTP status code **200** if Patroni heartbeat loop is properly running and **503** if the last run was more than ``ttl`` seconds ago on the primary or ``2*ttl`` on the replica. Could be used for ``livenessProbe``.
+
+- ``GET /readiness``: returns HTTP status code **200** when the Patroni node is running as the leader or when PostgreSQL is up and running. The endpoint could be used for ``readinessProbe`` when it is not possible to use Kubernetes endpoints for leader elections (OpenShift).
+
+Both, ``readiness`` and ``liveness`` endpoints are very light-weight and not executing any SQL. Probes should be configured in such a way that they start failing about time when the leader key is expiring. With the default value of ``ttl``, which is ``30s`` example probes would look like:
+
+.. code-block:: yaml
+
+    readinessProbe:
+      httpGet:
+        scheme: HTTP
+        path: /readiness
+        port: 8008
+      initialDelaySeconds: 3
+      periodSeconds: 10
+      timeoutSeconds: 5
+      successThreshold: 1
+      failureThreshold: 3
+    livenessProbe:
+      httpGet:
+        scheme: HTTP
+        path: /liveness
+        port: 8008
+      initialDelaySeconds: 3
+      periodSeconds: 10
+      timeoutSeconds: 5
+      successThreshold: 1
+      failureThreshold: 3
+
+
+Monitoring endpoint
+-------------------
+
+The ``GET /patroni`` is used by Patroni during the leader race. It also could be used by your monitoring system. The JSON document produced by this endpoint has the same structure as the JSON produced by the health check endpoints.
+
+.. code-block:: bash
+
+    $ curl -s http://localhost:8008/patroni | jq .
+    {
+      "state": "running",
+      "postmaster_start_time": "2019-09-24 09:22:32.555 CEST",
+      "role": "master",
+      "server_version": 110005,
+      "cluster_unlocked": false,
+      "xlog": {
+        "location": 25624640
+      },
+      "timeline": 3,
+      "database_system_identifier": "6739877027151648096",
+      "patroni": {
+        "version": "1.6.0",
+        "scope": "batman"
+      }
+    }
+
+Cluster status endpoints
+------------------------
+
+- The ``GET /cluster`` endpoint generates a JSON document describing the current cluster topology and state:
+
+.. code-block:: bash
+
+    $ curl -s http://localhost:8008/cluster | jq .
+    {
+      "members": [
+        {
+          "name": "postgresql0",
+          "host": "127.0.0.1",
+          "port": 5432,
+          "role": "leader",
+          "state": "running",
+          "api_url": "http://127.0.0.1:8008/patroni",
+          "timeline": 5,
+          "tags": {
+            "clonefrom": true
+          }
+        },
+        {
+          "name": "postgresql1",
+          "host": "127.0.0.1",
+          "port": 5433,
+          "role": "replica",
+          "state": "running",
+          "api_url": "http://127.0.0.1:8009/patroni",
+          "timeline": 5,
+          "tags": {
+            "clonefrom": true
+          },
+          "lag": 0
+        }
+      ],
+      "scheduled_switchover": {
+        "at": "2019-09-24T10:36:00+02:00",
+        "from": "postgresql0"
+      }
+    }
+
+
+- The ``GET /history`` endpoint provides a view on the history of cluster switchovers/failovers. The format is very similar to the content of history files in the ``pg_wal`` directory. The only difference is the timestamp field showing when the new timeline was created.
+
+.. code-block:: bash
+
+    $ curl -s http://localhost:8008/history | jq .
+    [
+      [
+        1,
+        25623960,
+        "no recovery target specified",
+        "2019-09-23T16:57:57+02:00"
+      ],
+      [
+        2,
+        25624344,
+        "no recovery target specified",
+        "2019-09-24T09:22:33+02:00"
+      ],
+      [
+        3,
+        25624752,
+        "no recovery target specified",
+        "2019-09-24T09:26:15+02:00"
+      ],
+      [
+        4,
+        50331856,
+        "no recovery target specified",
+        "2019-09-24T09:35:52+02:00"
+      ]
+    ]
+
+
+Config endpoint
+---------------
+
+``GET /config``: Get the current version of the dynamic configuration:
+
+.. code-block:: bash
+
+	$ curl -s localhost:8008/config | jq .
+	{
+	  "ttl": 30,
+	  "loop_wait": 10,
+	  "retry_timeout": 10,
+	  "maximum_lag_on_failover": 1048576,
+	  "postgresql": {
+	    "use_slots": true,
+	    "use_pg_rewind": true,
+	    "parameters": {
+	      "hot_standby": "on",
+	      "wal_log_hints": "on",
+	      "wal_level": "hot_standby",
+	      "max_wal_senders": 5,
+	      "max_replication_slots": 5,
+	      "max_connections": "100"
+	    }
+	  }
+	}
+
+
+``PATCH /config``: Change the existing configuration.
+
+.. code-block:: bash
+
+	$ curl -s -XPATCH -d \
+		'{"loop_wait":5,"ttl":20,"postgresql":{"parameters":{"max_connections":"101"}}}' \
+		http://localhost:8008/config | jq .
+	{
+	  "ttl": 20,
+	  "loop_wait": 5,
+	  "maximum_lag_on_failover": 1048576,
+	  "retry_timeout": 10,
+	  "postgresql": {
+	    "use_slots": true,
+	    "use_pg_rewind": true,
+	    "parameters": {
+	      "hot_standby": "on",
+	      "wal_log_hints": "on",
+	      "wal_level": "hot_standby",
+	      "max_wal_senders": 5,
+	      "max_replication_slots": 5,
+	      "max_connections": "101"
+	    }
+	  }
+	}
+
+The above REST API call patches the existing configuration and returns the new configuration.
+
+Let's check that the node processed this configuration. First of all it should start printing log lines every 5 seconds (loop_wait=5). The change of "max_connections" requires a restart, so the "pending_restart" flag should be exposed:
+
+.. code-block:: bash
+
+	$ curl -s http://localhost:8008/patroni | jq .
+	{
+	  "pending_restart": true,
+	  "database_system_identifier": "6287881213849985952",
+	  "postmaster_start_time": "2016-06-13 13:13:05.211 CEST",
+	  "xlog": {
+	    "location": 2197818976
+	  },
+	  "patroni": {
+	    "scope": "batman",
+	    "version": "1.0"
+	  },
+	  "state": "running",
+	  "role": "master",
+	  "server_version": 90503
+	}
+
+Removing parameters:
+
+If you want to remove (reset) some setting just patch it with ``null``:
+
+.. code-block:: bash
+
+	$ curl -s -XPATCH -d \
+		'{"postgresql":{"parameters":{"max_connections":null}}}' \
+		http://localhost:8008/config | jq .
+	{
+	  "ttl": 20,
+	  "loop_wait": 5,
+	  "retry_timeout": 10,
+	  "maximum_lag_on_failover": 1048576,
+	  "postgresql": {
+	    "use_slots": true,
+	    "use_pg_rewind": true,
+	    "parameters": {
+	      "hot_standby": "on",
+	      "unix_socket_directories": ".",
+	      "wal_level": "hot_standby",
+	      "wal_log_hints": "on",
+	      "max_wal_senders": 5,
+	      "max_replication_slots": 5
+	    }
+	  }
+	}
+
+The above call removes ``postgresql.parameters.max_connections`` from the dynamic configuration.
+
+``PUT /config``: It's also possible to perform the full rewrite of an existing dynamic configuration unconditionally:
+
+.. code-block:: bash
+
+	$ curl -s -XPUT -d \
+		'{"maximum_lag_on_failover":1048576,"retry_timeout":10,"postgresql":{"use_slots":true,"use_pg_rewind":true,"parameters":{"hot_standby":"on","wal_log_hints":"on","wal_level":"hot_standby","unix_socket_directories":".","max_wal_senders":5}},"loop_wait":3,"ttl":20}' \
+		http://localhost:8008/config | jq .
+	{
+	  "ttl": 20,
+	  "maximum_lag_on_failover": 1048576,
+	  "retry_timeout": 10,
+	  "postgresql": {
+	    "use_slots": true,
+	    "parameters": {
+	      "hot_standby": "on",
+	      "unix_socket_directories": ".",
+	      "wal_level": "hot_standby",
+	      "wal_log_hints": "on",
+	      "max_wal_senders": 5
+	    },
+	    "use_pg_rewind": true
+	  },
+	  "loop_wait": 3
+	}
+
+
+Switchover and failover endpoints
+---------------------------------
+
+``POST /switchover`` or ``POST /failover``. These endpoints are very similar to each other. There are a couple of minor differences though:
+
+1. The failover endpoint allows to perform a manual failover when there are no healthy nodes, but at the same time it will not allow you to schedule a switchover.
+
+2. The switchover endpoint is the opposite. It works only when the cluster is healthy (there is a leader) and allows to schedule a switchover at a given time.
+
+
+In the JSON body of the ``POST`` request you must specify at least the ``leader`` or ``candidate`` fields and optionally the ``scheduled_at`` field if you want to schedule a switchover at a specific time.
+
+
+Example: perform a failover to the specific node:
+
+.. code-block:: bash
+
+    $ curl -s http://localhost:8009/failover -XPOST -d '{"candidate":"postgresql1"}'
+    Successfully failed over to "postgresql1"
+
+
+Example: schedule a switchover from the leader to any other healthy replica in the cluster at a specific time:
+
+.. code-block:: bash
+
+    $ curl -s http://localhost:8008/switchover -XPOST -d \
+	    '{"leader":"postgresql0","scheduled_at":"2019-09-24T12:00+00"}'
+    Switchover scheduled
+
+
+Depending on the situation the request might finish with a different HTTP status code and body. The status code **200** is returned when the switchover or failover successfully completed. If the switchover was successfully scheduled, Patroni will return HTTP status code **202**. In case something went wrong, the error status code (one of **400**, **412** or **503**) will be returned with some details in the response body. For more information please check the source code of ``patroni/api.py:do_POST_failover()`` method.
+
+- ``DELETE /switchover``: delete the scheduled switchover
+
+The ``POST /switchover`` and ``POST failover`` endpoints are used by ``patronictl switchover`` and ``patronictl failover``, respectively.
+The ``DELETE /switchover`` is used by ``patronictl flush <cluster-name> switchover``.
+
+
+Restart endpoint
+----------------
+
+- ``POST /restart``: You can restart Postgres on the specific node by performing the ``POST /restart`` call. In the JSON body of ``POST`` request it is possible to optionally specify some restart conditions:
+
+  - **restart_pending**: boolean, if set to ``true`` Patroni will restart PostgreSQL only when restart is pending in order to apply some changes in the PostgreSQL config.
+  - **role**: perform restart only if the current role of the node matches with the role from the POST request.
+  - **postgres_version**: perform restart only if the current version of postgres is smaller than specified in the POST request.
+  - **timeout**: how long we should wait before PostgreSQL starts accepting connections. Overrides ``primary_start_timeout``.
+  - **schedule**: timestamp with time zone, schedule the restart somewhere in the future.
+
+- ``DELETE /restart``: delete the scheduled restart
+
+``POST /restart`` and ``DELETE /restart`` endpoints are used by ``patronictl restart`` and ``patronictl flush <cluster-name> restart`` respectively.
+
+
+Reload endpoint
+---------------
+
+The ``POST /reload`` call will order Patroni to re-read and apply the configuration file. This is the equivalent of sending the ``SIGHUP`` signal to the Patroni process. In case you changed some of the Postgres parameters which require a restart (like **shared_buffers**), you still have to explicitly do the restart of Postgres by either calling the ``POST /restart`` endpoint or with the help of ``patronictl restart``.
+
+The reload endpoint is used by ``patronictl reload``.
+
+
+Reinitialize endpoint
+---------------------
+
+``POST /reinitialize``: reinitialize the PostgreSQL data directory on the specified node. It is allowed to be executed only on replicas. Once called, it will remove the data directory and start ``pg_basebackup`` or some alternative :ref:`replica creation method <custom_replica_creation>`.
+
+The call might fail if Patroni is in a loop trying to recover (restart) a failed Postgres. In order to overcome this problem one can specify ``{"force":true}`` in the request body.
+
+The reinitialize endpoint is used by ``patronictl reinit``.

docs/security.rst

@@ -0,0 +1,37 @@
+.. _security:
+
+=======================
+Security Considerations
+=======================
+
+A Patroni cluster has two interfaces to be protected from unauthorized access: the distributed configuration storage (DCS) and the Patroni REST API.
+
+Protecting DCS
+==============
+
+Patroni and patronictl both store and retrieve data to/from the DCS.
+
+Despite DCS doesn't contain any sensitive information, it allows changing some of Patroni/Postgres configuration. Therefore the very first thing that should be protected is DCS itself.
+
+The details of protection depend on the type of DCS used. The authentication and encryption parameters (tokens/basic-auth/client certificates) for the supported types of DCS are covered in :ref:`SETTINGS <bootstrap_settings>`
+
+The general recommendation is to enable TLS for all DCS communication.
+
+Protecting the REST API
+=======================
+
+Protecting the REST API is a more complicated task.
+
+The Patroni REST API is used by Patroni itself during the leader race, by the ``patronictl`` tool in order to perform failovers/switchovers/reinitialize/restarts/reloads, by HAProxy or any other kind of load balancer to perform HTTP health checks, and of course could also be used for monitoring.
+
+From the point of view of security, REST API contains safe (``GET`` requests, only retrieve information) and unsafe (``PUT``, ``POST``, ``PATCH`` and ``DELETE`` requests, change the state of nodes) endpoints.
+
+The unsafe endpoints can be protected with HTTP basic-auth by setting the ``restapi.authentication.username`` and ``restapi.authentication.password`` parameters. There is no way to protect the safe endpoints without enabling TLS.
+
+When TLS for the REST API is enabled and a PKI is established, mutual authentication of the API server and API client is possible for all endpoints.
+
+The ``restapi`` section parameters enable TLS client authentication to the server. Depending on the value of the ``verify_client`` parameter, the API server requires a successful client certificate verification for both safe and unsafe API calls (``verify_client: required``), or only for unsafe API calls (``verify_client: optional``), or for no API calls (``verify_client: none``).
+
+The ``ctl`` section parameters enable TLS server authentication to the client (the ``patronictl`` tool which uses the same config as patroni). Set ``insecure: true`` to disable the server certificate verification by the client. See :ref:`SETTINGS <patronictl_settings>` for a detailed description of the TLS client parameters.
+
+Protecting the PostgreSQL database proper from unauthorized access is beyond the scope of this document and is covered in https://www.postgresql.org/docs/current/client-authentication.html

docs/watchdog.rst

@@ -3,7 +3,7 @@
 Watchdog support
 ================
 
-Having multiple PostgreSQL servers running as master can result in transactions lost due to diverging timelines. This situation is also called a split-brain problem. To avoid split-brain Patroni needs to ensure PostgreSQL will not accept any transaction commits after leader key expires in the DCS. Under normal circumstances Patroni will try to achieve this by stopping PostgreSQL when leader lock update fails for any reason. However, this may fail to happen due to various reasons:
+Having multiple PostgreSQL servers running as primary can result in transactions lost due to diverging timelines. This situation is also called a split-brain problem. To avoid split-brain Patroni needs to ensure PostgreSQL will not accept any transaction commits after leader key expires in the DCS. Under normal circumstances Patroni will try to achieve this by stopping PostgreSQL when leader lock update fails for any reason. However, this may fail to happen due to various reasons:
 
 - Patroni has crashed due to a bug, out-of-memory condition or by being accidentally killed by a system administrator.
 
@@ -13,7 +13,7 @@ Having multiple PostgreSQL servers running as master can result in transactions
 
 To guarantee correct behavior under these conditions Patroni supports watchdog devices. Watchdog devices are software or hardware mechanisms that will reset the whole system when they do not get a keepalive heartbeat within a specified timeframe. This adds an additional layer of fail safe in case usual Patroni split-brain protection mechanisms fail.
 
-Patroni will try to activate the watchdog before promoting PostgreSQL to master. If watchdog activation fails and watchdog mode is ``required`` then the node will refuse to become master. When deciding to participate in leader election Patroni will also check that watchdog configuration will allow it to become leader at all. After demoting PostgreSQL (for example due to a manual failover) Patroni will disable the watchdog again. Watchdog will also be disabled while Patroni is in paused state.
+Patroni will try to activate the watchdog before promoting PostgreSQL to primary. If watchdog activation fails and watchdog mode is ``required`` then the node will refuse to become leader. When deciding to participate in leader election Patroni will also check that watchdog configuration will allow it to become leader at all. After demoting PostgreSQL (for example due to a manual failover) Patroni will disable the watchdog again. Watchdog will also be disabled while Patroni is in paused state.
 
 By default Patroni will set up the watchdog to expire 5 seconds before TTL expires. With the default setup of ``loop_wait=10`` and ``ttl=30`` this gives HA loop at least 15 seconds (``ttl`` - ``safety_margin`` - ``loop_wait``) to complete before the system gets forcefully reset. By default accessing DCS is configured to time out after 10 seconds. This means that when DCS is unavailable, for example due to network issues, Patroni and PostgreSQL will have at least 5 seconds (``ttl`` - ``safety_margin`` - ``loop_wait`` - ``retry_timeout``) to come to a state where all client connections are terminated.
 

extras/confd/conf.d/haproxy.toml

@@ -9,5 +9,5 @@ check_cmd = "/usr/sbin/haproxy -c -f {{ .src }}"
 reload_cmd = "haproxy -f /etc/haproxy/haproxy.cfg -p /var/run/haproxy.pid -D -sf $(cat /var/run/haproxy.pid)"
 
 keys = [
-    "/members/",
+    "/",
 ]

extras/confd/templates/haproxy-citus.tmpl

@@ -0,0 +1,32 @@
+global
+	maxconn 100
+
+defaults
+	log	global
+	mode	tcp
+	retries 2
+	timeout client 30m
+	timeout connect 4s
+	timeout server 30m
+	timeout check 5s
+
+listen stats
+	mode http
+	bind *:7000
+	stats enable
+	stats uri /
+
+listen coordinator
+	bind *:5000
+	option httpchk HEAD /primary
+	http-check expect status 200
+	default-server inter 3s fall 3 rise 2 on-marked-down shutdown-sessions
+{{range gets "/0/members/*"}}	server {{base .Key}} {{$data := json .Value}}{{base (replace (index (split $data.conn_url "/") 2) "@" "/" -1)}} maxconn 100 check check-ssl port {{index (split (index (split $data.api_url "/") 2) ":") 1}} verify required ca-file /etc/ssl/certs/ssl-cert-snakeoil.pem crt /etc/ssl/private/ssl-cert-snakeoil.crt
+{{end}}
+listen workers
+	bind *:5001
+	option httpchk HEAD /primary
+	http-check expect status 200
+	default-server inter 3s fall 3 rise 2 on-marked-down shutdown-sessions
+{{range gets "/*/members/*"}}{{$group := index (split .Key "/") 1}}{{if ne $group "0"}}	server {{base .Key}} {{$data := json .Value}}{{base (replace (index (split $data.conn_url "/") 2) "@" "/" -1)}} maxconn 100 check check-ssl port {{index (split (index (split $data.api_url "/") 2) ":") 1}} verify required ca-file /etc/ssl/certs/ssl-cert-snakeoil.pem crt /etc/ssl/private/ssl-cert-snakeoil.crt
+{{end}}{{end}}

extras/confd/templates/haproxy.tmpl

@@ -16,16 +16,16 @@ listen stats
 	stats enable
 	stats uri /
 
-listen master
+listen primary
 	bind *:5000
-	option httpchk OPTIONS /master
+	option httpchk HEAD /primary
 	http-check expect status 200
 	default-server inter 3s fall 3 rise 2 on-marked-down shutdown-sessions
 {{range gets "/members/*"}}	server {{base .Key}} {{$data := json .Value}}{{base (replace (index (split $data.conn_url "/") 2) "@" "/" -1)}} maxconn 100 check port {{index (split (index (split $data.api_url "/") 2) ":") 1}}
 {{end}}
 listen replicas
 	bind *:5001
-	option httpchk OPTIONS /replica
+	option httpchk HEAD /replica
 	http-check expect status 200
 	default-server inter 3s fall 3 rise 2 on-marked-down shutdown-sessions
 {{range gets "/members/*"}}	server {{base .Key}} {{$data := json .Value}}{{base (replace (index (split $data.conn_url "/") 2) "@" "/" -1)}} maxconn 100 check port {{index (split (index (split $data.api_url "/") 2) ":") 1}}

extras/startup-scripts/README.md

@@ -1,6 +1,6 @@
 # startup scripts for Patroni
 
-This directory contains sample startup scripts for various OSes 
+This directory contains sample startup scripts for various OSes
 and management tools for Patroni.
 
 Scripts supplied:
@@ -10,7 +10,7 @@ Scripts supplied:
 Upstart job for Ubuntu 12.04 or 14.04.  Requires Upstart > 1.4. Intended for systems where Patroni has been installed on a base system, rather than in Docker.
 
 ### patroni.service
-Systemd service file, to be copied to /etc/systemd/system/patroni.service, tested on Centos 7.1 with Patroni installed from pip. 
+Systemd service file, to be copied to /etc/systemd/system/patroni.service, tested on Centos 7.1 with Patroni installed from pip.
 
 ### patroni
 Init.d service file for Debian-like distributions. Copy it to /etc/init.d/, make executable:

extras/startup-scripts/patroni.service

@@ -14,7 +14,8 @@ Group=postgres
 # Read in configuration file if it exists, otherwise proceed
 EnvironmentFile=-/etc/patroni_env.conf
 
-WorkingDirectory=~
+# The default is the user's home directory, and if you want to change it, you must provide an absolute path.
+# WorkingDirectory=/home/sameuser
 
 # Where to send early-startup messages from the server
 # This is normally controlled by the global default set by systemd
@@ -31,14 +32,14 @@ ExecStart=/bin/patroni /etc/patroni.yml
 # Send HUP to reload from patroni.yml
 ExecReload=/bin/kill -s HUP $MAINPID
 
-# only kill the patroni process, not it's children, so it will gracefully stop postgres
+# Only kill the patroni process, not it's children, so it will gracefully stop postgres
 KillMode=process
 
 # Give a reasonable amount of time for the server to start up/shut down
 TimeoutSec=30
 
-# Do not restart the service if it crashes, we want to manually inspect database on failure
+# Restart the service if it crashed
-Restart=no
+Restart=on-failure
 
 [Install]
 WantedBy=multi-user.target

features/archive-restore.py

@@ -0,0 +1,21 @@
+#!/usr/bin/env python
+import os
+import argparse
+import shutil
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--dirname", required=True)
+    parser.add_argument("--pathname", required=True)
+    parser.add_argument("--filename", required=True)
+    parser.add_argument("--mode", required=True, choices=("archive", "restore"))
+    args, _ = parser.parse_known_args()
+
+    full_filename = os.path.join(args.dirname, args.filename)
+    if args.mode == "archive":
+        if not os.path.isdir(args.dirname):
+            os.makedirs(args.dirname)
+        if not os.path.exists(full_filename):
+            shutil.copy(args.pathname, full_filename)
+    else:
+        shutil.copy(full_filename, args.pathname)

features/backup_create.py

@@ -0,0 +1,14 @@
+#!/usr/bin/env python
+import argparse
+import subprocess
+import sys
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--datadir", required=True)
+    parser.add_argument("--dbname", required=True)
+    parser.add_argument("--walmethod", required=True, choices=("fetch", "stream", "none"))
+    args, _ = parser.parse_known_args()
+
+    walmethod = ["-X", args.walmethod] if args.walmethod != "none" else []
+    sys.exit(subprocess.call(["pg_basebackup", "-D", args.datadir, "-c", "fast", "-d", args.dbname] + walmethod))

features/backup_create.sh

@@ -1,22 +0,0 @@
-#!/bin/bash
-
-while getopts ":-:" optchar; do
-    [[ "${optchar}" == "-" ]] || continue
-    case "${OPTARG}" in
-        datadir=* )
-            PGDATA=${OPTARG#*=}
-            ;;
-        dbname=* )
-            DBNAME=${OPTARG#*=}
-            ;;
-        walmethod=* )
-            WALMETHOD=${OPTARG#*=}
-            ;;
-    esac
-done
-
-[[ -z $PGDATA || -z $DBNAME || -z $WALMETHOD ]] && exit 1
-
-[[ $WALMETHOD != "none" ]] && WALMETHOD="-X $WALMETHOD" || WALMETHOD=""
-
-exec pg_basebackup -D $PGDATA $WALMETHOD -c fast -d $DBNAME

features/backup_restore.py

@@ -0,0 +1,11 @@
+#!/usr/bin/env python
+import argparse
+import shutil
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--datadir", required=True)
+    parser.add_argument("--sourcedir", required=True)
+    args, _ = parser.parse_known_args()
+
+    shutil.copytree(args.sourcedir, args.datadir)

features/backup_restore.sh

@@ -1,21 +0,0 @@
-#!/bin/bash
-
-set -x
-
-while getopts ":-:" optchar; do
-    [[ "${optchar}" == "-" ]] || continue
-    case "${OPTARG}" in
-        datadir=* )
-            PGDATA=${OPTARG#*=}
-            ;;
-        sourcedir=* )
-            SOURCE=${OPTARG#*=}
-            ;;
-    esac
-done
-
-[[ -z $PGDATA || -z $SOURCE ]] && exit 1
-
-mkdir -p $(dirname $PGDATA)
-
-exec cp -af $SOURCE $PGDATA

features/basic_replication.feature

@@ -4,7 +4,8 @@ Feature: basic replication
   Scenario: check replication of a single table
     Given I start postgres0
     Then postgres0 is a leader after 10 seconds
-    When I issue a PATCH request to http://127.0.0.1:8008/config with {"ttl": 20, "loop_wait": 2, "synchronous_mode": true}
+    And there is a non empty initialize key in DCS after 15 seconds
+    When I issue a PATCH request to http://127.0.0.1:8008/config with {"ttl": 20, "synchronous_mode": true}
     Then I receive a response code 200
     When I start postgres1
     And I configure and start postgres2 with a tag replicatefrom postgres0
@@ -20,12 +21,40 @@ Feature: basic replication
     And I shut down postgres1
     Then "sync" key in DCS has sync_standby=postgres2 after 10 seconds
     When I start postgres1
-    And "members/postgres1" key in DCS has state=running after 10 seconds
+    Then "members/postgres1" key in DCS has state=running after 10 seconds
-    And I sleep for 2 seconds
+    And Status code on GET http://127.0.0.1:8010/sync is 200 after 3 seconds
-    When I issue a GET request to http://127.0.0.1:8010/sync
+    And Status code on GET http://127.0.0.1:8009/async is 200 after 3 seconds
+
+  Scenario: check stuck sync replica
+    Given I issue a PATCH request to http://127.0.0.1:8008/config with {"pause": true, "maximum_lag_on_syncnode": 15000000, "postgresql": {"parameters": {"synchronous_commit": "remote_apply"}}}
     Then I receive a response code 200
-    When I issue a GET request to http://127.0.0.1:8009/async
+    And I create table on postgres0
+    And table mytest is present on postgres1 after 2 seconds
+    And table mytest is present on postgres2 after 2 seconds
+    When I pause wal replay on postgres2
+    And I load data on postgres0
+    Then "sync" key in DCS has sync_standby=postgres1 after 15 seconds
+    And I resume wal replay on postgres2
+    And Status code on GET http://127.0.0.1:8009/sync is 200 after 3 seconds
+    And Status code on GET http://127.0.0.1:8010/async is 200 after 3 seconds
+    When I issue a PATCH request to http://127.0.0.1:8008/config with {"pause": null, "maximum_lag_on_syncnode": -1, "postgresql": {"parameters": {"synchronous_commit": "on"}}}
     Then I receive a response code 200
+    And I drop table on postgres0
+
+  Scenario: check multi sync replication
+    Given I issue a PATCH request to http://127.0.0.1:8008/config with {"synchronous_node_count": 2}
+    Then I receive a response code 200
+    Then "sync" key in DCS has sync_standby=postgres1,postgres2 after 10 seconds
+    And Status code on GET http://127.0.0.1:8010/sync is 200 after 3 seconds
+    And Status code on GET http://127.0.0.1:8009/sync is 200 after 3 seconds
+    When I issue a PATCH request to http://127.0.0.1:8008/config with {"synchronous_node_count": 1}
+    Then I receive a response code 200
+    And I shut down postgres1
+    Then "sync" key in DCS has sync_standby=postgres2 after 10 seconds
+    When I start postgres1
+    Then "members/postgres1" key in DCS has state=running after 10 seconds
+    And Status code on GET http://127.0.0.1:8010/sync is 200 after 3 seconds
+    And Status code on GET http://127.0.0.1:8009/async is 200 after 3 seconds
 
   Scenario: check the basic failover in synchronous mode
     Given I run patronictl.py pause batman
@@ -35,17 +64,19 @@ Feature: basic replication
     And I run patronictl.py resume batman
     Then I receive a response returncode 0
     And postgres2 role is the primary after 24 seconds
+    And Response on GET http://127.0.0.1:8010/history contains recovery after 10 seconds
     When I issue a PATCH request to http://127.0.0.1:8010/config with {"synchronous_mode": null, "master_start_timeout": 0}
     Then I receive a response code 200
     When I add the table bar to postgres2
     Then table bar is present on postgres1 after 20 seconds
+    And Response on GET http://127.0.0.1:8010/config contains master_start_timeout after 10 seconds
 
   Scenario: check immediate failover when master_start_timeout=0
     Given I kill postmaster on postgres2
     Then postgres1 is a leader after 10 seconds
     And postgres1 role is the primary after 10 seconds
 
-  Scenario: check rejoin of the former master with pg_rewind
+  Scenario: check rejoin of the former primary with pg_rewind
     Given I add the table splitbrain to postgres0
     And I start postgres0
     Then postgres0 role is the secondary after 20 seconds

features/callback.sh

@@ -1,5 +0,0 @@
-#!/bin/bash
-
-[[ "$3" == "master" ]] || exit
-
-PGPASSWORD=zalando psql -h localhost -U postgres -p $1 -w -tAc "SELECT slot_name FROM pg_replication_slots WHERE slot_type = 'logical'" >> data/postgres0/label

features/callback2.py

@@ -0,0 +1,5 @@
+#!/usr/bin/env python
+
+import sys
+with open("data/{0}/{0}_cb.log".format(sys.argv[1]), "a+") as log:
+    log.write(" ".join(sys.argv[-3:]) + "\n")

features/citus.feature

@@ -0,0 +1,72 @@
+Feature: citus
+  We should check that coordinator discovers and registers workers and clients don't have errors when worker cluster switches over
+
+  Scenario: check that worker cluster is registered in the coordinator
+    Given I start postgres0 in citus group 0
+    And I start postgres2 in citus group 1
+    Then postgres0 is a leader in a group 0 after 10 seconds
+    And postgres2 is a leader in a group 1 after 10 seconds
+    When I start postgres1 in citus group 0
+    And I start postgres3 in citus group 1
+    Then replication works from postgres0 to postgres1 after 15 seconds
+    Then replication works from postgres2 to postgres3 after 15 seconds
+    And postgres0 is registered in the coordinator postgres0 as the worker in group 0
+    And postgres2 is registered in the coordinator postgres0 as the worker in group 1
+
+  Scenario: coordinator failover updates pg_dist_node
+    Given I run patronictl.py failover batman --group 0 --candidate postgres1 --force
+    Then postgres1 role is the primary after 10 seconds
+    And replication works from postgres1 to postgres0 after 15 seconds
+    And "sync" key in a group 0 in DCS has sync_standby=postgres0 after 15 seconds
+    And postgres1 is registered in the coordinator postgres1 as the worker in group 0
+    When I run patronictl.py failover batman --group 0 --candidate postgres0 --force
+    Then postgres0 role is the primary after 10 seconds
+    And replication works from postgres0 to postgres1 after 15 seconds
+    And "sync" key in a group 0 in DCS has sync_standby=postgres1 after 15 seconds
+    And postgres0 is registered in the coordinator postgres0 as the worker in group 0
+
+  Scenario: worker switchover doesn't break client queries on the coordinator
+    Given I create a distributed table on postgres0
+    And I start a thread inserting data on postgres0
+    When I run patronictl.py switchover batman --group 1 --force
+    Then I receive a response returncode 0
+    And postgres3 role is the primary after 10 seconds
+    And replication works from postgres3 to postgres2 after 15 seconds
+    And "sync" key in a group 1 in DCS has sync_standby=postgres2 after 15 seconds
+    And postgres3 is registered in the coordinator postgres0 as the worker in group 1
+    And a thread is still alive
+    When I run patronictl.py switchover batman --group 1 --force
+    Then I receive a response returncode 0
+    And postgres2 role is the primary after 10 seconds
+    And replication works from postgres2 to postgres3 after 15 seconds
+    And "sync" key in a group 1 in DCS has sync_standby=postgres3 after 15 seconds
+    And postgres2 is registered in the coordinator postgres0 as the worker in group 1
+    And a thread is still alive
+    When I stop a thread
+    Then a distributed table on postgres0 has expected rows
+
+  Scenario: worker primary restart doesn't break client queries on the coordinator
+    Given I cleanup a distributed table on postgres0
+    And I start a thread inserting data on postgres0
+    When I run patronictl.py restart batman postgres2 --group 1 --force
+    Then I receive a response returncode 0
+    And postgres2 role is the primary after 10 seconds
+    And replication works from postgres2 to postgres3 after 15 seconds
+    And postgres2 is registered in the coordinator postgres0 as the worker in group 1
+    And a thread is still alive
+    When I stop a thread
+    Then a distributed table on postgres0 has expected rows
+
+  Scenario: check that in-flight transaction is rolled back after timeout when other workers need to change pg_dist_node
+    Given I start postgres4 in citus group 2
+    Then postgres4 is a leader in a group 2 after 10 seconds
+    And "members/postgres4" key in a group 2 in DCS has role=master after 3 seconds
+    When I run patronictl.py edit-config batman --group 2 -s ttl=20 --force
+    Then I receive a response returncode 0
+    And I receive a response output "+ttl: 20"
+    When I sleep for 2 seconds
+    Then postgres4 is registered in the coordinator postgres0 as the worker in group 2
+    When I shut down postgres4
+    Then There is a transaction in progress on postgres0 changing pg_dist_node
+    When I run patronictl.py restart batman postgres2 --group 1 --force
+    Then a transaction finishes in 20 seconds

features/custom_bootstrap.feature

@@ -13,5 +13,5 @@ Scenario: make a backup and do a restore into a new cluster
     Given I add the table bar to postgres1
     And I do a backup of postgres1
     When I start postgres2 in a cluster batman2 from backup
-    Then postgres2 is a leader of batman2 after 10 seconds
+    Then postgres2 is a leader of batman2 after 30 seconds
     And table bar is present on postgres2 after 10 seconds

features/dcs_failsafe_mode.feature

@@ -0,0 +1,85 @@
+Feature: dcs failsafe mode
+  We should check the basic dcs failsafe mode functioning
+
+  Scenario: check failsafe mode can be successfully enabled
+    Given I start postgres0
+    And postgres0 is a leader after 10 seconds
+    And I sleep for 3 seconds
+    When I issue a PATCH request to http://127.0.0.1:8008/config with {"loop_wait": 2, "ttl": 20, "retry_timeout": 5, "failsafe_mode": true}
+    Then I receive a response code 200
+    And Response on GET http://127.0.0.1:8008/failsafe contains postgres0 after 10 seconds
+    When I issue a GET request to http://127.0.0.1:8008/failsafe
+    Then I receive a response code 200
+    And I receive a response postgres0 http://127.0.0.1:8008/patroni
+    When I issue a PATCH request to http://127.0.0.1:8008/config with {"postgresql": {"parameters": {"wal_level": "logical"}}}
+    Then I receive a response code 200
+    When I issue a PATCH request to http://127.0.0.1:8008/config with {"slots": {"dcs_slot_0": {"type": "logical", "database": "postgres", "plugin": "test_decoding"}}}
+    Then I receive a response code 200
+
+  @dcs-failsafe
+  Scenario: check one-node cluster is functioning while DCS is down
+    Given DCS is down
+    Then Response on GET http://127.0.0.1:8008/primary contains failsafe_mode_is_active after 12 seconds
+    And postgres0 role is the primary after 10 seconds
+
+  @dcs-failsafe
+  Scenario: check new replica isn't promoted when leader is down and DCS is up
+    Given DCS is up
+    When I do a backup of postgres0
+    And I shut down postgres0
+    When I start postgres1 in a cluster batman from backup with no_leader
+    And I sleep for 2 seconds
+    Then postgres1 role is the replica after 12 seconds
+
+  Scenario: check leader and replica are both in /failsafe key after leader is back
+    Given I start postgres0
+    And I start postgres1
+    Then "members/postgres0" key in DCS has state=running after 10 seconds
+    And "members/postgres1" key in DCS has state=running after 2 seconds
+    And Response on GET http://127.0.0.1:8009/failsafe contains postgres1 after 10 seconds
+    When I issue a GET request to http://127.0.0.1:8009/failsafe
+    Then I receive a response code 200
+    And I receive a response postgres0 http://127.0.0.1:8008/patroni
+    And I receive a response postgres1 http://127.0.0.1:8009/patroni
+
+  @dcs-failsafe
+  @slot-advance
+  Scenario: check leader and replica are functioning while DCS is down
+    Given logical slot dcs_slot_0 is in sync between postgres0 and postgres1 after 10 seconds
+    And DCS is down
+    Then Response on GET http://127.0.0.1:8008/primary contains failsafe_mode_is_active after 12 seconds
+    Then postgres0 role is the primary after 10 seconds
+    And postgres1 role is the replica after 2 seconds
+    And replication works from postgres0 to postgres1 after 10 seconds
+    And I get all changes from logical slot dcs_slot_0 on postgres0
+    And logical slot dcs_slot_0 is in sync between postgres0 and postgres1 after 20 seconds
+
+  @dcs-failsafe
+  Scenario: check primary is demoted when one replica is shut down and DCS is down
+    Given DCS is down
+    And I kill postgres1
+    And I kill postmaster on postgres1
+    And I sleep for 2 seconds
+    Then postgres0 role is the replica after 12 seconds
+
+  @dcs-failsafe
+  Scenario: check known replica is promoted when leader is down and DCS is up
+    Given I shut down postgres0
+    And DCS is up
+    When I start postgres1
+    Then "members/postgres1" key in DCS has state=running after 10 seconds
+    And postgres1 role is the primary after 25 seconds
+
+  @dcs-failsafe
+  Scenario: check three-node cluster is functioning while DCS is down
+    Given I start postgres0
+    And I start postgres2
+    Then "members/postgres2" key in DCS has state=running after 10 seconds
+    And "members/postgres0" key in DCS has state=running after 20 seconds
+    And Response on GET http://127.0.0.1:8008/failsafe contains postgres2 after 10 seconds
+    And replication works from postgres1 to postgres0 after 10 seconds
+    Given DCS is down
+    Then Response on GET http://127.0.0.1:8008/primary contains failsafe_mode_is_active after 12 seconds
+    Then postgres1 role is the primary after 10 seconds
+    And postgres0 role is the replica after 2 seconds
+    And postgres2 role is the replica after 2 seconds

features/environment.py

@@ -1,13 +1,10 @@
 import abc
-import consul
 import datetime
-import etcd
+import glob
-import kazoo.client
-import kazoo.exceptions
 import os
-import psutil
-import psycopg2
 import json
+import psutil
+import re
 import shutil
 import signal
 import six
@@ -18,6 +15,11 @@ import threading
 import time
 import yaml
 
+import patroni.psycopg as psycopg
+
+from patroni.request import PatroniRequest
+from six.moves.BaseHTTPServer import BaseHTTPRequestHandler, HTTPServer
+
 
 @six.add_metaclass(abc.ABCMeta)
 class AbstractController(object):
@@ -101,6 +103,7 @@ class PatroniController(AbstractController):
             self.watchdog = None
 
         self._scope = (custom_config or {}).get('scope', 'batman')
+        self._citus_group = (custom_config or {}).get('citus', {}).get('group')
         self._config = self._make_patroni_test_config(name, custom_config)
         self._closables = []
 
@@ -140,12 +143,24 @@ class PatroniController(AbstractController):
     def _start(self):
         if self.watchdog:
             self.watchdog.start()
+        env = os.environ.copy()
         if isinstance(self._context.dcs_ctl, KubernetesController):
-            self._context.dcs_ctl.create_pod(self._name[8:], self._scope)
+            self._context.dcs_ctl.create_pod(self._name[8:], self._scope, self._citus_group)
-            os.environ['PATRONI_KUBERNETES_POD_IP'] = '10.0.0.' + self._name[-1]
+            env['PATRONI_KUBERNETES_POD_IP'] = '10.0.0.' + self._name[-1]
-        return subprocess.Popen([sys.executable, '-m', 'coverage', 'run',
+        if os.name == 'nt':
-                                '--source=patroni', '-p', 'patroni.py', self._config],
+            env['BEHAVE_DEBUG'] = 'true'
-                                stdout=self._log, stderr=subprocess.STDOUT, cwd=self._work_directory)
+        patroni = subprocess.Popen([sys.executable, '-m', 'coverage', 'run',
+                                   '--source=patroni', '-p', 'patroni.py', self._config], env=env,
+                                   stdout=self._log, stderr=subprocess.STDOUT, cwd=self._work_directory)
+        if os.name == 'nt':
+            patroni.terminate = self.terminate
+        return patroni
+
+    def terminate(self):
+        try:
+            self._context.request_executor.request('POST', self._restapi_url + '/sigterm')
+        except Exception:
+            pass
 
     def stop(self, kill=False, timeout=15, postgres=False):
         if postgres:
@@ -166,21 +181,59 @@ class PatroniController(AbstractController):
         patroni_config_name = self.PATRONI_CONFIG.format(name)
         patroni_config_path = os.path.join(self._output_dir, patroni_config_name)
 
-        with open(patroni_config_name) as f:
+        with open('postgres0.yml') as f:
             config = yaml.safe_load(f)
             config.pop('etcd', None)
 
-        host = config['postgresql']['listen'].split(':')[0]
+        raft_port = os.environ.get('RAFT_PORT')
+        # If patroni_raft_controller is suspended two Patroni members is enough to get a quorum,
+        # therefore we don't want Patroni to join as a voting member when testing dcs_failsafe_mode.
+        if raft_port and not self._output_dir.endswith('dcs_failsafe_mode'):
+            os.environ['RAFT_PORT'] = str(int(raft_port) + 1)
+            config['raft'] = {'data_dir': self._output_dir, 'self_addr': 'localhost:' + os.environ['RAFT_PORT']}
 
+        host = config['restapi']['listen'].rsplit(':', 1)[0]
+        config['restapi']['listen'] = config['restapi']['connect_address'] = '{0}:{1}'.format(host, 8008+int(name[-1]))
+
+        host = config['postgresql']['listen'].rsplit(':', 1)[0]
         config['postgresql']['listen'] = config['postgresql']['connect_address'] = '{0}:{1}'.format(host, self.__PORT)
 
         config['name'] = name
-        config['postgresql']['data_dir'] = self._data_dir
+        config['postgresql']['data_dir'] = self._data_dir.replace('\\', '/')
-        config['postgresql']['use_unix_socket'] = True
+        config['postgresql']['basebackup'] = [{'checkpoint': 'fast'}]
+        config['postgresql']['use_unix_socket'] = os.name != 'nt'  # windows doesn't yet support unix-domain sockets
+        config['postgresql']['use_unix_socket_repl'] = os.name != 'nt'
+        config['postgresql']['pgpass'] = os.path.join(tempfile.gettempdir(), 'pgpass_' + name).replace('\\', '/')
         config['postgresql']['parameters'].update({
-            'logging_collector': 'on', 'log_destination': 'csvlog', 'log_directory': self._output_dir,
+            'logging_collector': 'on', 'log_destination': 'csvlog',
+            'log_directory': self._output_dir.replace('\\', '/'),
             'log_filename': name + '.log', 'log_statement': 'all', 'log_min_messages': 'debug1',
-            'unix_socket_directories': self._data_dir})
+            'shared_buffers': '1MB', 'unix_socket_directories': tempfile.gettempdir().replace('\\', '/')})
+        config['postgresql']['pg_hba'] = [
+            'local all all trust',
+            'local replication all trust',
+            'host replication replicator all md5',
+            'host all all all md5'
+        ]
+
+        if self._context.postgres_supports_ssl and self._context.certfile:
+            config['postgresql']['parameters'].update({
+                'ssl': 'on',
+                'ssl_ca_file': self._context.certfile.replace('\\', '/'),
+                'ssl_cert_file': self._context.certfile.replace('\\', '/'),
+                'ssl_key_file': self._context.keyfile.replace('\\', '/')
+            })
+            for user in config['postgresql'].get('authentication').keys():
+                config['postgresql'].get('authentication', {}).get(user, {}).update({
+                    'sslmode': 'verify-ca',
+                    'sslrootcert': self._context.certfile,
+                    'sslcert': self._context.certfile,
+                    'sslkey': self._context.keyfile
+                })
+            for i, line in enumerate(list(config['postgresql']['pg_hba'])):
+                if line.endswith('md5'):
+                    # we want to verify client cert first and than password
+                    config['postgresql']['pg_hba'][i] = 'hostssl' + line[4:] + ' clientcert=verify-ca'
 
         if 'bootstrap' in config:
             config['bootstrap']['post_bootstrap'] = 'psql -w -c "SELECT 1"'
@@ -190,25 +243,44 @@ class PatroniController(AbstractController):
         if custom_config is not None:
             self.recursive_update(config, custom_config)
 
+        self.recursive_update(config, {
+            'bootstrap': {
+                'dcs': {
+                    'loop_wait': 2,
+                    'postgresql': {
+                        'parameters': {
+                            'wal_keep_segments': 100,
+                            'archive_mode': 'on',
+                            'archive_command': (PatroniPoolController.ARCHIVE_RESTORE_SCRIPT +
+                                                ' --mode archive ' +
+                                                '--dirname {} --filename %f --pathname %p').format(
+                                                    os.path.join(self._work_directory, 'data', 'wal_archive'))
+                        }
+                    }
+                }
+            }
+        })
         if config['postgresql'].get('callbacks', {}).get('on_role_change'):
             config['postgresql']['callbacks']['on_role_change'] += ' ' + str(self.__PORT)
 
         with open(patroni_config_path, 'w') as f:
             yaml.safe_dump(config, f, default_flow_style=False)
 
-        user = config['postgresql'].get('authentication', config['postgresql']).get('superuser', {})
+        self._connkwargs = config['postgresql'].get('authentication', config['postgresql']).get('superuser', {})
-        self._connkwargs = {k: user[n] for n, k in [('username', 'user'), ('password', 'password')] if n in user}
+        self._connkwargs.update({'host': host, 'port': self.__PORT, 'dbname': 'postgres',
-        self._connkwargs.update({'host': host, 'port': self.__PORT, 'database': 'postgres'})
+                                 'user': self._connkwargs.pop('username', None)})
 
         self._replication = config['postgresql'].get('authentication', config['postgresql']).get('replication', {})
-        self._replication.update({'host': host, 'port': self.__PORT, 'database': 'postgres'})
+        self._replication.update({'host': host, 'port': self.__PORT, 'user': self._replication.pop('username', None)})
+        self._restapi_url = 'http://{0}'.format(config['restapi']['connect_address'])
+        if self._context.certfile:
+            self._restapi_url = self._restapi_url.replace('http://', 'https://')
 
         return patroni_config_path
 
     def _connection(self):
         if not self._conn or self._conn.closed != 0:
-            self._conn = psycopg2.connect(**self._connkwargs)
+            self._conn = psycopg.connect(**self._connkwargs)
-            self._conn.autocommit = True
         return self._conn
 
     def _cursor(self):
@@ -221,7 +293,7 @@ class PatroniController(AbstractController):
             cursor = self._cursor()
             cursor.execute(query)
             return cursor
-        except psycopg2.Error:
+        except psycopg.Error:
             if not fail_ok:
                 raise
 
@@ -249,59 +321,27 @@ class PatroniController(AbstractController):
         except Exception:
             return None
 
-    def database_is_running(self):
-        pid = self._get_pid()
-        if not pid:
-            return False
-        try:
-            os.kill(pid, 0)
-        except OSError:
-            return False
-        return True
-
     def patroni_hang(self, timeout):
         hang = ProcessHang(self._handle.pid, timeout)
         self._closables.append(hang)
         hang.start()
 
-    def checkpoint_hang(self, timeout):
-        pid = self._get_pid()
-        if not pid:
-            return False
-        proc = psutil.Process(pid)
-        for child in proc.children():
-            if 'checkpoint' in child.cmdline()[0]:
-                checkpointer = child
-                break
-        else:
-            return False
-        hang = ProcessHang(checkpointer.pid, timeout)
-        self._closables.append(hang)
-        hang.start()
-        return True
-
     def cancel_background(self):
         for obj in self._closables:
             obj.close()
         self._closables = []
 
-    def terminate_backends(self):
-        pid = self._get_pid()
-        if not pid:
-            return False
-        proc = psutil.Process(pid)
-        for p in proc.children():
-            if 'process' not in p.cmdline()[0]:
-                p.terminate()
-
     @property
     def backup_source(self):
-        return 'postgres://{username}:{password}@{host}:{port}/{database}'.format(**self._replication)
+        def escape(value):
+            return re.sub(r'([\'\\ ])', r'\\\1', str(value))
 
-    def backup(self, dest='data/basebackup'):
+        return ' '.join('{0}={1}'.format(k, escape(v)) for k, v in self._replication.items())
-        subprocess.call([PatroniPoolController.BACKUP_SCRIPT, '--walmethod=none',
+
-                         '--datadir=' + os.path.join(self._work_directory, dest),
+    def backup(self, dest=os.path.join('data', 'basebackup')):
-                         '--dbname=' + self.backup_source])
+        subprocess.call(PatroniPoolController.BACKUP_SCRIPT + ['--walmethod=none',
+                        '--datadir=' + os.path.join(self._work_directory, dest),
+                        '--dbname=' + self.backup_source])
 
 
 class ProcessHang(object):
@@ -335,6 +375,7 @@ class AbstractDcsController(AbstractController):
 
     def __init__(self, context, mktemp=True):
         work_directory = mktemp and tempfile.mkdtemp() or None
+        self._paused = False
         super(AbstractDcsController, self).__init__(context, self.name(), work_directory, context.pctl.output_dir)
 
     def _is_accessible(self):
@@ -346,11 +387,22 @@ class AbstractDcsController(AbstractController):
         if self._work_directory:
             shutil.rmtree(self._work_directory)
 
-    def path(self, key=None, scope='batman'):
+    def path(self, key=None, scope='batman', group=None):
-        return self._CLUSTER_NODE.format(scope) + (key and '/' + key or '')
+        citus_group = '/{0}'.format(group) if group is not None else ''
+        return self._CLUSTER_NODE.format(scope) + citus_group + (key and '/' + key or '')
+
+    def start_outage(self):
+        if not self._paused and self._handle:
+            self._handle.suspend()
+            self._paused = True
+
+    def stop_outage(self):
+        if self._paused and self._handle:
+            self._handle.resume()
+            self._paused = False
 
     @abc.abstractmethod
-    def query(self, key, scope='batman'):
+    def query(self, key, scope='batman', group=None):
         """ query for a value of a given key """
 
     @abc.abstractmethod
@@ -375,15 +427,17 @@ class ConsulController(AbstractDcsController):
         super(ConsulController, self).__init__(context)
         os.environ['PATRONI_CONSUL_HOST'] = 'localhost:8500'
         os.environ['PATRONI_CONSUL_REGISTER_SERVICE'] = 'on'
-        self._client = consul.Consul()
         self._config_file = None
 
+        import consul
+        self._client = consul.Consul()
+
     def _start(self):
         self._config_file = self._work_directory + '.json'
         with open(self._config_file, 'wb') as f:
             f.write(b'{"session_ttl_min":"5s","server":true,"bootstrap":true,"advertise_addr":"127.0.0.1"}')
-        return subprocess.Popen(['consul', 'agent', '-config-file', self._config_file, '-data-dir',
+        return psutil.Popen(['consul', 'agent', '-config-file', self._config_file, '-data-dir',
-                                 self._work_directory], stdout=self._log, stderr=subprocess.STDOUT)
+                             self._work_directory], stdout=self._log, stderr=subprocess.STDOUT)
 
     def stop(self, kill=False, timeout=15):
         super(ConsulController, self).stop(kill=kill, timeout=timeout)
@@ -396,11 +450,11 @@ class ConsulController(AbstractDcsController):
         except Exception:
             return False
 
-    def path(self, key=None, scope='batman'):
+    def path(self, key=None, scope='batman', group=None):
-        return super(ConsulController, self).path(key, scope)[1:]
+        return super(ConsulController, self).path(key, scope, group)[1:]
 
-    def query(self, key, scope='batman'):
+    def query(self, key, scope='batman', group=None):
-        _, value = self._client.kv.get(self.path(key, scope))
+        _, value = self._client.kv.get(self.path(key, scope, group))
         return value and value['Value'].decode('utf-8')
 
     def cleanup_service_tree(self):
@@ -410,26 +464,45 @@ class ConsulController(AbstractDcsController):
         super(ConsulController, self).start(max_wait_limit)
 
 
-class EtcdController(AbstractDcsController):
+class AbstractEtcdController(AbstractDcsController):
 
     """ handles all etcd related tasks, used for the tests setup and cleanup """
 
-    def __init__(self, context):
+    def __init__(self, context, client_cls):
-        super(EtcdController, self).__init__(context)
+        super(AbstractEtcdController, self).__init__(context)
-        os.environ['PATRONI_ETCD_HOST'] = 'localhost:2379'
+        self._client_cls = client_cls
-        self._client = etcd.Client(port=2379)
 
     def _start(self):
-        return subprocess.Popen(["etcd", "--debug", "--data-dir", self._work_directory],
+        return psutil.Popen(["etcd", "--enable-v2=true", "--data-dir", self._work_directory],
-                                stdout=self._log, stderr=subprocess.STDOUT)
+                            stdout=self._log, stderr=subprocess.STDOUT)
 
-    def query(self, key, scope='batman'):
+    def _is_running(self):
+        from patroni.dcs.etcd import DnsCachingResolver
+        # if etcd is running, but we didn't start it
         try:
-            return self._client.get(self.path(key, scope)).value
+            self._client = self._client_cls({'host': 'localhost', 'port': 2379, 'retry_timeout': 30,
+                                             'patronictl': 1}, DnsCachingResolver())
+            return True
+        except Exception:
+            return False
+
+
+class EtcdController(AbstractEtcdController):
+
+    def __init__(self, context):
+        from patroni.dcs.etcd import EtcdClient
+        super(EtcdController, self).__init__(context, EtcdClient)
+        os.environ['PATRONI_ETCD_HOST'] = 'localhost:2379'
+
+    def query(self, key, scope='batman', group=None):
+        import etcd
+        try:
+            return self._client.get(self.path(key, scope, group)).value
         except etcd.EtcdKeyNotFound:
             return None
 
     def cleanup_service_tree(self):
+        import etcd
         try:
             self._client.delete(self.path(scope=''), recursive=True)
         except (etcd.EtcdKeyNotFound, etcd.EtcdConnectionFailed):
@@ -437,15 +510,64 @@ class EtcdController(AbstractDcsController):
         except Exception as e:
             assert False, "exception when cleaning up etcd contents: {0}".format(e)
 
-    def _is_running(self):
+
-        # if etcd is running, but we didn't start it
+class Etcd3Controller(AbstractEtcdController):
+
+    def __init__(self, context):
+        from patroni.dcs.etcd3 import Etcd3Client
+        super(Etcd3Controller, self).__init__(context, Etcd3Client)
+        os.environ['PATRONI_ETCD3_HOST'] = 'localhost:2379'
+
+    def query(self, key, scope='batman', group=None):
+        import base64
+        response = self._client.range(self.path(key, scope, group))
+        for k in response.get('kvs', []):
+            return base64.b64decode(k['value']).decode('utf-8') if 'value' in k else None
+
+    def cleanup_service_tree(self):
         try:
-            return bool(self._client.machines)
+            self._client.deleteprefix(self.path(scope=''))
-        except Exception:
+        except Exception as e:
+            assert False, "exception when cleaning up etcd contents: {0}".format(e)
+
+
+class AbstractExternalDcsController(AbstractDcsController):
+
+    def __init__(self, context, mktemp=True):
+        super(AbstractExternalDcsController, self).__init__(context, mktemp)
+        self._wrapper = ['sudo']
+
+    def _start(self):
+        return self._external_pid
+
+    def start_outage(self):
+        if not self._paused:
+            subprocess.call(self._wrapper + ['kill', '-SIGSTOP', self._external_pid])
+            self._paused = True
+
+    def stop_outage(self):
+        if self._paused:
+            subprocess.call(self._wrapper + ['kill', '-SIGCONT', self._external_pid])
+            self._paused = False
+
+    def _has_started(self):
+        return True
+
+    @abc.abstractmethod
+    def process_name():
+        """process name to search with pgrep"""
+
+    def _is_running(self):
+        if not self._handle:
+            self._external_pid = subprocess.check_output(['pgrep', '-nf', self.process_name()]).decode('utf-8').strip()
             return False
+        return True
+
+    def stop(self):
+        pass
 
 
-class KubernetesController(AbstractDcsController):
+class KubernetesController(AbstractExternalDcsController):
 
     def __init__(self, context):
         super(KubernetesController, self).__init__(context)
@@ -454,18 +576,48 @@ class KubernetesController(AbstractDcsController):
         self._label_selector = ','.join('{0}={1}'.format(k, v) for k, v in self._labels.items())
         os.environ['PATRONI_KUBERNETES_LABELS'] = json.dumps(self._labels)
         os.environ['PATRONI_KUBERNETES_USE_ENDPOINTS'] = 'true'
+        os.environ.setdefault('PATRONI_KUBERNETES_BYPASS_API_SERVICE', 'true')
 
-        from kubernetes import client as k8s_client, config as k8s_config
+        from patroni.dcs.kubernetes import k8s_client, k8s_config
-        k8s_config.load_kube_config(context='local')
+        k8s_config.load_kube_config(context=os.environ.setdefault('PATRONI_KUBERNETES_CONTEXT', 'kind-kind'))
         self._client = k8s_client
         self._api = self._client.CoreV1Api()
 
-    def _start(self):
+    def process_name(self):
-        pass
+        return "localkube"
 
-    def create_pod(self, name, scope):
+    def _is_running(self):
+        if not self._handle:
+            context = os.environ.get('PATRONI_KUBERNETES_CONTEXT')
+            if context.startswith('kind-'):
+                container = '{0}-control-plane'.format(context[5:])
+                api_process = 'kube-apiserver'
+            elif context.startswith('k3d-'):
+                container = '{0}-server-0'.format(context)
+                api_process = 'k3s'
+            else:
+                return super(KubernetesController, self)._is_running()
+            try:
+                docker = 'docker'
+                with open(os.devnull, 'w') as null:
+                    if subprocess.call([docker, 'info'], stdout=null, stderr=null) != 0:
+                        raise Exception
+            except Exception:
+                docker = 'podman'
+                with open(os.devnull, 'w') as null:
+                    if subprocess.call([docker, 'info'], stdout=null, stderr=null) != 0:
+                        raise Exception
+            self._wrapper = [docker, 'exec', container]
+            self._external_pid = subprocess.check_output(self._wrapper + ['pidof', api_process]).decode('utf-8').strip()
+            return False
+        return True
+
+    def create_pod(self, name, scope, group=None):
+        self.delete_pod(name)
         labels = self._labels.copy()
         labels['cluster-name'] = scope
+        if group is not None:
+            labels['citus-group'] = str(group)
         metadata = self._client.V1ObjectMeta(namespace=self._namespace, name=name, labels=labels)
         spec = self._client.V1PodSpec(containers=[self._client.V1Container(name=name, image='empty')])
         body = self._client.V1Pod(metadata=metadata, spec=spec)
@@ -473,37 +625,40 @@ class KubernetesController(AbstractDcsController):
 
     def delete_pod(self, name):
         try:
-            self._api.delete_namespaced_pod(name, self._namespace, self._client.V1DeleteOptions())
+            self._api.delete_namespaced_pod(name, self._namespace, body=self._client.V1DeleteOptions())
-        except:
+        except Exception:
             pass
         while True:
             try:
                 self._api.read_namespaced_pod(name, self._namespace)
-            except:
+            except Exception:
                 break
 
-    def query(self, key, scope='batman'):
+    def query(self, key, scope='batman', group=None):
         if key.startswith('members/'):
             pod = self._api.read_namespaced_pod(key[8:], self._namespace)
             return (pod.metadata.annotations or {}).get('status', '')
         else:
             try:
-                e = self._api.read_namespaced_endpoints(scope + ('' if key == 'leader' else '-' + key), self._namespace)
+                if group is not None:
-                if key == 'leader':
+                    scope = '{0}-{1}'.format(scope, group)
+                ep = scope + {'leader': '', 'history': '-config', 'initialize': '-config'}.get(key, '-' + key)
+                e = self._api.read_namespaced_endpoints(ep, self._namespace)
+                if key != 'sync':
                     return e.metadata.annotations[key]
                 else:
                     return json.dumps(e.metadata.annotations)
-            except:
+            except Exception:
                 return None
 
     def cleanup_service_tree(self):
         try:
             self._api.delete_collection_namespaced_pod(self._namespace, label_selector=self._label_selector)
-        except:
+        except Exception:
             pass
         try:
             self._api.delete_collection_namespaced_endpoints(self._namespace, label_selector=self._label_selector)
-        except:
+        except Exception:
             pass
 
         while True:
@@ -511,11 +666,8 @@ class KubernetesController(AbstractDcsController):
             if len(result.items) < 1:
                 break
 
-    def _is_running(self):
-        return True
 
-
+class ZooKeeperController(AbstractExternalDcsController):
-class ZooKeeperController(AbstractDcsController):
 
     """ handles all zookeeper related tasks, used for the tests setup and cleanup """
 
@@ -523,18 +675,22 @@ class ZooKeeperController(AbstractDcsController):
         super(ZooKeeperController, self).__init__(context, False)
         if export_env:
             os.environ['PATRONI_ZOOKEEPER_HOSTS'] = "'localhost:2181'"
+
+        import kazoo.client
         self._client = kazoo.client.KazooClient()
 
-    def _start(self):
+    def process_name(self):
-        pass  # TODO: implement later
+        return "zookeeper"
 
-    def query(self, key, scope='batman'):
+    def query(self, key, scope='batman', group=None):
+        import kazoo.exceptions
         try:
-            return self._client.get(self.path(key, scope))[0].decode('utf-8')
+            return self._client.get(self.path(key, scope, group))[0].decode('utf-8')
         except kazoo.exceptions.NoNodeError:
             return None
 
     def cleanup_service_tree(self):
+        import kazoo.exceptions
         try:
             self._client.delete(self.path(scope=''), recursive=True)
         except (kazoo.exceptions.NoNodeError):
@@ -543,6 +699,9 @@ class ZooKeeperController(AbstractDcsController):
             assert False, "exception when cleaning up zookeeper contents: {0}".format(e)
 
     def _is_running(self):
+        if not super(ZooKeeperController, self)._is_running():
+            return False
+
         # if zookeeper is running, but we didn't start it
         if self._client.connected:
             return True
@@ -552,16 +711,79 @@ class ZooKeeperController(AbstractDcsController):
             return False
 
 
+class MockExhibitor(BaseHTTPRequestHandler):
+
+    def do_GET(self):
+        self.send_response(200)
+        self.end_headers()
+        self.wfile.write(b'{"servers":["127.0.0.1"],"port":2181}')
+
+    def log_message(self, fmt, *args):
+        pass
+
+
 class ExhibitorController(ZooKeeperController):
 
     def __init__(self, context):
         super(ExhibitorController, self).__init__(context, False)
-        os.environ.update({'PATRONI_EXHIBITOR_HOSTS': 'localhost', 'PATRONI_EXHIBITOR_PORT': '8181'})
+        port = 8181
+        exhibitor = HTTPServer(('', port), MockExhibitor)
+        exhibitor.daemon_thread = True
+        exhibitor_thread = threading.Thread(target=exhibitor.serve_forever)
+        exhibitor_thread.daemon = True
+        exhibitor_thread.start()
+        os.environ.update({'PATRONI_EXHIBITOR_HOSTS': 'localhost', 'PATRONI_EXHIBITOR_PORT': str(port)})
+
+
+class RaftController(AbstractDcsController):
+
+    CONTROLLER_ADDR = 'localhost:1234'
+    PASSWORD = '12345'
+
+    def __init__(self, context):
+        super(RaftController, self).__init__(context)
+        os.environ.update(PATRONI_RAFT_PARTNER_ADDRS="'" + self.CONTROLLER_ADDR + "'",
+                          PATRONI_RAFT_PASSWORD=self.PASSWORD, RAFT_PORT='1234')
+        self._raft = None
+
+    def _start(self):
+        env = os.environ.copy()
+        del env['PATRONI_RAFT_PARTNER_ADDRS']
+        env['PATRONI_RAFT_SELF_ADDR'] = self.CONTROLLER_ADDR
+        env['PATRONI_RAFT_DATA_DIR'] = self._work_directory
+        return psutil.Popen([sys.executable, '-m', 'coverage', 'run',
+                             '--source=patroni', '-p', 'patroni_raft_controller.py'],
+                            stdout=self._log, stderr=subprocess.STDOUT, env=env)
+
+    def query(self, key, scope='batman', group=None):
+        ret = self._raft.get(self.path(key, scope, group))
+        return ret and ret['value']
+
+    def set(self, key, value):
+        self._raft.set(self.path(key), value)
+
+    def cleanup_service_tree(self):
+        from patroni.dcs.raft import KVStoreTTL
+
+        if self._raft:
+            self._raft.destroy()
+            self.stop()
+            os.makedirs(self._work_directory)
+            self.start()
+
+        ready_event = threading.Event()
+        self._raft = KVStoreTTL(ready_event.set, None, None,
+                                partner_addrs=[self.CONTROLLER_ADDR], password=self.PASSWORD)
+        self._raft.startAutoTick()
+        ready_event.wait()
 
 
 class PatroniPoolController(object):
 
-    BACKUP_SCRIPT = 'features/backup_create.sh'
+    PYTHON = sys.executable.replace('\\', '/')
+    BACKUP_SCRIPT = [PYTHON, 'features/backup_create.py']
+    BACKUP_RESTORE_SCRIPT = ' '.join((PYTHON, os.path.abspath('features/backup_restore.py'))).replace('\\', '/')
+    ARCHIVE_RESTORE_SCRIPT = ' '.join((PYTHON, os.path.abspath('features/archive-restore.py')))
 
     def __init__(self, context):
         self._context = context
@@ -570,8 +792,17 @@ class PatroniPoolController(object):
         self._patroni_path = None
         self._processes = {}
         self.create_and_set_output_directory('')
+        self._check_postgres_ssl()
         self.known_dcs = {subclass.name(): subclass for subclass in AbstractDcsController.get_subclasses()}
 
+    def _check_postgres_ssl(self):
+        try:
+            subprocess.check_output(['postgres', '-D', os.devnull, '-c', 'ssl=on'], stderr=subprocess.STDOUT)
+            raise Exception  # this one should never happen because the previous line will always raise and exception
+        except Exception as e:
+            self._context.postgres_supports_ssl = isinstance(e, subprocess.CalledProcessError)\
+                    and 'SSL is not supported by this build' not in e.output.decode()
+
     @property
     def patroni_path(self):
         if self._patroni_path is None:
@@ -587,16 +818,15 @@ class PatroniPoolController(object):
     def output_dir(self):
         return self._output_dir
 
-    def start(self, name, max_wait_limit=20, custom_config=None):
+    def start(self, name, max_wait_limit=40, custom_config=None):
         if name not in self._processes:
             self._processes[name] = PatroniController(self._context, name, self.patroni_path,
                                                       self._output_dir, custom_config)
         self._processes[name].start(max_wait_limit)
 
     def __getattr__(self, func):
-        if func not in ['stop', 'query', 'write_label', 'read_label', 'check_role_has_changed_to', 'add_tag_to_config',
+        if func not in ['stop', 'query', 'write_label', 'read_label', 'check_role_has_changed_to',
-                        'get_watchdog', 'database_is_running', 'checkpoint_hang', 'patroni_hang',
+                        'add_tag_to_config', 'get_watchdog', 'patroni_hang', 'backup']:
-                        'terminate_backends', 'backup']:
             raise AttributeError("PatroniPoolController instance has no attribute '{0}'".format(func))
 
         def wrapper(name, *args, **kwargs):
@@ -610,7 +840,7 @@ class PatroniPoolController(object):
         self._processes.clear()
 
     def create_and_set_output_directory(self, feature_name):
-        feature_dir = os.path.join(self.patroni_path, 'features/output', feature_name.replace(' ', '_'))
+        feature_dir = os.path.join(self.patroni_path, 'features', 'output', feature_name.replace(' ', '_'))
         if os.path.exists(feature_dir):
             shutil.rmtree(feature_dir)
         os.makedirs(feature_dir)
@@ -623,7 +853,8 @@ class PatroniPoolController(object):
             'bootstrap': {
                 'method': 'pg_basebackup',
                 'pg_basebackup': {
-                    'command': self.BACKUP_SCRIPT + ' --walmethod=stream --dbname=' + f.backup_source
+                    'command': " ".join(self.BACKUP_SCRIPT +
+                                        ['--walmethod=stream', '--dbname="{0}"'.format(f.backup_source)])
                 },
                 'dcs': {
                     'postgresql': {
@@ -636,8 +867,9 @@ class PatroniPoolController(object):
             'postgresql': {
                 'parameters': {
                     'archive_mode': 'on',
-                    'archive_command': 'mkdir -p {0} && test ! -f {0}/%f && cp %p {0}/%f'.format(
+                    'archive_command': (self.ARCHIVE_RESTORE_SCRIPT + ' --mode archive ' +
-                            os.path.join(self.patroni_path, 'data/wal_archive'))
+                                        '--dirname {} --filename %f --pathname %p').format(
+                                        os.path.join(self.patroni_path, 'data', 'wal_archive_clone').replace('\\', '/'))
                 },
                 'authentication': {
                     'superuser': {'password': 'zalando1'},
@@ -653,12 +885,14 @@ class PatroniPoolController(object):
             'bootstrap': {
                 'method': 'backup_restore',
                 'backup_restore': {
-                    'command': 'features/backup_restore.sh --sourcedir=' + os.path.join(self.patroni_path,
+                    'command': (self.BACKUP_RESTORE_SCRIPT + ' --sourcedir=' +
-                                                                                        'data/basebackup'),
+                                os.path.join(self.patroni_path, 'data', 'basebackup').replace('\\', '/')),
                     'recovery_conf': {
                         'recovery_target_action': 'promote',
                         'recovery_target_timeline': 'latest',
-                        'restore_command': 'cp {0}/data/wal_archive/%f %p'.format(self.patroni_path)
+                        'restore_command': (self.ARCHIVE_RESTORE_SCRIPT + ' --mode restore ' +
+                                            '--dirname {} --filename %f --pathname %p').format(
+                            os.path.join(self.patroni_path, 'data', 'wal_archive_clone').replace('\\', '/'))
                     }
                 }
             },
@@ -671,6 +905,25 @@ class PatroniPoolController(object):
         }
         self.start(name, custom_config=custom_config)
 
+    def bootstrap_from_backup_no_leader(self, name, cluster_name):
+        custom_config = {
+            'scope': cluster_name,
+            'postgresql': {
+                'recovery_conf': {
+                    'restore_command': (self.ARCHIVE_RESTORE_SCRIPT + ' --mode restore ' +
+                                        '--dirname {} --filename %f --pathname %p').format(
+                                        os.path.join(self.patroni_path, 'data', 'wal_archive').replace('\\', '/'))
+                },
+                'create_replica_methods': ['no_leader_bootstrap'],
+                'no_leader_bootstrap': {
+                    'command': (self.BACKUP_RESTORE_SCRIPT + ' --sourcedir=' +
+                                os.path.join(self.patroni_path, 'data', 'basebackup').replace('\\', '/')),
+                    'no_leader': '1'
+                }
+            }
+        }
+        self.start(name, custom_config=custom_config)
+
     @property
     def dcs(self):
         if self._dcs is None:
@@ -795,12 +1048,34 @@ class WatchdogMonitor(object):
         return triggered
 
 
-# actions to execute on start/stop of the tests and before running invidual features
+# actions to execute on start/stop of the tests and before running individual features
 def before_all(context):
-    os.environ.update({'PATRONI_RESTAPI_USERNAME': 'username', 'PATRONI_RESTAPI_PASSWORD': 'password'})
+    context.ci = os.name == 'nt' or\
-    context.ci = 'TRAVIS_BUILD_NUMBER' in os.environ or 'BUILD_NUMBER' in os.environ
+        any(a in os.environ for a in ('TRAVIS_BUILD_NUMBER', 'BUILD_NUMBER', 'GITHUB_ACTIONS'))
-    context.timeout_multiplier = 2 if context.ci else 1
+    context.timeout_multiplier = 5 if context.ci else 1  # MacOS sometimes is VERY slow
     context.pctl = PatroniPoolController(context)
+
+    context.keyfile = os.path.join(context.pctl.output_dir, 'patroni.key')
+    context.certfile = os.path.join(context.pctl.output_dir, 'patroni.crt')
+    try:
+        with open(os.devnull, 'w') as null:
+            ret = subprocess.call(['openssl', 'req', '-nodes', '-new', '-x509', '-subj', '/CN=batman.patroni',
+                                   '-keyout', context.keyfile, '-out', context.certfile], stdout=null, stderr=null)
+            if ret != 0:
+                raise Exception
+    except Exception:
+        context.keyfile = context.certfile = None
+
+    os.environ.update({'PATRONI_RESTAPI_USERNAME': 'username', 'PATRONI_RESTAPI_PASSWORD': 'password'})
+    ctl = {'auth': os.environ['PATRONI_RESTAPI_USERNAME'] + ':' + os.environ['PATRONI_RESTAPI_PASSWORD']}
+    if context.certfile:
+        os.environ.update({'PATRONI_RESTAPI_CAFILE': context.certfile,
+                           'PATRONI_RESTAPI_CERTFILE': context.certfile,
+                           'PATRONI_RESTAPI_KEYFILE': context.keyfile,
+                           'PATRONI_RESTAPI_VERIFY_CLIENT': 'required',
+                           'PATRONI_CTL_INSECURE': 'on'})
+        ctl.update({'cacert': context.certfile, 'certfile': context.certfile, 'keyfile': context.keyfile})
+    context.request_executor = PatroniRequest({'ctl': ctl}, True)
     context.dcs_ctl = context.pctl.known_dcs[context.pctl.dcs](context)
     context.dcs_ctl.start()
     try:
@@ -818,13 +1093,45 @@ def after_all(context):
 
 def before_feature(context, feature):
     """ create per-feature output directory to collect Patroni and PostgreSQL logs """
+    if feature.name == 'watchdog' and os.name == 'nt':
+        return feature.skip("Watchdog isn't supported on Windows")
+    elif feature.name == 'citus':
+        lib = subprocess.check_output(['pg_config', '--pkglibdir']).decode('utf-8').strip()
+        if not os.path.exists(os.path.join(lib, 'citus.so')):
+            return feature.skip("Citus extenstion isn't available")
     context.pctl.create_and_set_output_directory(feature.name)
 
 
 def after_feature(context, feature):
-    """ stop all Patronis, remove their data directory and cleanup the keys in etcd """
+    """ send SIGCONT to a dcs if neccessary,
+    stop all Patronis remove their data directory and cleanup the keys in etcd """
+    context.dcs_ctl.stop_outage()
     context.pctl.stop_all()
-    shutil.rmtree(os.path.join(context.pctl.patroni_path, 'data'))
+    data = os.path.join(context.pctl.patroni_path, 'data')
+    if os.path.exists(data):
+        shutil.rmtree(data)
     context.dcs_ctl.cleanup_service_tree()
-    if feature.status == 'failed':
+
+    found = False
+    logs = glob.glob(context.pctl.output_dir + '/patroni_*.log')
+    for log in logs:
+        with open(log) as f:
+            for line in f:
+                if 'please report it as a BUG' in line:
+                    print(':'.join([log, line.rstrip()]))
+                    found = True
+
+    if feature.status == 'failed' or found:
         shutil.copytree(context.pctl.output_dir, context.pctl.output_dir + '_failed')
+    if found:
+        raise Exception('Unexpected errors in Patroni log files')
+
+
+def before_scenario(context, scenario):
+    if 'slot-advance' in scenario.effective_tags:
+        for p in context.pctl._processes.values():
+            if p._conn and p._conn.server_version < 110000:
+                scenario.skip('pg_replication_slot_advance() is not supported on {0}'.format(p._conn.server_version))
+                break
+    if 'dcs-failsafe' in scenario.effective_tags and not context.dcs_ctl._handle:
+        scenario.skip('it is not possible to control state of {0} from tests'.format(context.dcs_ctl.name()))

features/ignored_slots.feature

@@ -0,0 +1,61 @@
+Feature: ignored slots
+  Scenario: check ignored slots aren't removed on failover/switchover
+    Given I start postgres1
+    Then postgres1 is a leader after 10 seconds
+    And there is a non empty initialize key in DCS after 15 seconds
+    When I issue a PATCH request to http://127.0.0.1:8009/config with {"ignore_slots": [{"name": "unmanaged_slot_0", "database": "postgres", "plugin": "test_decoding", "type": "logical"}, {"name": "unmanaged_slot_1", "database": "postgres", "plugin": "test_decoding"}, {"name": "unmanaged_slot_2", "database": "postgres"}, {"name": "unmanaged_slot_3"}], "postgresql": {"parameters": {"wal_level": "logical"}}}
+    Then I receive a response code 200
+    And Response on GET http://127.0.0.1:8009/config contains ignore_slots after 10 seconds
+    # Make sure the wal_level has been changed.
+    When I shut down postgres1
+    And I start postgres1
+    Then postgres1 is a leader after 10 seconds
+    And "members/postgres1" key in DCS has role=master after 3 seconds
+    # Make sure Patroni has finished telling Postgres it should be accepting writes.
+    And postgres1 role is the primary after 20 seconds
+    # 1. Create our test logical replication slot.
+    # Test that ny subset of attributes in the ignore slots matcher is enough to match a slot
+    # by using 3 different slots.
+    When I create a logical replication slot unmanaged_slot_0 on postgres1 with the test_decoding plugin
+    And I create a logical replication slot unmanaged_slot_1 on postgres1 with the test_decoding plugin
+    And I create a logical replication slot unmanaged_slot_2 on postgres1 with the test_decoding plugin
+    And I create a logical replication slot unmanaged_slot_3 on postgres1 with the test_decoding plugin
+    And I create a logical replication slot dummy_slot on postgres1 with the test_decoding plugin
+    # It seems like it'd be obvious that these slots exist since we just created them,
+    # but Patroni can actually end up dropping them almost immediately, so it's helpful
+    # to verify they exist before we begin testing whether they persist through failover
+    # cycles.
+    Then postgres1 has a logical replication slot named unmanaged_slot_0 with the test_decoding plugin
+    And postgres1 has a logical replication slot named unmanaged_slot_1 with the test_decoding plugin
+    And postgres1 has a logical replication slot named unmanaged_slot_2 with the test_decoding plugin
+    And postgres1 has a logical replication slot named unmanaged_slot_3 with the test_decoding plugin
+
+    When I start postgres0
+    Then "members/postgres0" key in DCS has role=replica after 3 seconds
+    And postgres0 role is the secondary after 20 seconds
+    # Verify that the replica has advanced beyond the point in the WAL
+    # where we created the replication slot so that on the next failover
+    # cycle we don't accidentally rewind to before the slot creation.
+    And replication works from postgres1 to postgres0 after 20 seconds
+    When I shut down postgres1
+    Then "members/postgres0" key in DCS has role=master after 3 seconds
+
+    # 2. After a failover the server (now a replica) still has the slot.
+    When I start postgres1
+    Then postgres1 role is the secondary after 20 seconds
+    And "members/postgres1" key in DCS has role=replica after 3 seconds
+    # give Patroni time to sync replication slots
+    And I sleep for 2 seconds
+    And postgres1 has a logical replication slot named unmanaged_slot_0 with the test_decoding plugin
+    And postgres1 has a logical replication slot named unmanaged_slot_1 with the test_decoding plugin
+    And postgres1 has a logical replication slot named unmanaged_slot_2 with the test_decoding plugin
+    And postgres1 has a logical replication slot named unmanaged_slot_3 with the test_decoding plugin
+    And postgres1 does not have a logical replication slot named dummy_slot
+
+    # 3. After a failover the server (now a primary) still has the slot.
+    When I shut down postgres0
+    Then "members/postgres1" key in DCS has role=master after 3 seconds
+    And postgres1 has a logical replication slot named unmanaged_slot_0 with the test_decoding plugin
+    And postgres1 has a logical replication slot named unmanaged_slot_1 with the test_decoding plugin
+    And postgres1 has a logical replication slot named unmanaged_slot_2 with the test_decoding plugin
+    And postgres1 has a logical replication slot named unmanaged_slot_3 with the test_decoding plugin

features/patroni_api.feature

@@ -8,13 +8,15 @@ Scenario: check API requests on a stand-alone server
 	Then I receive a response code 200
 	And I receive a response state running
 	And I receive a response role master
+	When I issue a GET request to http://127.0.0.1:8008/standby_leader
+	Then I receive a response code 503
 	When I issue a GET request to http://127.0.0.1:8008/health
 	Then I receive a response code 200
 	When I issue a GET request to http://127.0.0.1:8008/replica
 	Then I receive a response code 503
-	When I run patronictl.py reinit batman postgres0 --force
+	When I issue a POST request to http://127.0.0.1:8008/reinitialize with {"force": true}
-	Then I receive a response returncode 0
+	Then I receive a response code 503
-	And I receive a response output "Failed: reinitialize for member postgres0, status code=503, (I am the leader, can not reinitialize)"
+	And I receive a response text I am the leader, can not reinitialize
 	When I run patronictl.py switchover batman --master postgres0 --force
 	Then I receive a response returncode 1
 	And I receive a response output "Error: No candidates found to switchover to"
@@ -28,30 +30,48 @@ Scenario: check API requests on a stand-alone server
 	And I receive a response text "Failover could be performed only to a specific candidate"
 
 Scenario: check local configuration reload
-	Given I issue an empty POST request to http://127.0.0.1:8008/reload
+	Given I add tag new_tag new_value to postgres0 config
-	Then I receive a response code 200
-	And I receive a response text nothing changed
-	When I add tag new_tag new_value to postgres0 config
 	And I issue an empty POST request to http://127.0.0.1:8008/reload
 	Then I receive a response code 202
 
 Scenario: check dynamic configuration change via DCS
-	Given I run patronictl.py edit-config -s 'ttl=10' -s 'loop_wait=2' -p 'max_connections=101' --force batman
+	Given I run patronictl.py edit-config -s 'ttl=10' -p 'max_connections=101' --force batman
 	Then I receive a response returncode 0
-	And I receive a response output "+loop_wait: 2"
+	And I receive a response output "+ttl: 10"
 	And Response on GET http://127.0.0.1:8008/patroni contains pending_restart after 11 seconds
 	When I issue a GET request to http://127.0.0.1:8008/config
 	Then I receive a response code 200
-	And I receive a response loop_wait 2
+	And I receive a response ttl 10
 	When I issue a GET request to http://127.0.0.1:8008/patroni
 	Then I receive a response code 200
 	And I receive a response tags {'new_tag': 'new_value'}
+	And I sleep for 4 seconds
+
+Scenario: check the scheduled restart
+	Given I issue a PATCH request to http://127.0.0.1:8008/config with {"postgresql": {"parameters": {"superuser_reserved_connections": "6"}}}
+	Then I receive a response code 200
+		And Response on GET http://127.0.0.1:8008/patroni contains pending_restart after 5 seconds
+	Given I issue a scheduled restart at http://127.0.0.1:8008 in 5 seconds with {"role": "replica"}
+	Then I receive a response code 202
+		And I sleep for 8 seconds
+		And Response on GET http://127.0.0.1:8008/patroni contains pending_restart after 10 seconds
+	Given I issue a scheduled restart at http://127.0.0.1:8008 in 5 seconds with {"restart_pending": "True"}
+	Then I receive a response code 202
+		And Response on GET http://127.0.0.1:8008/patroni does not contain pending_restart after 10 seconds
+		And postgres0 role is the primary after 10 seconds
 
 Scenario: check API requests for the primary-replica pair in the pause mode
-	Given I run patronictl.py pause batman
+	Given I start postgres1
-	Then I receive a response returncode 0
-	When I start postgres1
 	Then replication works from postgres0 to postgres1 after 20 seconds
+	When I run patronictl.py pause batman
+	Then I receive a response returncode 0
+	When I kill postmaster on postgres1
+	And I issue a GET request to http://127.0.0.1:8009/replica
+	Then I receive a response code 503
+	When I run patronictl.py restart batman postgres1 --force
+	Then I receive a response returncode 0
+	Then replication works from postgres0 to postgres1 after 20 seconds
+	And I sleep for 2 seconds
 	When I issue a GET request to http://127.0.0.1:8009/replica
 	Then I receive a response code 200
 	And I receive a response state running
@@ -73,45 +93,34 @@ Scenario: check the switchover via the API in the pause mode
 	And postgres1 role is the primary after 10 seconds
 	And postgres0 role is the secondary after 10 seconds
 	And replication works from postgres1 to postgres0 after 20 seconds
-	When I issue a GET request to http://127.0.0.1:8008/master
+	And "members/postgres0" key in DCS has state=running after 10 seconds
+	When I issue a GET request to http://127.0.0.1:8008/primary
 	Then I receive a response code 503
 	When I issue a GET request to http://127.0.0.1:8008/replica
 	Then I receive a response code 200
-	When I issue a GET request to http://127.0.0.1:8009/master
+	When I issue a GET request to http://127.0.0.1:8009/primary
 	Then I receive a response code 200
 	When I issue a GET request to http://127.0.0.1:8009/replica
 	Then I receive a response code 503
 
 Scenario: check the scheduled switchover
-	Given I issue a scheduled switchover from postgres1 to postgres0 in 3 seconds
+	Given I issue a scheduled switchover from postgres1 to postgres0 in 10 seconds
 	Then I receive a response returncode 1
 	And I receive a response output "Can't schedule switchover in the paused state"
 	When I run patronictl.py resume batman
 	Then I receive a response returncode 0
-	Given I issue a scheduled switchover from postgres1 to postgres0 in 3 seconds
+	Given I issue a scheduled switchover from postgres1 to postgres0 in 10 seconds
 	Then I receive a response returncode 0
 	And postgres0 is a leader after 20 seconds
 	And postgres0 role is the primary after 10 seconds
 	And postgres1 role is the secondary after 10 seconds
 	And replication works from postgres0 to postgres1 after 25 seconds
-	When I issue a GET request to http://127.0.0.1:8008/master
+	And "members/postgres1" key in DCS has state=running after 10 seconds
+	When I issue a GET request to http://127.0.0.1:8008/primary
 	Then I receive a response code 200
 	When I issue a GET request to http://127.0.0.1:8008/replica
 	Then I receive a response code 503
-	When I issue a GET request to http://127.0.0.1:8009/master
+	When I issue a GET request to http://127.0.0.1:8009/primary
 	Then I receive a response code 503
 	When I issue a GET request to http://127.0.0.1:8009/replica
 	Then I receive a response code 200
-
-Scenario: check the scheduled restart
-	Given I issue a PATCH request to http://127.0.0.1:8008/config with {"postgresql": {"parameters": {"superuser_reserved_connections": "6"}}}
-	Then I receive a response code 200
-		And Response on GET http://127.0.0.1:8008/patroni contains pending_restart after 5 seconds
-	Given I issue a scheduled restart at http://127.0.0.1:8008 in 3 seconds with {"role": "replica"}
-	Then I receive a response code 202
-		And I sleep for 4 seconds
-		And Response on GET http://127.0.0.1:8008/patroni contains pending_restart after 10 seconds
-	Given I issue a scheduled restart at http://127.0.0.1:8008 in 3 seconds with {"restart_pending": "True"}
-	Then I receive a response code 202
-		And Response on GET http://127.0.0.1:8008/patroni does not contain pending_restart after 10 seconds
-

features/standby_cluster.feature

@@ -1,46 +1,58 @@
 Feature: standby cluster
-  Scenario: check permanent logical slots are preserved on failover/switchover
+  Scenario: prepare the cluster with logical slots
     Given I start postgres1
     Then postgres1 is a leader after 10 seconds
-    And I sleep for 3 seconds
+    And there is a non empty initialize key in DCS after 15 seconds
-    When I issue a PATCH request to http://127.0.0.1:8009/config with {"loop_wait": 2, "slots": {"pm_1": {"type": "physical"}}, "postgresql": {"parameters": {"wal_level": "logical"}}}
+    When I issue a PATCH request to http://127.0.0.1:8009/config with {"slots": {"pm_1": {"type": "physical"}}, "postgresql": {"parameters": {"wal_level": "logical"}}}
     Then I receive a response code 200
     And Response on GET http://127.0.0.1:8009/config contains slots after 10 seconds
     And I sleep for 3 seconds
     When I issue a PATCH request to http://127.0.0.1:8009/config with {"slots": {"test_logical": {"type": "logical", "database": "postgres", "plugin": "test_decoding"}}}
     Then I receive a response code 200
     And I do a backup of postgres1
-    When I start postgres0 with callback configured
+    When I start postgres0
     Then "members/postgres0" key in DCS has state=running after 10 seconds
     And replication works from postgres1 to postgres0 after 15 seconds
+
+  @slot-advance
+  Scenario: check permanent logical slots are synced to the replica
+    Given I run patronictl.py restart batman postgres1 --force
+    Then Logical slot test_logical is in sync between postgres0 and postgres1 after 10 seconds
+    When I add the table replicate_me to postgres1
+    And I get all changes from logical slot test_logical on postgres1
+    Then Logical slot test_logical is in sync between postgres0 and postgres1 after 10 seconds
+
+  Scenario: Detach exiting node from the cluster
     When I shut down postgres1
     Then postgres0 is a leader after 10 seconds
     And "members/postgres0" key in DCS has role=master after 3 seconds
     When I issue a GET request to http://127.0.0.1:8008/
     Then I receive a response code 200
-    And there is a label with "test_logical" in postgres0 data directory
 
   Scenario: check replication of a single table in a standby cluster
     Given I start postgres1 in a standby cluster batman1 as a clone of postgres0
     Then postgres1 is a leader of batman1 after 10 seconds
     When I add the table foo to postgres0
     Then table foo is present on postgres1 after 20 seconds
-    When I issue a GET request to http://127.0.0.1:8009/master
+    And I sleep for 3 seconds
-    Then I receive a response code 200
+    When I issue a GET request to http://127.0.0.1:8009/primary
+    Then I receive a response code 503
     When I issue a GET request to http://127.0.0.1:8009/standby_leader
     Then I receive a response code 200
     And I receive a response role standby_leader
-    And there is a postgres1_cb.log with "on_start replica batman1\non_role_change standby_leader batman1" in postgres1 data directory
+    And there is a postgres1_cb.log with "on_role_change standby_leader batman1" in postgres1 data directory
     When I start postgres2 in a cluster batman1
     Then postgres2 role is the replica after 24 seconds
     And table foo is present on postgres2 after 20 seconds
+    And postgres1 does not have a logical replication slot named test_logical
 
   Scenario: check failover
     When I kill postgres1
     And I kill postmaster on postgres1
     Then postgres2 is replicating from postgres0 after 32 seconds
-    When I issue a GET request to http://127.0.0.1:8010/master
+    When I issue a GET request to http://127.0.0.1:8010/primary
-    Then I receive a response code 200
+    Then I receive a response code 503
+    And I sleep for 3 seconds
     When I issue a GET request to http://127.0.0.1:8010/standby_leader
     Then I receive a response code 200
     And I receive a response role standby_leader

features/steps/basic_replication.py

@@ -1,4 +1,4 @@
-import psycopg2 as pg
+import patroni.psycopg as pg
 
 from behave import step, then
 from time import sleep, time
@@ -28,16 +28,47 @@ def stop_postgres(context, name):
 def add_table(context, table_name, pg_name):
     # parse the configuration file and get the port
     try:
-        context.pctl.query(pg_name, "CREATE TABLE {0}()".format(table_name))
+        context.pctl.query(pg_name, "CREATE TABLE public.{0}()".format(table_name))
     except pg.Error as e:
         assert False, "Error creating table {0} on {1}: {2}".format(table_name, pg_name, e)
 
 
+@step('I {action:w} wal replay on {pg_name:w}')
+def toggle_wal_replay(context, action, pg_name):
+    # pause or resume the wal replay process
+    try:
+        version = context.pctl.query(pg_name, "SHOW server_version_num").fetchone()[0]
+        wal_name = 'xlog' if int(version)/10000 < 10 else 'wal'
+        context.pctl.query(pg_name, "SELECT pg_{0}_replay_{1}()".format(wal_name, action))
+    except pg.Error as e:
+        assert False, "Error during {0} wal recovery on {1}: {2}".format(action, pg_name, e)
+
+
+@step('I {action:w} table on {pg_name:w}')
+def crdr_mytest(context, action, pg_name):
+    try:
+        if (action == "create"):
+            context.pctl.query(pg_name, "create table if not exists public.mytest(id numeric)")
+        else:
+            context.pctl.query(pg_name, "drop table if exists public.mytest")
+    except pg.Error as e:
+        assert False, "Error {0} table mytest on {1}: {2}".format(action, pg_name, e)
+
+
+@step('I load data on {pg_name:w}')
+def initiate_load(context, pg_name):
+    # perform dummy load
+    try:
+        context.pctl.query(pg_name, "insert into public.mytest select r::numeric from generate_series(1, 350000) r")
+    except pg.Error as e:
+        assert False, "Error loading test data on {0}: {1}".format(pg_name, e)
+
+
 @then('Table {table_name:w} is present on {pg_name:w} after {max_replication_delay:d} seconds')
 def table_is_present_on(context, table_name, pg_name, max_replication_delay):
     max_replication_delay *= context.timeout_multiplier
     for _ in range(int(max_replication_delay)):
-        if context.pctl.query(pg_name, "SELECT 1 FROM {0}".format(table_name), fail_ok=True) is not None:
+        if context.pctl.query(pg_name, "SELECT 1 FROM public.{0}".format(table_name), fail_ok=True) is not None:
             break
         sleep(1)
     else:
@@ -52,10 +83,10 @@ def check_role(context, pg_name, pg_role, max_promotion_timeout):
         "{0} role didn't change to {1} after {2} seconds".format(pg_name, pg_role, max_promotion_timeout)
 
 
-@step('replication works from {master:w} to {replica:w} after {time_limit:d} seconds')
+@step('replication works from {primary:w} to {replica:w} after {time_limit:d} seconds')
-@then('replication works from {master:w} to {replica:w} after {time_limit:d} seconds')
+@then('replication works from {primary:w} to {replica:w} after {time_limit:d} seconds')
-def replication_works(context, master, replica, time_limit):
+def replication_works(context, primary, replica, time_limit):
     context.execute_steps(u"""
         When I add the table test_{0} to {1}
         Then table test_{0} is present on {2} after {3} seconds
-    """.format(int(time()), master, replica, time_limit))
+    """.format(int(time()), primary, replica, time_limit))

features/steps/cascading_replication.py

@@ -12,8 +12,10 @@ def start_patroni_with_a_name_value_tag(context, name, tag_name, tag_value):
 @then('There is a {label} with "{content}" in {name:w} data directory')
 def check_label(context, label, content, name):
     label = context.pctl.read_label(name, label)
+    if label is None:
+        label = ""
     label = label.replace('\n', '\\n')
-    assert label == content, "{0} is not equal to {1}".format(label, content)
+    assert content in label, "\"{0}\" doesn't contain {1}".format(label, content)
 
 
 @step('I create label with "{content:w}" in {name:w} data directory')
@@ -21,16 +23,33 @@ def write_label(context, content, name):
     context.pctl.write_label(name, content)
 
 
-@step('"{name}" key in DCS has {key:w}={value:w} after {time_limit:d} seconds')
+@step('"{name}" key in DCS has {key:w}={value} after {time_limit:d} seconds')
 def check_member(context, name, key, value, time_limit):
     time_limit *= context.timeout_multiplier
     max_time = time.time() + int(time_limit)
+    dcs_value = None
     while time.time() < max_time:
         try:
             response = json.loads(context.dcs_ctl.query(name))
-            if response.get(key) == value:
+            dcs_value = response.get(key)
+            if dcs_value == value:
                 return
         except Exception:
             pass
         time.sleep(1)
-    assert False, "{0} does not have {1}={2} in dcs after {3} seconds".format(name, key, value, time_limit)
+    assert False, "{0} does not have {1}={2} (found {3}) in dcs after {4} seconds".format(name, key, value,
+                                                                                          dcs_value, time_limit)
+
+
+@step('there is a non empty {key:w} key in DCS after {time_limit:d} seconds')
+def check_initialize(context, key, time_limit):
+    time_limit *= context.timeout_multiplier
+    max_time = time.time() + int(time_limit)
+    while time.time() < max_time:
+        try:
+            if context.dcs_ctl.query(key):
+                return
+        except Exception:
+            pass
+        time.sleep(1)
+    assert False, "There is no {0} in dcs after {1} seconds".format(key, time_limit)

features/steps/citus.py

@@ -0,0 +1,117 @@
+import json
+import time
+
+from behave import step, then
+from dateutil import tz
+from datetime import datetime
+from functools import partial
+from threading import Thread, Event
+
+tzutc = tz.tzutc()
+
+
+@step('{name:w} is a leader in a group {group:d} after {time_limit:d} seconds')
+@then('{name:w} is a leader in a group {group:d} after {time_limit:d} seconds')
+def is_a_group_leader(context, name, group, time_limit):
+    time_limit *= context.timeout_multiplier
+    max_time = time.time() + int(time_limit)
+    while (context.dcs_ctl.query("leader", group=group) != name):
+        time.sleep(1)
+        assert time.time() < max_time, "{0} is not a leader in dcs after {1} seconds".format(name, time_limit)
+
+
+@step('"{name}" key in a group {group:d} in DCS has {key:w}={value} after {time_limit:d} seconds')
+def check_group_member(context, name, group, key, value, time_limit):
+    time_limit *= context.timeout_multiplier
+    max_time = time.time() + int(time_limit)
+    dcs_value = None
+    response = None
+    while time.time() < max_time:
+        try:
+            response = json.loads(context.dcs_ctl.query(name, group=group))
+            dcs_value = response.get(key)
+            if dcs_value == value:
+                return
+        except Exception:
+            pass
+        time.sleep(1)
+    assert False, ("{0} in a group {1} does not have {2}={3} (found {4}) in dcs" +
+                   " after {5} seconds").format(name, group, key, value, response, time_limit)
+
+
+@step('I start {name:w} in citus group {group:d}')
+def start_citus(context, name, group):
+    return context.pctl.start(name, custom_config={"citus": {"database": "postgres", "group": int(group)}})
+
+
+@step('{name1:w} is registered in the coordinator {name2:w} as the worker in group {group:d}')
+def check_registration(context, name1, name2, group):
+    worker_port = int(context.pctl.query(name1, "SHOW port").fetchone()[0])
+    r = context.pctl.query(name2, "SELECT nodeport FROM pg_catalog.pg_dist_node WHERE groupid = {0}".format(group))
+    assert worker_port == r.fetchone()[0],\
+        "Worker {0} is not registered in pg_dist_node on the coordinator {1}".format(name1, name2)
+
+
+@step('I create a distributed table on {name:w}')
+def create_distributed_table(context, name):
+    context.pctl.query(name, 'CREATE TABLE public.d(id int not null)')
+    context.pctl.query(name, "SELECT create_distributed_table('public.d', 'id')")
+
+
+@step('I cleanup a distributed table on {name:w}')
+def cleanup_distributed_table(context, name):
+    context.pctl.query(name, 'TRUNCATE public.d')
+
+
+def insert_thread(query_func, context):
+    while True:
+        if context.thread_stop_event.is_set():
+            break
+
+        context.insert_counter += 1
+        query_func('INSERT INTO public.d VALUES({0})'.format(context.insert_counter))
+
+        context.thread_stop_event.wait(0.01)
+
+
+@step('I start a thread inserting data on {name:w}')
+def start_insert_thread(context, name):
+    context.thread_stop_event = Event()
+    context.insert_counter = 0
+    query_func = partial(context.pctl.query, name)
+    thread_func = partial(insert_thread, query_func, context)
+    context.thread = Thread(target=thread_func)
+    context.thread.daemon = True
+    context.thread.start()
+
+
+@then('a thread is still alive')
+def thread_is_alive(context):
+    assert context.thread.is_alive(), "Thread is not alive"
+
+
+@step("I stop a thread")
+def stop_insert_thread(context):
+    context.thread_stop_event.set()
+    context.thread.join(1*context.timeout_multiplier)
+    assert not context.thread.is_alive(), "Thread is still alive"
+
+
+@step("a distributed table on {name:w} has expected rows")
+def count_rows(context, name):
+    rows = context.pctl.query(name, "SELECT COUNT(*) FROM public.d").fetchone()[0]
+    assert rows == context.insert_counter, "Distributed table doesn't have expected amount of rows"
+
+
+@step("There is a transaction in progress on {name:w} changing pg_dist_node")
+def check_transaction(context, name):
+    cur = context.pctl.query(name, "SELECT xact_start FROM pg_stat_activity WHERE pid <> pg_backend_pid()"
+                                   " AND state = 'idle in transaction' AND query ~ 'citus_update_node'")
+    assert cur.rowcount == 1, "There is no idle in transaction updating pg_dist_node"
+    context.xact_start = cur.fetchone()[0]
+
+
+@step("a transaction finishes in {timeout:d} seconds")
+def check_transaction_timeout(context, timeout):
+    assert (datetime.now(tzutc) - context.xact_start).seconds > timeout,\
+            "a transaction finished earlier than in {0} seconds".format(timeout)

features/steps/dcs_failsafe_mode.py

@@ -0,0 +1,16 @@
+from behave import step
+
+
+@step('DCS is down')
+def start_dcs_outage(context):
+    context.dcs_ctl.start_outage()
+
+
+@step('DCS is up')
+def stop_dcs_outage(context):
+    context.dcs_ctl.stop_outage()
+
+
+@step('I start {name:w} in a cluster {cluster_name:w} from backup with no_leader')
+def start_cluster_from_backup_no_leader(context, name, cluster_name):
+    context.pctl.bootstrap_from_backup_no_leader(name, cluster_name)

features/steps/patroni_api.py

@@ -1,8 +1,5 @@
-import base64
 import json
-import os
 import parse
-import requests
 import shlex
 import subprocess
 import sys
@@ -45,9 +42,9 @@ def sleep_for_n_seconds(context, value):
 
 
 def _set_response(context, response):
-    context.status_code = response.status_code
+    context.status_code = response.status
-    data = response.content.decode('utf-8')
+    data = response.data.decode('utf-8')
-    ct = response.headers.get('content-type', '')
+    ct = response.getheader('content-type', '')
     if ct.startswith('application/json') or\
             ct.startswith('text/yaml') or\
             ct.startswith('text/x-yaml') or\
@@ -63,13 +60,7 @@ def _set_response(context, response):
 
 @step('I issue a GET request to {url:url}')
 def do_get(context, url):
-    try:
+    do_request(context, 'GET', url, None)
-        r = requests.get(url)
-    except requests.exceptions.RequestException:
-        context.status_code = None
-        context.response = None
-    else:
-        _set_response(context, r)
 
 
 @step('I issue an empty POST request to {url:url}')
@@ -79,17 +70,15 @@ def do_post_empty(context, url):
 
 @step('I issue a {request_method:w} request to {url:url} with {data}')
 def do_request(context, request_method, url, data):
-    data = data and json.loads(data) or {}
+    if context.certfile:
-    headers = {'Authorization': 'Basic ' + base64.b64encode('username:password'.encode('utf-8')).decode('utf-8'),
+        url = url.replace('http://', 'https://')
-               'Content-Type': 'application/json'}
+    data = data and json.loads(data)
     try:
-        if request_method == 'PATCH':
+        r = context.request_executor.request(request_method, url, data)
-            r = requests.patch(url, headers=headers, json=data)
+        if request_method == 'PATCH' and r.status == 409:
-        else:
+            r = context.request_executor.request(request_method, url, data)
-            r = requests.post(url, headers=headers, json=data)
+    except Exception:
-    except requests.exceptions.RequestException:
+        context.status_code = context.response = None
-        context.status_code = None
-        context.response = None
     else:
         _set_response(context, r)
 
@@ -98,10 +87,7 @@ def do_request(context, request_method, url, data):
 def do_run(context, cmd):
     cmd = [sys.executable, '-m', 'coverage', 'run', '--source=patroni', '-p'] + shlex.split(cmd)
     try:
-        # XXX: Dirty hack! We need to take name/passwd from the config!
+        response = subprocess.check_output(cmd, stderr=subprocess.STDOUT)
-        env = os.environ.copy()
-        env.update({'PATRONI_RESTAPI_USERNAME': 'username', 'PATRONI_RESTAPI_PASSWORD': 'password'})
-        response = subprocess.check_output(cmd, stderr=subprocess.STDOUT, env=env)
         context.status_code = 0
     except subprocess.CalledProcessError as e:
         response = e.output
@@ -123,6 +109,8 @@ def check_response(context, component, data):
         assert data.strip('"') in context.response, "response {0} does not contain {1}".format(context.response, data)
     else:
         assert component in context.response, "{0} is not part of the response".format(component)
+        if context.certfile:
+            data = data.replace('http://', 'https://')
         assert str(context.response[component]) == str(data), "{0} does not contain {1}".format(component, data)
 
 
@@ -145,12 +133,28 @@ def add_tag_to_config(context, tag, value, pg_name):
     context.pctl.add_tag_to_config(pg_name, tag, value)
 
 
-@then('Response on GET {url} contains {value} after {timeout:d} seconds')
+@then('Status code on GET {url:url} is {code:d} after {timeout:d} seconds')
-def check_http_response(context, url, value, timeout, negate=False):
+def check_http_code(context, url, code, timeout):
+    if context.certfile:
+        url = url.replace('http://', 'https://')
     timeout *= context.timeout_multiplier
     for _ in range(int(timeout)):
-        r = requests.get(url)
+        r = context.request_executor.request('GET', url)
-        if (value in r.content.decode('utf-8')) != negate:
+        if int(code) == int(r.status):
+            break
+        time.sleep(1)
+    else:
+        assert False, "HTTP Status Code is not {0} after {1} seconds".format(code, timeout)
+
+
+@then('Response on GET {url:url} contains {value} after {timeout:d} seconds')
+def check_http_response(context, url, value, timeout, negate=False):
+    if context.certfile:
+        url = url.replace('http://', 'https://')
+    timeout *= context.timeout_multiplier
+    for _ in range(int(timeout)):
+        r = context.request_executor.request('GET', url)
+        if (value in r.data.decode('utf-8')) != negate:
             break
         time.sleep(1)
     else:

features/steps/slots.py

@@ -0,0 +1,60 @@
+import time
+
+from behave import step, then
+import patroni.psycopg as pg
+
+
+@step('I create a logical replication slot {slot_name} on {pg_name:w} with the {plugin:w} plugin')
+def create_logical_replication_slot(context, slot_name, pg_name, plugin):
+    try:
+        output = context.pctl.query(pg_name, ("SELECT pg_create_logical_replication_slot('{0}', '{1}'),"
+                                              " current_database()").format(slot_name, plugin))
+        print(output.fetchone())
+    except pg.Error as e:
+        print(e)
+        assert False, "Error creating slot {0} on {1} with plugin {2}".format(slot_name, pg_name, plugin)
+
+
+@then('{pg_name:w} has a logical replication slot named {slot_name} with the {plugin:w} plugin')
+def has_logical_replication_slot(context, pg_name, slot_name, plugin):
+    try:
+        row = context.pctl.query(pg_name, ("SELECT slot_type, plugin FROM pg_replication_slots"
+                                           " WHERE slot_name = '{0}'").format(slot_name)).fetchone()
+        assert row, "Couldn't find replication slot named {0}".format(slot_name)
+        assert row[0] == "logical", "Found replication slot named {0} but wasn't a logical slot".format(slot_name)
+        assert row[1] == plugin, ("Found replication slot named {0} but was using plugin "
+                                  "{1} rather than {2}").format(slot_name, row[1], plugin)
+    except pg.Error:
+        assert False, "Error looking for slot {0} on {1} with plugin {2}".format(slot_name, pg_name, plugin)
+
+
+@then('{pg_name:w} does not have a logical replication slot named {slot_name}')
+def does_not_have_logical_replication_slot(context, pg_name, slot_name):
+    try:
+        row = context.pctl.query(pg_name, ("SELECT 1 FROM pg_replication_slots"
+                                           " WHERE slot_name = '{0}'").format(slot_name)).fetchone()
+        assert not row, "Found unexpected replication slot named {0}".format(slot_name)
+    except pg.Error:
+        assert False, "Error looking for slot {0} on {1}".format(slot_name, pg_name)
+
+
+@step('Logical slot {slot_name:w} is in sync between {pg_name1:w} and {pg_name2:w} after {time_limit:d} seconds')
+def logical_slots_in_sync(context, slot_name, pg_name1, pg_name2, time_limit):
+    time_limit *= context.timeout_multiplier
+    max_time = time.time() + int(time_limit)
+    while time.time() < max_time:
+        try:
+            query = "SELECT confirmed_flush_lsn FROM pg_replication_slots WHERE slot_name = '{0}'".format(slot_name)
+            slot1 = context.pctl.query(pg_name1, query).fetchone()
+            slot2 = context.pctl.query(pg_name2, query).fetchone()
+            if slot1[0] == slot2[0]:
+                return
+        except Exception:
+            pass
+        time.sleep(1)
+    assert False, "Logical slot {0} is not in sync between {1} and {2}".format(slot_name, pg_name1, pg_name2)
+
+
+@step('I get all changes from logical slot {slot_name:w} on {pg_name:w}')
+def logical_slot_get_changes(context, slot_name, pg_name):
+    context.pctl.query(pg_name, "SELECT * FROM pg_logical_slot_get_changes('{0}', NULL, NULL)".format(slot_name))

features/steps/standby_cluster.py

@@ -4,23 +4,9 @@ import time
 from behave import step
 
 
-select_replication_query = """
+def callbacks(context, name):
-SELECT * FROM pg_catalog.pg_stat_replication
+    return {c: '{0} features/callback2.py {1}'.format(context.pctl.PYTHON, name)
-WHERE application_name = '{0}'
+            for c in ('on_start', 'on_stop', 'on_restart', 'on_role_change')}
-"""
-
-callback = "bash -c 'echo \"${*: -3:1} ${*: -2:1} ${*: -1:1}\" >> data/$1/$1_cb.log' -- "
-
-
-@step('I start {name:w} with callback configured')
-def start_patroni_with_callbacks(context, name):
-    return context.pctl.start(name, custom_config={
-        "postgresql": {
-            "callbacks": {
-                "on_role_change": "features/callback.sh"
-            }
-        }
-    })
 
 
 @step('I start {name:w} in a cluster {cluster_name:w}')
@@ -28,10 +14,10 @@ def start_patroni(context, name, cluster_name):
     return context.pctl.start(name, custom_config={
         "scope": cluster_name,
         "postgresql": {
-            "callbacks": {c: callback + name for c in ('on_start', 'on_stop', 'on_restart', 'on_role_change')},
+            "callbacks": callbacks(context, name),
             "backup_restore": {
-                "command": "features/backup_restore.sh --sourcedir=" + os.path.join(context.pctl.patroni_path,
+                "command": (context.pctl.PYTHON + " features/backup_restore.py --sourcedir=" +
-                                                                                    "data/basebackup")}
+                            os.path.join(context.pctl.patroni_path, 'data', 'basebackup').replace('\\', '/'))}
         }
     })
 
@@ -53,11 +39,12 @@ def start_patroni_standby_cluster(context, name, cluster_name, name2):
                     "port": port,
                     "primary_slot_name": "pm_1",
                     "create_replica_methods": ["backup_restore", "basebackup"]
-                }
+                },
+                "postgresql": {"parameters": {"wal_level": "logical"}}
             }
         },
         "postgresql": {
-            "callbacks": {c: callback + name for c in ('on_start', 'on_stop', 'on_restart', 'on_role_change')}
+            "callbacks": callbacks(context, name)
         }
     })
     return context.pctl.start(name)
@@ -65,12 +52,12 @@ def start_patroni_standby_cluster(context, name, cluster_name, name2):
 
 @step('{pg_name1:w} is replicating from {pg_name2:w} after {timeout:d} seconds')
 def check_replication_status(context, pg_name1, pg_name2, timeout):
-    bound_time = time.time() + timeout
+    bound_time = time.time() + timeout * context.timeout_multiplier
 
     while time.time() < bound_time:
         cur = context.pctl.query(
             pg_name2,
-            select_replication_query.format(pg_name1),
+            "SELECT * FROM pg_catalog.pg_stat_replication WHERE application_name = '{0}'".format(pg_name1),
             fail_ok=True
         )
 

features/steps/watchdog.py

@@ -15,7 +15,7 @@ def polling_loop(timeout, interval=1):
 
 @step('I start {name:w} with watchdog')
 def start_patroni_with_watchdog(context, name):
-    return context.pctl.start(name, custom_config={'watchdog': True})
+    return context.pctl.start(name, custom_config={'watchdog': True, 'bootstrap': {'dcs': {'ttl': 20}}})
 
 
 @step('{name:w} watchdog has been pinged after {timeout:d} seconds')
@@ -31,6 +31,11 @@ def watchdog_was_closed(context, name):
     assert context.pctl.get_watchdog(name).was_closed
 
 
+@step('{name:w} watchdog has a {timeout:d} second timeout')
+def watchdog_has_timeout(context, name, timeout):
+    assert context.pctl.get_watchdog(name).timeout == timeout
+
+
 @step('I reset {name:w} watchdog state')
 def watchdog_reset_pinged(context, name):
     context.pctl.get_watchdog(name).reset()
@@ -44,31 +49,6 @@ def watchdog_was_triggered(context, name, timeout):
     assert False
 
 
-@then('{name:w} watchdog was not triggered')
-def watchdog_was_not_triggered(context, name):
-    assert not context.pctl.get_watchdog(name).was_triggered
-
-
-@step('{name:w} checkpoint takes {timeout:d} seconds')
-def checkpoint_hang(context, name, timeout):
-    assert context.pctl.checkpoint_hang(name, timeout)
-
-
 @step('{name:w} hangs for {timeout:d} seconds')
 def patroni_hang(context, name, timeout):
     return context.pctl.patroni_hang(name, timeout)
-
-
-@step('I terminate {name:w} user processes')
-def terminate_backends(context, name):
-    return context.pctl.terminate_backends(name)
-
-
-@step('Sleep for {timeout:d} seconds')
-def dcs_connection_lost(context, timeout):
-    time.sleep(timeout)
-
-
-@then('{name:w} database is running')
-def database_is_running(context, name):
-    assert context.pctl.database_is_running(name)

features/watchdog.feature

@@ -6,6 +6,14 @@ Feature: watchdog
     Then postgres0 is a leader after 10 seconds
     And postgres0 role is the primary after 10 seconds
     And postgres0 watchdog has been pinged after 10 seconds
+    And postgres0 watchdog has a 15 second timeout
+
+  Scenario: watchdog is reconfigured after global ttl changed
+    Given I run patronictl.py edit-config batman -s ttl=30 --force
+    Then I receive a response returncode 0
+    And I receive a response output "+ttl: 30"
+    When I sleep for 4 seconds
+    Then postgres0 watchdog has a 25 second timeout
 
   Scenario: watchdog is disabled during pause
     Given I run patronictl.py pause batman

kubernetes/Dockerfile

@@ -1,10 +1,9 @@
-FROM postgres:11
+FROM postgres:15
-MAINTAINER Alexander Kukushkin <alexander.kukushkin@zalando.de>
+LABEL maintainer="Alexander Kukushkin <akukushkin@microsoft.com>"
 
 RUN export DEBIAN_FRONTEND=noninteractive \
     && echo 'APT::Install-Recommends "0";\nAPT::Install-Suggests "0";' > /etc/apt/apt.conf.d/01norecommend \
     && apt-get update -y \
-    && apt-get upgrade -y \
     && apt-cache depends patroni | sed -n -e 's/.* Depends: \(python3-.\+\)$/\1/p' \
             | grep -Ev '^python3-(sphinx|etcd|consul|kazoo|kubernetes)' \
             | xargs apt-get install -y vim-tiny curl jq locales git python3-pip python3-wheel \
@@ -25,7 +24,7 @@ RUN export DEBIAN_FRONTEND=noninteractive \
     && apt-get clean -y \
     && rm -rf /var/lib/apt/lists/* /root/.cache
 
-ADD entrypoint.sh /
+COPY entrypoint.sh /
 
 EXPOSE 5432 8008
 ENV LC_ALL=en_US.UTF-8 LANG=en_US.UTF-8 EDITOR=/usr/bin/editor

kubernetes/Dockerfile.citus

@@ -0,0 +1,42 @@
+FROM postgres:15
+LABEL maintainer="Alexander Kukushkin <[email protected]>"
+
+RUN export DEBIAN_FRONTEND=noninteractive \
+    && echo 'APT::Install-Recommends "0";\nAPT::Install-Suggests "0";' > /etc/apt/apt.conf.d/01norecommend \
+    && apt-get update -y \
+    && apt-get upgrade -y \
+    && apt-cache depends patroni | sed -n -e 's/.* Depends: \(python3-.\+\)$/\1/p' \
+            | grep -Ev '^python3-(sphinx|etcd|consul|kazoo|kubernetes)' \
+            | xargs apt-get install -y busybox vim-tiny curl jq less locales git python3-pip python3-wheel \
+    ## Make sure we have a en_US.UTF-8 locale available
+    && localedef -i en_US -c -f UTF-8 -A /usr/share/locale/locale.alias en_US.UTF-8 \
+    && curl https://install.citusdata.com/community/deb.sh | bash \
+    && apt-get -y install postgresql-15-citus-11.1 \
+    && pip3 install setuptools \
+    && pip3 install 'git+https://github.com/zalando/patroni.git#egg=patroni[kubernetes]' \
+    && PGHOME=/home/postgres \
+    && mkdir -p $PGHOME \
+    && chown postgres $PGHOME \
+    && sed -i "s|/var/lib/postgresql.*|$PGHOME:/bin/bash|" /etc/passwd \
+    && /bin/busybox --install -s \
+    # Set permissions for OpenShift
+    && chmod 775 $PGHOME \
+    && chmod 664 /etc/passwd \
+    # Clean up
+    && apt-get remove -y git python3-pip python3-wheel \
+    && apt-get autoremove -y \
+    && apt-get clean -y \
+    && rm -rf /var/lib/apt/lists/* /root/.cache
+
+ADD entrypoint.sh /
+ENV PGSSLMODE=verify-ca PGSSLKEY=/etc/ssl/private/ssl-cert-snakeoil.key PGSSLCERT=/etc/ssl/certs/ssl-cert-snakeoil.pem PGSSLROOTCERT=/etc/ssl/certs/ssl-cert-snakeoil.pem
+
+RUN sed -i 's/^postgresql:/&\n  basebackup:\n    checkpoint: fast/' /entrypoint.sh \
+    && sed -i "s|^    postgresql:|&\n      pg_hba:\n      - local all all trust\n      - hostssl replication all all md5 clientcert=$PGSSLMODE\n      - hostssl all all all md5 clientcert=$PGSSLMODE\n      parameters:\n        max_connections: 100\n        shared_buffers: 16MB\n        ssl: 'on'\n        ssl_ca_file: $PGSSLROOTCERT\n        ssl_cert_file: $PGSSLCERT\n        ssl_key_file: $PGSSLKEY\n        citus.node_conninfo: 'sslrootcert=$PGSSLROOTCERT sslkey=$PGSSLKEY sslcert=$PGSSLCERT sslmode=$PGSSLMODE'|"  /entrypoint.sh \
+    && sed -i "s#^    \(superuser\|replication\):#&\n      sslmode: $PGSSLMODE\n      sslkey: $PGSSLKEY\n      sslcert: $PGSSLCERT\n      sslrootcert: $PGSSLROOTCERT#" /entrypoint.sh
+
+EXPOSE 5432 8008
+ENV LC_ALL=en_US.UTF-8 LANG=en_US.UTF-8 EDITOR=/usr/bin/editor
+USER postgres
+WORKDIR /home/postgres
+CMD ["/bin/bash", "/entrypoint.sh"]

kubernetes/README.md

@@ -0,0 +1,154 @@
+# Kubernetes deployment examples
+Below you will find examples of Patroni deployments using [kind](https://kind.sigs.k8s.io/).
+
+# Patroni on K8s
+The Patroni cluster deployment with a StatefulSet consisting of three Pods.
+
+Example session:
+
+    $ kind create cluster
+    Creating cluster "kind" ...
+     ✓ Ensuring node image (kindest/node:v1.25.3) 🖼
+     ✓ Preparing nodes 📦
+     ✓ Writing configuration 📜
+     ✓ Starting control-plane 🕹️
+     ✓ Installing CNI 🔌
+     ✓ Installing StorageClass 💾
+    Set kubectl context to "kind-kind"
+    You can now use your cluster with:
+
+    kubectl cluster-info --context kind-kind
+
+    Thanks for using kind! 😊
+
+    $ docker build -t patroni .
+    Sending build context to Docker daemon  138.8kB
+    Step 1/9 : FROM postgres:15
+    ...
+    Successfully built e9bfe69c5d2b
+    Successfully tagged patroni:latest
+
+    $ kind load docker-image patroni
+    Image: "" with ID "sha256:e9bfe69c5d2b319dec0cf564fb895484537664775e18f37f9b707914cc5537e6" not yet present on node "kind-control-plane", loading...
+
+    $ kubectl apply -f patroni_k8s.yaml
+    service/patronidemo-config created
+    statefulset.apps/patronidemo created
+    endpoints/patronidemo created
+    service/patronidemo created
+    service/patronidemo-repl created
+    secret/patronidemo created
+    serviceaccount/patronidemo created
+    role.rbac.authorization.k8s.io/patronidemo created
+    rolebinding.rbac.authorization.k8s.io/patronidemo created
+    clusterrole.rbac.authorization.k8s.io/patroni-k8s-ep-access created
+    clusterrolebinding.rbac.authorization.k8s.io/patroni-k8s-ep-access created
+
+    $ kubectl get pods -L role
+    NAME            READY   STATUS    RESTARTS   AGE   ROLE
+    patronidemo-0   1/1     Running   0          34s   master
+    patronidemo-1   1/1     Running   0          30s   replica
+    patronidemo-2   1/1     Running   0          26s   replica
+
+    $ kubectl exec -ti patronidemo-0 -- bash
+    postgres@patronidemo-0:~$ patronictl list
+    + Cluster: patronidemo (7186662553319358497) ----+----+-----------+
+    | Member        | Host       | Role    | State   | TL | Lag in MB |
+    +---------------+------------+---------+---------+----+-----------+
+    | patronidemo-0 | 10.244.0.5 | Leader  | running |  1 |           |
+    | patronidemo-1 | 10.244.0.6 | Replica | running |  1 |         0 |
+    | patronidemo-2 | 10.244.0.7 | Replica | running |  1 |         0 |
+    +---------------+------------+---------+---------+----+-----------+
+
+# Citus on K8s
+The Citus cluster with the StatefulSets, one coordinator with three Pods and two workers with two pods each.
+
+Example session:
+
+    $ kind create cluster
+    Creating cluster "kind" ...
+     ✓ Ensuring node image (kindest/node:v1.25.3) 🖼
+     ✓ Preparing nodes 📦
+     ✓ Writing configuration 📜
+     ✓ Starting control-plane 🕹️
+     ✓ Installing CNI 🔌
+     ✓ Installing StorageClass 💾
+    Set kubectl context to "kind-kind"
+    You can now use your cluster with:
+
+    kubectl cluster-info --context kind-kind
+
+    Thanks for using kind! 😊
+
+    demo@localhost:~/git/patroni/kubernetes$ docker build -f Dockerfile.citus -t patroni-citus-k8s .
+    Sending build context to Docker daemon  138.8kB
+    Step 1/11 : FROM postgres:15
+    ...
+    Successfully built 8cd73e325028
+    Successfully tagged patroni-citus-k8s:latest
+
+    $ kind load docker-image patroni-citus-k8s
+    Image: "" with ID "sha256:8cd73e325028d7147672494965e53453f5540400928caac0305015eb2c7027c7" not yet present on node "kind-control-plane", loading...
+
+    $ kubectl apply -f citus_k8s.yaml
+    service/citusdemo-0-config created
+    service/citusdemo-1-config created
+    service/citusdemo-2-config created
+    statefulset.apps/citusdemo-0 created
+    statefulset.apps/citusdemo-1 created
+    statefulset.apps/citusdemo-2 created
+    endpoints/citusdemo-0 created
+    service/citusdemo-0 created
+    endpoints/citusdemo-1 created
+    service/citusdemo-1 created
+    endpoints/citusdemo-2 created
+    service/citusdemo-2 created
+    service/citusdemo-workers created
+    secret/citusdemo created
+    serviceaccount/citusdemo created
+    role.rbac.authorization.k8s.io/citusdemo created
+    rolebinding.rbac.authorization.k8s.io/citusdemo created
+    clusterrole.rbac.authorization.k8s.io/patroni-k8s-ep-access created
+    clusterrolebinding.rbac.authorization.k8s.io/patroni-k8s-ep-access created
+
+    $ kubectl get sts
+    NAME          READY   AGE
+    citusdemo-0   1/3     6s  # coodinator (group=0)
+    citusdemo-1   1/2     6s  # worker (group=1)
+    citusdemo-2   1/2     6s  # worker (group=2)
+
+    $ kubectl get pods -l cluster-name=citusdemo -L role
+    NAME            READY   STATUS    RESTARTS   AGE    ROLE
+    citusdemo-0-0   1/1     Running   0          105s   master
+    citusdemo-0-1   1/1     Running   0          101s   replica
+    citusdemo-0-2   1/1     Running   0          96s    replica
+    citusdemo-1-0   1/1     Running   0          105s   master
+    citusdemo-1-1   1/1     Running   0          101s   replica
+    citusdemo-2-0   1/1     Running   0          105s   master
+    citusdemo-2-1   1/1     Running   0          101s   replica
+
+    $ kubectl exec -ti citusdemo-0-0 -- bash
+    postgres@citusdemo-0-0:~$ patronictl list
+    + Citus cluster: citusdemo -----------+--------------+---------+----+-----------+
+    | Group | Member        | Host        | Role         | State   | TL | Lag in MB |
+    +-------+---------------+-------------+--------------+---------+----+-----------+
+    |     0 | citusdemo-0-0 | 10.244.0.10 | Leader       | running |  1 |           |
+    |     0 | citusdemo-0-1 | 10.244.0.12 | Replica      | running |  1 |         0 |
+    |     0 | citusdemo-0-2 | 10.244.0.14 | Sync Standby | running |  1 |         0 |
+    |     1 | citusdemo-1-0 | 10.244.0.8  | Leader       | running |  1 |           |
+    |     1 | citusdemo-1-1 | 10.244.0.11 | Sync Standby | running |  1 |         0 |
+    |     2 | citusdemo-2-0 | 10.244.0.9  | Leader       | running |  1 |           |
+    |     2 | citusdemo-2-1 | 10.244.0.13 | Sync Standby | running |  1 |         0 |
+    +-------+---------------+-------------+--------------+---------+----+-----------+
+
+    postgres@citusdemo-0-0:~$ psql citus
+    psql (15.1 (Debian 15.1-1.pgdg110+1))
+    Type "help" for help.
+
+    citus=# table pg_dist_node;
+     nodeid | groupid |  nodename   | nodeport | noderack | hasmetadata | isactive | noderole | nodecluster | metadatasynced | shouldhaveshards
+    --------+---------+-------------+----------+----------+-------------+----------+----------+-------------+----------------+------------------
+          1 |       0 | 10.244.0.10 |     5432 | default  | t           | t        | primary  | default     | t              | f
+          2 |       1 | 10.244.0.8  |     5432 | default  | t           | t        | primary  | default     | t              | t
+          3 |       2 | 10.244.0.9  |     5432 | default  | t           | t        | primary  | default     | t              | t
+    (3 rows)

kubernetes/citus_k8s.yaml

@@ -0,0 +1,590 @@
+# headless services to avoid deletion of citusdemo-*-config endpoints
+apiVersion: v1
+kind: Service
+metadata:
+  name: citusdemo-0-config
+  labels:
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '0'
+spec:
+  clusterIP: None
+
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: citusdemo-1-config
+  labels:
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '1'
+spec:
+  clusterIP: None
+
+---
+
+apiVersion: v1
+kind: Service
+metadata:
+  name: citusdemo-2-config
+  labels:
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '2'
+spec:
+  clusterIP: None
+
+---
+apiVersion: apps/v1
+kind: StatefulSet
+metadata:
+  name: &cluster_name citusdemo-0
+  labels: &labels
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '0'
+    citus-type: coordinator
+spec:
+  replicas: 3
+  serviceName: *cluster_name
+  selector:
+    matchLabels:
+      <<: *labels
+  template:
+    metadata:
+      labels:
+        <<: *labels
+    spec:
+      serviceAccountName: citusdemo
+      containers:
+      - name: *cluster_name
+        image: patroni-citus-k8s  # docker build -f Dockerfile.citus -t patroni-citus-k8s .
+        imagePullPolicy: IfNotPresent
+        readinessProbe:
+          httpGet:
+            scheme: HTTP
+            path: /readiness
+            port: 8008
+          initialDelaySeconds: 3
+          periodSeconds: 10
+          timeoutSeconds: 5
+          successThreshold: 1
+          failureThreshold: 3
+        ports:
+        - containerPort: 8008
+          protocol: TCP
+        - containerPort: 5432
+          protocol: TCP
+        volumeMounts:
+        - mountPath: /home/postgres/pgdata
+          name: pgdata
+        env:
+        - name: PATRONI_KUBERNETES_POD_IP
+          valueFrom:
+            fieldRef:
+              fieldPath: status.podIP
+        - name: PATRONI_KUBERNETES_NAMESPACE
+          valueFrom:
+            fieldRef:
+              fieldPath: metadata.namespace
+        - name: PATRONI_KUBERNETES_BYPASS_API_SERVICE
+          value: 'true'
+        - name: PATRONI_KUBERNETES_USE_ENDPOINTS
+          value: 'true'
+        - name: PATRONI_KUBERNETES_LABELS
+          value: '{application: patroni, cluster-name: citusdemo}'
+        - name: PATRONI_CITUS_DATABASE
+          value: citus
+        - name: PATRONI_CITUS_GROUP
+          value: '0'
+        - name: PATRONI_SUPERUSER_USERNAME
+          value: postgres
+        - name: PATRONI_SUPERUSER_PASSWORD
+          valueFrom:
+            secretKeyRef:
+              name: citusdemo
+              key: superuser-password
+        - name: PATRONI_REPLICATION_USERNAME
+          value: standby
+        - name: PATRONI_REPLICATION_PASSWORD
+          valueFrom:
+            secretKeyRef:
+              name: citusdemo
+              key: replication-password
+        - name: PATRONI_SCOPE
+          value: citusdemo
+        - name: PATRONI_NAME
+          valueFrom:
+            fieldRef:
+              fieldPath: metadata.name
+        - name: PATRONI_POSTGRESQL_DATA_DIR
+          value: /home/postgres/pgdata/pgroot/data
+        - name: PATRONI_POSTGRESQL_PGPASS
+          value: /tmp/pgpass
+        - name: PATRONI_POSTGRESQL_LISTEN
+          value: '0.0.0.0:5432'
+        - name: PATRONI_RESTAPI_LISTEN
+          value: '0.0.0.0:8008'
+      terminationGracePeriodSeconds: 0
+      volumes:
+      - name: pgdata
+        emptyDir: {}
+#  volumeClaimTemplates:
+#  - metadata:
+#      labels:
+#        application: spilo
+#        spilo-cluster: *cluster_name
+#      annotations:
+#        volume.alpha.kubernetes.io/storage-class: anything
+#      name: pgdata
+#    spec:
+#      accessModes:
+#      - ReadWriteOnce
+#      resources:
+#        requests:
+#          storage: 5Gi
+
+---
+apiVersion: apps/v1
+kind: StatefulSet
+metadata:
+  name: &cluster_name citusdemo-1
+  labels: &labels
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '1'
+    citus-type: worker
+spec:
+  replicas: 2
+  serviceName: *cluster_name
+  selector:
+    matchLabels:
+      <<: *labels
+  template:
+    metadata:
+      labels:
+        <<: *labels
+    spec:
+      serviceAccountName: citusdemo
+      containers:
+      - name: *cluster_name
+        image: patroni-citus-k8s  # docker build -f Dockerfile.citus -t patroni-citus-k8s .
+        imagePullPolicy: IfNotPresent
+        readinessProbe:
+          httpGet:
+            scheme: HTTP
+            path: /readiness
+            port: 8008
+          initialDelaySeconds: 3
+          periodSeconds: 10
+          timeoutSeconds: 5
+          successThreshold: 1
+          failureThreshold: 3
+        ports:
+        - containerPort: 8008
+          protocol: TCP
+        - containerPort: 5432
+          protocol: TCP
+        volumeMounts:
+        - mountPath: /home/postgres/pgdata
+          name: pgdata
+        env:
+        - name: PATRONI_KUBERNETES_POD_IP
+          valueFrom:
+            fieldRef:
+              fieldPath: status.podIP
+        - name: PATRONI_KUBERNETES_NAMESPACE
+          valueFrom:
+            fieldRef:
+              fieldPath: metadata.namespace
+        - name: PATRONI_KUBERNETES_BYPASS_API_SERVICE
+          value: 'true'
+        - name: PATRONI_KUBERNETES_USE_ENDPOINTS
+          value: 'true'
+        - name: PATRONI_KUBERNETES_LABELS
+          value: '{application: patroni, cluster-name: citusdemo}'
+        - name: PATRONI_CITUS_DATABASE
+          value: citus
+        - name: PATRONI_CITUS_GROUP
+          value: '1'
+        - name: PATRONI_SUPERUSER_USERNAME
+          value: postgres
+        - name: PATRONI_SUPERUSER_PASSWORD
+          valueFrom:
+            secretKeyRef:
+              name: citusdemo
+              key: superuser-password
+        - name: PATRONI_REPLICATION_USERNAME
+          value: standby
+        - name: PATRONI_REPLICATION_PASSWORD
+          valueFrom:
+            secretKeyRef:
+              name: citusdemo
+              key: replication-password
+        - name: PATRONI_SCOPE
+          value: citusdemo
+        - name: PATRONI_NAME
+          valueFrom:
+            fieldRef:
+              fieldPath: metadata.name
+        - name: PATRONI_POSTGRESQL_DATA_DIR
+          value: /home/postgres/pgdata/pgroot/data
+        - name: PATRONI_POSTGRESQL_PGPASS
+          value: /tmp/pgpass
+        - name: PATRONI_POSTGRESQL_LISTEN
+          value: '0.0.0.0:5432'
+        - name: PATRONI_RESTAPI_LISTEN
+          value: '0.0.0.0:8008'
+      terminationGracePeriodSeconds: 0
+      volumes:
+      - name: pgdata
+        emptyDir: {}
+#  volumeClaimTemplates:
+#  - metadata:
+#      labels:
+#        application: spilo
+#        spilo-cluster: *cluster_name
+#      annotations:
+#        volume.alpha.kubernetes.io/storage-class: anything
+#      name: pgdata
+#    spec:
+#      accessModes:
+#      - ReadWriteOnce
+#      resources:
+#        requests:
+#          storage: 5Gi
+
+---
+apiVersion: apps/v1
+kind: StatefulSet
+metadata:
+  name: &cluster_name citusdemo-2
+  labels: &labels
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '2'
+    citus-type: worker
+spec:
+  replicas: 2
+  serviceName: *cluster_name
+  selector:
+    matchLabels:
+      <<: *labels
+  template:
+    metadata:
+      labels:
+        <<: *labels
+    spec:
+      serviceAccountName: citusdemo
+      containers:
+      - name: *cluster_name
+        image: patroni-citus-k8s  # docker build -f Dockerfile.citus -t patroni-citus-k8s .
+        imagePullPolicy: IfNotPresent
+        readinessProbe:
+          httpGet:
+            scheme: HTTP
+            path: /readiness
+            port: 8008
+          initialDelaySeconds: 3
+          periodSeconds: 10
+          timeoutSeconds: 5
+          successThreshold: 1
+          failureThreshold: 3
+        ports:
+        - containerPort: 8008
+          protocol: TCP
+        - containerPort: 5432
+          protocol: TCP
+        volumeMounts:
+        - mountPath: /home/postgres/pgdata
+          name: pgdata
+        env:
+        - name: PATRONI_KUBERNETES_POD_IP
+          valueFrom:
+            fieldRef:
+              fieldPath: status.podIP
+        - name: PATRONI_KUBERNETES_NAMESPACE
+          valueFrom:
+            fieldRef:
+              fieldPath: metadata.namespace
+        - name: PATRONI_KUBERNETES_BYPASS_API_SERVICE
+          value: 'true'
+        - name: PATRONI_KUBERNETES_USE_ENDPOINTS
+          value: 'true'
+        - name: PATRONI_KUBERNETES_LABELS
+          value: '{application: patroni, cluster-name: citusdemo}'
+        - name: PATRONI_CITUS_DATABASE
+          value: citus
+        - name: PATRONI_CITUS_GROUP
+          value: '2'
+        - name: PATRONI_SUPERUSER_USERNAME
+          value: postgres
+        - name: PATRONI_SUPERUSER_PASSWORD
+          valueFrom:
+            secretKeyRef:
+              name: citusdemo
+              key: superuser-password
+        - name: PATRONI_REPLICATION_USERNAME
+          value: standby
+        - name: PATRONI_REPLICATION_PASSWORD
+          valueFrom:
+            secretKeyRef:
+              name: citusdemo
+              key: replication-password
+        - name: PATRONI_SCOPE
+          value: citusdemo
+        - name: PATRONI_NAME
+          valueFrom:
+            fieldRef:
+              fieldPath: metadata.name
+        - name: PATRONI_POSTGRESQL_DATA_DIR
+          value: /home/postgres/pgdata/pgroot/data
+        - name: PATRONI_POSTGRESQL_PGPASS
+          value: /tmp/pgpass
+        - name: PATRONI_POSTGRESQL_LISTEN
+          value: '0.0.0.0:5432'
+        - name: PATRONI_RESTAPI_LISTEN
+          value: '0.0.0.0:8008'
+      terminationGracePeriodSeconds: 0
+      volumes:
+      - name: pgdata
+        emptyDir: {}
+#  volumeClaimTemplates:
+#  - metadata:
+#      labels:
+#        application: spilo
+#        spilo-cluster: *cluster_name
+#      annotations:
+#        volume.alpha.kubernetes.io/storage-class: anything
+#      name: pgdata
+#    spec:
+#      accessModes:
+#      - ReadWriteOnce
+#      resources:
+#        requests:
+#          storage: 5Gi
+          
+---
+apiVersion: v1
+kind: Endpoints
+metadata:
+  name: citusdemo-0
+  labels:
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '0'
+    citus-type: coordinator
+subsets: []
+
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: citusdemo-0
+  labels:
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '0'
+    citus-type: coordinator
+spec:
+  type: ClusterIP
+  ports:
+  - port: 5432
+    targetPort: 5432
+
+---
+apiVersion: v1
+kind: Endpoints
+metadata:
+  name: citusdemo-1
+  labels:
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '1'
+    citus-type: worker
+subsets: []
+
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: citusdemo-1
+  labels:
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '1'
+    citus-type: worker
+spec:
+  type: ClusterIP
+  ports:
+  - port: 5432
+    targetPort: 5432
+
+---
+apiVersion: v1
+kind: Endpoints
+metadata:
+  name: citusdemo-2
+  labels:
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '2'
+    citus-type: worker
+subsets: []
+
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: citusdemo-2
+  labels:
+    application: patroni
+    cluster-name: citusdemo
+    citus-group: '2'
+    citus-type: worker
+spec:
+  type: ClusterIP
+  ports:
+  - port: 5432
+    targetPort: 5432
+
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: citusdemo-workers
+  labels: &labels
+    application: patroni
+    cluster-name: citusdemo
+    citus-type: worker
+    role: master
+spec:
+  type: ClusterIP
+  selector:
+    <<: *labels
+  ports:
+  - port: 5432
+    targetPort: 5432
+
+---
+apiVersion: v1
+kind: Secret
+metadata:
+  name: &cluster_name citusdemo
+  labels:
+    application: patroni
+    cluster-name: *cluster_name
+type: Opaque
+data:
+  superuser-password: emFsYW5kbw==
+  replication-password: cmVwLXBhc3M=
+
+---
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: citusdemo
+
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: Role
+metadata:
+  name: citusdemo
+rules:
+- apiGroups:
+  - ""
+  resources:
+  - configmaps
+  verbs:
+  - create
+  - get
+  - list
+  - patch
+  - update
+  - watch
+  # delete and deletecollection are required only for 'patronictl remove'
+  - delete
+  - deletecollection
+- apiGroups:
+  - ""
+  resources:
+  - endpoints
+  verbs:
+  - get
+  - patch
+  - update
+  # the following three privileges are necessary only when using endpoints
+  - create
+  - list
+  - watch
+  # delete and deletecollection are required only for for 'patronictl remove'
+  - delete
+  - deletecollection
+- apiGroups:
+  - ""
+  resources:
+  - pods
+  verbs:
+  - get
+  - list
+  - patch
+  - update
+  - watch
+# The following privilege is only necessary for creation of headless service
+# for citusdemo-config endpoint, in order to prevent cleaning it up by the
+# k8s master. You can avoid giving this privilege by explicitly creating the
+# service like it is done in this manifest (lines 2..10)
+- apiGroups:
+  - ""
+  resources:
+  - services
+  verbs:
+  - create
+
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: RoleBinding
+metadata:
+  name: citusdemo
+roleRef:
+  apiGroup: rbac.authorization.k8s.io
+  kind: Role
+  name: citusdemo
+subjects:
+- kind: ServiceAccount
+  name: citusdemo
+
+# Following privileges are only required if deployed not in the "default"
+# namespace and you want Patroni to bypass kubernetes service
+# (PATRONI_KUBERNETES_BYPASS_API_SERVICE=true)
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRole
+metadata:
+  name: patroni-k8s-ep-access
+rules:
+- apiGroups:
+  - ""
+  resources:
+  - endpoints
+  resourceNames:
+  - kubernetes
+  verbs:
+  - get
+
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRoleBinding
+metadata:
+  name: patroni-k8s-ep-access
+roleRef:
+  apiGroup: rbac.authorization.k8s.io
+  kind: ClusterRole
+  name: patroni-k8s-ep-access
+subjects:
+- kind: ServiceAccount
+  name: citusdemo
+# The namespace must be specified explicitly.
+# If deploying to the different namespace you have to change it.
+  namespace: default

kubernetes/entrypoint.sh

@@ -33,7 +33,5 @@ postgresql:
 __EOF__
 
 unset PATRONI_SUPERUSER_PASSWORD PATRONI_REPLICATION_PASSWORD
-export KUBERNETES_NAMESPACE=$PATRONI_KUBERNETES_NAMESPACE
-export POD_NAME=$PATRONI_NAME
 
-exec /usr/bin/python3 /usr/local/bin/patroni /home/postgres/patroni.yml
+exec /usr/bin/python3 /usr/local/bin/patroni /home/postgres/patroni.yml

kubernetes/openshift-example/README.md

@@ -1,5 +1,5 @@
 # Patroni OpenShift Configuration
-Patroni can be run in OpenShift. Based on the kubernetes configuration, the Dockerfile and Entrypoint has been modified to support the dynamic UID/GID configuration that is applied in OpenShift. This can be run under the standard `restricted` SCC. 
+Patroni can be run in OpenShift. Based on the kubernetes configuration, the Dockerfile and Entrypoint has been modified to support the dynamic UID/GID configuration that is applied in OpenShift. This can be run under the standard `restricted` SCC.
 
 # Examples
 
@@ -11,39 +11,39 @@ oc new-project patroni-test
 
 ## Build the image
 
-Note: Update the references when merged upstream. 
+Note: Update the references when merged upstream.
-Note: If deploying as a template for multiple users, the following commands should be performed in a shared namespace like `openshift`. 
+Note: If deploying as a template for multiple users, the following commands should be performed in a shared namespace like `openshift`.
 
 ```
 oc import-image postgres:10 --confirm -n openshift
 oc new-build https://github.com/zalando/patroni --context-dir=kubernetes -n openshift
 ```
 
-## Deploy the Image 
+## Deploy the Image
-Two configuration templates exist in [templates](templates) directory: 
+Two configuration templates exist in [templates](templates) directory:
-- Patroni Ephemeral
+-   Patroni Ephemeral
-- Patroni Persistent
+-   Patroni Persistent
 
-The only difference is whether or not the statefulset requests persistent storage. 
+The only difference is whether or not the statefulset requests persistent storage.
 
 ## Create the Template
-Install the template into the `openshift` namespace if this should be shared across projects: 
+Install the template into the `openshift` namespace if this should be shared across projects:
 
 ```
 oc create -f templates/template_patroni_ephemeral.yml -n openshift
 ```
 
-Then, from your own project: 
+Then, from your own project:
 
 ```
 oc new-app patroni-pgsql-ephemeral
 ```
 
-Once the pods are running, two configmaps should be available: 
+Once the pods are running, two configmaps should be available:
 
 ```
 $ oc get configmap
 NAME                DATA      AGE
 patroniocp-config   0         1m
 patroniocp-leader   0         1m
-```
+```

kubernetes/openshift-example/templates/template_patroni_ephemeral.yml

@@ -118,6 +118,8 @@ objects:
               fieldRef:
                 apiVersion: v1
                 fieldPath: metadata.namespace
+          - name: PATRONI_KUBERNETES_BYPASS_API_SERVICE
+            value: 'true'
           - name: PATRONI_KUBERNETES_LABELS
             value: '{application: ${APPLICATION_NAME}, cluster-name: ${PATRONI_CLUSTER_NAME}}'
           - name: PATRONI_SUPERUSER_USERNAME
@@ -152,6 +154,16 @@ objects:
           image: docker-registry.default.svc:5000/${NAMESPACE}/patroni:latest
           imagePullPolicy: IfNotPresent
           name: ${APPLICATION_NAME}
+          readinessProbe:
+            httpGet:
+              scheme: HTTP
+              path: /readiness
+              port: 8008
+            initialDelaySeconds: 3
+            periodSeconds: 10
+            timeoutSeconds: 5
+            successThreshold: 1
+            failureThreshold: 3
           ports:
           - containerPort: 8008
             protocol: TCP
@@ -240,6 +252,34 @@ objects:
   subjects:
   - kind: ServiceAccount
     name: ${SERVICE_ACCOUNT}
+# Following privileges are only required if deployed not in the "default"
+# namespace and you want Patroni to bypass kubernetes service
+# (PATRONI_KUBERNETES_BYPASS_API_SERVICE=true)
+- apiVersion: rbac.authorization.k8s.io/v1
+  kind: ClusterRole
+  metadata:
+    name: patroni-k8s-ep-access
+  rules:
+  - apiGroups:
+    - ""
+    resources:
+    - endpoints
+    resourceNames:
+    - kubernetes
+    verbs:
+    - get
+- apiVersion: rbac.authorization.k8s.io/v1
+  kind: ClusterRoleBinding
+  metadata:
+    name: ${NAMESPACE}-${SERVICE_ACCOUNT}-k8s-ep-access
+  roleRef:
+    apiGroup: rbac.authorization.k8s.io
+    kind: ClusterRole
+    name: patroni-k8s-ep-access
+  subjects:
+  - kind: ServiceAccount
+    name: ${SERVICE_ACCOUNT}
+    namespace: ${NAMESPACE}
 parameters:
 - description: The name of the application for labelling all artifacts.
   displayName: Application Name

kubernetes/openshift-example/templates/template_patroni_persistent.yaml

@@ -5,11 +5,9 @@ metadata:
   annotations:
     description: |-
       Patroni Postgresql database cluster, with persistent storage.
-
-      WARNING: Any data stored will be lost upon pod destruction. Only use this template for testing.
     iconClass: icon-postgresql
     openshift.io/display-name: Patroni Postgresql (Persistent)
-    openshift.io/long-description: This template deploys a a patroni postgresql HA cluster without persistent storage.
+    openshift.io/long-description: This template deploys a a patroni postgresql HA cluster with persistent storage.
     tags: postgresql
 objects:
 - apiVersion: v1
@@ -106,6 +104,20 @@ objects:
           application: ${APPLICATION_NAME}
           cluster-name: ${PATRONI_CLUSTER_NAME}
       spec:
+        initContainers:
+        - command:
+          - sh
+          - -c 
+          - "mkdir -p /home/postgres/pgdata/pgroot/data && chmod 0700 /home/postgres/pgdata/pgroot/data"
+          image: docker-registry.default.svc:5000/${NAMESPACE}/patroni:latest
+          imagePullPolicy: IfNotPresent
+          name: fix-perms
+          resources: {}
+          terminationMessagePath: /dev/termination-log
+          terminationMessagePolicy: File
+          volumeMounts:
+          - mountPath: /home/postgres/pgdata
+            name: ${APPLICATION_NAME}
         containers:
         - env:
           - name: PATRONI_KUBERNETES_POD_IP
@@ -118,6 +130,8 @@ objects:
               fieldRef:
                 apiVersion: v1
                 fieldPath: metadata.namespace
+          - name: PATRONI_KUBERNETES_BYPASS_API_SERVICE
+            value: 'true'
           - name: PATRONI_KUBERNETES_LABELS
             value: '{application: ${APPLICATION_NAME}, cluster-name: ${PATRONI_CLUSTER_NAME}}'
           - name: PATRONI_SUPERUSER_USERNAME
@@ -152,6 +166,16 @@ objects:
           image: docker-registry.default.svc:5000/${NAMESPACE}/patroni:latest
           imagePullPolicy: IfNotPresent
           name: ${APPLICATION_NAME}
+          readinessProbe:
+            httpGet:
+              scheme: HTTP
+              path: /readiness
+              port: 8008
+            initialDelaySeconds: 3
+            periodSeconds: 10
+            timeoutSeconds: 5
+            successThreshold: 1
+            failureThreshold: 3
           ports:
           - containerPort: 8008
             protocol: TCP
@@ -252,6 +276,34 @@ objects:
   subjects:
   - kind: ServiceAccount
     name: ${SERVICE_ACCOUNT}
+# Following privileges are only required if deployed not in the "default"
+# namespace and you want Patroni to bypass kubernetes service
+# (PATRONI_KUBERNETES_BYPASS_API_SERVICE=true)
+- apiVersion: rbac.authorization.k8s.io/v1
+  kind: ClusterRole
+  metadata:
+    name: patroni-k8s-ep-access
+  rules:
+  - apiGroups:
+    - ""
+    resources:
+    - endpoints
+    resourceNames:
+    - kubernetes
+    verbs:
+    - get
+- apiVersion: rbac.authorization.k8s.io/v1
+  kind: ClusterRoleBinding
+  metadata:
+    name: ${NAMESPACE}-${SERVICE_ACCOUNT}-k8s-ep-access
+  roleRef:
+    apiGroup: rbac.authorization.k8s.io
+    kind: ClusterRole
+    name: patroni-k8s-ep-access
+  subjects:
+  - kind: ServiceAccount
+    name: ${SERVICE_ACCOUNT}
+    namespace: ${NAMESPACE}
 parameters:
 - description: The name of the application for labelling all artifacts.
   displayName: Application Name
@@ -300,4 +352,4 @@ parameters:
 - description: The size of the persistent volume to create. 
   displayName: Persistent Volume Size
   name: PVC_SIZE
-  value: 5Gi
+  value: 5Gi

kubernetes/openshift-example/test/README.md

@@ -1,2 +1,2 @@
 # Jenkins Test
-This pipeline test will create a separate deployment config for a pgbench pod and execute a test against the patroni cluster. This is a sample and should be customized. 
+This pipeline test will create a separate deployment config for a pgbench pod and execute a test against the patroni cluster. This is a sample and should be customized.

kubernetes/patroni_k8s.yaml

@@ -10,7 +10,7 @@ spec:
   clusterIP: None
 
 ---
-apiVersion: apps/v1beta1
+apiVersion: apps/v1
 kind: StatefulSet
 metadata:
   name: &cluster_name patronidemo
@@ -20,6 +20,10 @@ metadata:
 spec:
   replicas: 3
   serviceName: *cluster_name
+  selector:
+    matchLabels:
+      application: patroni
+      cluster-name: *cluster_name
   template:
     metadata:
       labels:
@@ -31,6 +35,16 @@ spec:
       - name: *cluster_name
         image: patroni  # docker build -t patroni .
         imagePullPolicy: IfNotPresent
+        readinessProbe:
+          httpGet:
+            scheme: HTTP
+            path: /readiness
+            port: 8008
+          initialDelaySeconds: 3
+          periodSeconds: 10
+          timeoutSeconds: 5
+          successThreshold: 1
+          failureThreshold: 3
         ports:
         - containerPort: 8008
           protocol: TCP
@@ -48,6 +62,8 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.namespace
+        - name: PATRONI_KUBERNETES_BYPASS_API_SERVICE
+          value: 'true'
         - name: PATRONI_KUBERNETES_USE_ENDPOINTS
           value: 'true'
         - name: PATRONI_KUBERNETES_LABELS
@@ -123,6 +139,25 @@ spec:
   - port: 5432
     targetPort: 5432
 
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: patronidemo-repl
+  labels:
+    application: patroni
+    cluster-name:  &cluster_name patronidemo
+    role: replica
+spec:
+  type: ClusterIP
+  selector:
+    application: patroni
+    cluster-name: *cluster_name
+    role: replica
+  ports:
+  - port: 5432
+    targetPort: 5432
+
 ---
 apiVersion: v1
 kind: Secret
@@ -210,3 +245,37 @@ roleRef:
 subjects:
 - kind: ServiceAccount
   name: patronidemo
+
+# Following privileges are only required if deployed not in the "default"
+# namespace and you want Patroni to bypass kubernetes service
+# (PATRONI_KUBERNETES_BYPASS_API_SERVICE=true)
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRole
+metadata:
+  name: patroni-k8s-ep-access
+rules:
+- apiGroups:
+  - ""
+  resources:
+  - endpoints
+  resourceNames:
+  - kubernetes
+  verbs:
+  - get
+
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRoleBinding
+metadata:
+  name: patroni-k8s-ep-access
+roleRef:
+  apiGroup: rbac.authorization.k8s.io
+  kind: ClusterRole
+  name: patroni-k8s-ep-access
+subjects:
+- kind: ServiceAccount
+  name: patronidemo
+# The namespace must be specified explicitly.
+# If deploying to the different namespace you have to change it.
+  namespace: default

mkbinary.sh

@@ -1,5 +1,5 @@
 #!/bin/sh
 set -e
 
-pip install --ignore-installed setuptools==19.2 pyinstaller
+pip install --ignore-installed pyinstaller
-pyinstaller --clean --onefile patroni.spec
+pyinstaller --clean patroni.spec

patroni.py

@@ -1,14 +1,6 @@
 #!/usr/bin/env python
-from patroni import main
+from patroni.__main__ import main
 
-import os
 
 if __name__ == '__main__':
-
-    if os.getenv("PATRONI_DEBUG_MODE"):
-        # XXX Visual Code specific https://github.com/microsoft/ptvsd/issues/1443
-        # create processes by spawning new Python interpreters instead of forking the current one 
-        import multiprocessing
-        multiprocessing.set_start_method('spawn', True)
-
     main()

patroni.spec

@@ -8,7 +8,7 @@ def hiddenimports():
     sys.path.insert(0, '.')
     try:
         import patroni.dcs
-        return patroni.dcs.dcs_modules()
+        return patroni.dcs.dcs_modules() + ['http.server']
     finally:
         sys.path.pop(0)
 

patroni/__init__.py

@@ -1,169 +1,8 @@
-import logging
-import os
-import signal
 import sys
-import time
 
-logger = logging.getLogger(__name__)
+PATRONI_ENV_PREFIX = 'PATRONI_'
-
+KUBERNETES_ENV_PREFIX = 'KUBERNETES_'
-
+MIN_PSYCOPG2 = (2, 5, 4)
-class Patroni(object):
-
-    def __init__(self):
-        from patroni.api import RestApiServer
-        from patroni.config import Config
-        from patroni.dcs import get_dcs
-        from patroni.ha import Ha
-        from patroni.log import PatroniLogger
-        from patroni.postgresql import Postgresql
-        from patroni.version import __version__
-        from patroni.watchdog import Watchdog
-
-        self.setup_signal_handlers()
-
-        self.version = __version__
-        self.logger = PatroniLogger()
-        self.config = Config()
-        self.logger.reload_config(self.config.get('log', {}))
-        self.dcs = get_dcs(self.config)
-        self.watchdog = Watchdog(self.config)
-        self.load_dynamic_configuration()
-
-        self.postgresql = Postgresql(self.config['postgresql'])
-        self.api = RestApiServer(self, self.config['restapi'])
-        self.ha = Ha(self)
-        self.is_in_debug_mode = bool(os.getenv("PATRONI_DEBUG_MODE"))
-
-        self.tags = self.get_tags()
-        self.next_run = time.time()
-        self.scheduled_restart = {}
-
-    def load_dynamic_configuration(self):
-        from patroni.exceptions import DCSError
-        while True:
-            try:
-                cluster = self.dcs.get_cluster()
-                if cluster and cluster.config and cluster.config.data:
-                    if self.config.set_dynamic_configuration(cluster.config):
-                        self.dcs.reload_config(self.config)
-                        self.watchdog.reload_config(self.config)
-                elif not self.config.dynamic_configuration and 'bootstrap' in self.config:
-                    if self.config.set_dynamic_configuration(self.config['bootstrap']['dcs']):
-                        self.dcs.reload_config(self.config)
-                break
-            except DCSError:
-                logger.warning('Can not get cluster from dcs')
-                time.sleep(5)
-
-    def get_tags(self):
-        return {tag: value for tag, value in self.config.get('tags', {}).items()
-                if tag not in ('clonefrom', 'nofailover', 'noloadbalance', 'nosync') or value}
-
-    @property
-    def nofailover(self):
-        return bool(self.tags.get('nofailover', False))
-
-    @property
-    def nosync(self):
-        return bool(self.tags.get('nosync', False))
-
-    def reload_config(self):
-        try:
-            self.tags = self.get_tags()
-            self.logger.reload_config(self.config.get('log', {}))
-            self.dcs.reload_config(self.config)
-            self.watchdog.reload_config(self.config)
-            self.api.reload_config(self.config['restapi'])
-            self.postgresql.reload_config(self.config['postgresql'])
-        except Exception:
-            logger.exception('Failed to reload config_file=%s', self.config.config_file)
-
-    @property
-    def replicatefrom(self):
-        return self.tags.get('replicatefrom')
-
-    def sighup_handler(self, *args):
-        self._received_sighup = True
-
-    def sigterm_handler(self, *args):
-        with self._sigterm_lock:
-            if not self._received_sigterm:
-                self._received_sigterm = True
-                sys.exit()
-
-    @property
-    def noloadbalance(self):
-        return bool(self.tags.get('noloadbalance', False))
-
-    def schedule_next_run(self):
-        self.next_run += self.dcs.loop_wait
-        current_time = time.time()
-        nap_time = self.next_run - current_time
-        if nap_time <= 0:
-            self.next_run = current_time
-            # Release the GIL so we don't starve anyone waiting on async_executor lock
-            time.sleep(0.001)
-            # Warn user that Patroni is not keeping up or runs in debug
-            msg = "Patroni runs in the debug mode: keys' TTL is infinite, loop wait disabled" if self.is_in_debug_mode else "Loop time exceeded, rescheduling immediately."
-            logger.warning(msg)
-        elif self.ha.watch(nap_time):
-            self.next_run = time.time()
-
-    @property
-    def received_sigterm(self):
-        with self._sigterm_lock:
-            return self._received_sigterm
-
-    def run(self):
-        self.api.start()
-        self.next_run = time.time()
-
-        while not self.received_sigterm:
-            if self._received_sighup:
-                self._received_sighup = False
-                if self.config.reload_local_configuration():
-                    self.reload_config()
-
-            logger.info(self.ha.run_cycle())
-
-            if self.dcs.cluster and self.dcs.cluster.config and self.dcs.cluster.config.data \
-                    and self.config.set_dynamic_configuration(self.dcs.cluster.config):
-                self.reload_config()
-
-            if self.postgresql.role != 'uninitialized':
-                self.config.save_cache()
-
-            self.schedule_next_run()
-
-    def setup_signal_handlers(self):
-        from threading import Lock
-
-        self._received_sighup = False
-        self._sigterm_lock = Lock()
-        self._received_sigterm = False
-        if os.name != 'nt':
-            signal.signal(signal.SIGHUP, self.sighup_handler)
-        signal.signal(signal.SIGTERM, self.sigterm_handler)
-
-    def shutdown(self):
-        with self._sigterm_lock:
-            self._received_sigterm = True
-        try:
-            self.api.shutdown()
-        except Exception:
-            logger.exception('Exception during RestApi.shutdown')
-        self.ha.shutdown()
-        self.logger.shutdown()
-
-
-def patroni_main():
-    patroni = Patroni()
-    try:
-        patroni.run()
-    except KeyboardInterrupt:
-        pass
-    finally:
-        patroni.shutdown()
 
 
 def fatal(string, *args):
@@ -171,63 +10,32 @@ def fatal(string, *args):
     sys.exit(1)
 
 
-def check_psycopg2():
+def parse_version(version):
-    min_psycopg2 = (2, 5, 4)
+    def _parse_version(version):
-    min_psycopg2_str = '.'.join(map(str, min_psycopg2))
-
-    def parse_version(version):
         for e in version.split('.'):
             try:
                 yield int(e)
             except ValueError:
                 break
+    return tuple(_parse_version(version.split(' ')[0]))
+
+
+# We pass MIN_PSYCOPG2 and parse_version as arguments to simplify usage of check_psycopg from the setup.py
+def check_psycopg(_min_psycopg2=MIN_PSYCOPG2, _parse_version=parse_version):
+    min_psycopg2_str = '.'.join(map(str, _min_psycopg2))
 
     try:
-        import psycopg2
+        from psycopg2 import __version__
-        version_str = psycopg2.__version__.split(' ')[0]
+        if _parse_version(__version__) >= _min_psycopg2:
-        version = tuple(parse_version(version_str))
+            return
-        if version < min_psycopg2:
+        version_str = __version__.split(' ')[0]
-            fatal('Patroni requires psycopg2>={0}, but only {1} is available', min_psycopg2_str, version_str)
     except ImportError:
-        fatal('Patroni requires psycopg2>={0} or psycopg2-binary', min_psycopg2_str)
+        version_str = None
 
-
+    try:
-def main():
+        from psycopg import __version__
-    check_psycopg2()
+    except ImportError:
-    if os.getpid() != 1:
+        error = 'Patroni requires psycopg2>={0}, psycopg2-binary, or psycopg>=3.0'.format(min_psycopg2_str)
-        return patroni_main()
+        if version_str:
-
+            error += ', but only psycopg2=={0} is available'.format(version_str)
-    # Patroni started with PID=1, it looks like we are in the container
+        fatal(error)
-    pid = 0
-
-    # Looks like we are in a docker, so we will act like init
-    def sigchld_handler(signo, stack_frame):
-        try:
-            while True:
-                ret = os.waitpid(-1, os.WNOHANG)
-                if ret == (0, 0):
-                    break
-                elif ret[0] != pid:
-                    logger.info('Reaped pid=%s, exit status=%s', *ret)
-        except OSError:
-            pass
-
-    def passtochild(signo, stack_frame):
-        if pid:
-            os.kill(pid, signo)
-
-    if os.name != 'nt':
-        signal.signal(signal.SIGCHLD, sigchld_handler)
-        signal.signal(signal.SIGHUP, passtochild)
-        signal.signal(signal.SIGQUIT, passtochild)
-        signal.signal(signal.SIGUSR1, passtochild)
-        signal.signal(signal.SIGUSR2, passtochild)
-    signal.signal(signal.SIGINT, passtochild)
-    signal.signal(signal.SIGABRT, passtochild)
-    signal.signal(signal.SIGTERM, passtochild)
-
-    import multiprocessing
-    patroni = multiprocessing.Process(target=patroni_main)
-    patroni.start()
-    pid = patroni.pid
-    patroni.join()

patroni/__main__.py

@@ -1,4 +1,182 @@
-from patroni import main
+import logging
+import os
+import signal
+import time
+
+from patroni.daemon import AbstractPatroniDaemon, abstract_main
+
+logger = logging.getLogger(__name__)
+
+
+class Patroni(AbstractPatroniDaemon):
+
+    def __init__(self, config):
+        from patroni.api import RestApiServer
+        from patroni.dcs import get_dcs
+        from patroni.ha import Ha
+        from patroni.postgresql import Postgresql
+        from patroni.request import PatroniRequest
+        from patroni.version import __version__
+        from patroni.watchdog import Watchdog
+
+        super(Patroni, self).__init__(config)
+
+        self.version = __version__
+        self.dcs = get_dcs(self.config)
+        self.watchdog = Watchdog(self.config)
+        self.load_dynamic_configuration()
+
+        self.postgresql = Postgresql(self.config['postgresql'])
+        self.api = RestApiServer(self, self.config['restapi'])
+        self.request = PatroniRequest(self.config, True)
+        self.ha = Ha(self)
+
+        self.tags = self.get_tags()
+        self.next_run = time.time()
+        self.scheduled_restart = {}
+
+    def load_dynamic_configuration(self):
+        from patroni.exceptions import DCSError
+        while True:
+            try:
+                cluster = self.dcs.get_cluster()
+                if cluster and cluster.config and cluster.config.data:
+                    if self.config.set_dynamic_configuration(cluster.config):
+                        self.dcs.reload_config(self.config)
+                        self.watchdog.reload_config(self.config)
+                elif not self.config.dynamic_configuration and 'bootstrap' in self.config:
+                    if self.config.set_dynamic_configuration(self.config['bootstrap']['dcs']):
+                        self.dcs.reload_config(self.config)
+                        self.watchdog.reload_config(self.config)
+                break
+            except DCSError:
+                logger.warning('Can not get cluster from dcs')
+                time.sleep(5)
+
+    def get_tags(self):
+        return {tag: value for tag, value in self.config.get('tags', {}).items()
+                if tag not in ('clonefrom', 'nofailover', 'noloadbalance', 'nosync') or value}
+
+    @property
+    def nofailover(self):
+        return bool(self.tags.get('nofailover', False))
+
+    @property
+    def nosync(self):
+        return bool(self.tags.get('nosync', False))
+
+    def reload_config(self, sighup=False, local=False):
+        try:
+            super(Patroni, self).reload_config(sighup, local)
+            if local:
+                self.tags = self.get_tags()
+                self.request.reload_config(self.config)
+            if local or sighup and self.api.reload_local_certificate():
+                self.api.reload_config(self.config['restapi'])
+            self.watchdog.reload_config(self.config)
+            self.postgresql.reload_config(self.config['postgresql'], sighup)
+            self.dcs.reload_config(self.config)
+        except Exception:
+            logger.exception('Failed to reload config_file=%s', self.config.config_file)
+
+    @property
+    def replicatefrom(self):
+        return self.tags.get('replicatefrom')
+
+    @property
+    def noloadbalance(self):
+        return bool(self.tags.get('noloadbalance', False))
+
+    def schedule_next_run(self):
+        self.next_run += self.dcs.loop_wait
+        current_time = time.time()
+        nap_time = self.next_run - current_time
+        if nap_time <= 0:
+            self.next_run = current_time
+            # Release the GIL so we don't starve anyone waiting on async_executor lock
+            time.sleep(0.001)
+            # Warn user that Patroni is not keeping up
+            logger.warning("Loop time exceeded, rescheduling immediately.")
+        elif self.ha.watch(nap_time):
+            self.next_run = time.time()
+
+    def run(self):
+        self.api.start()
+        self.next_run = time.time()
+        super(Patroni, self).run()
+
+    def _run_cycle(self):
+        logger.info(self.ha.run_cycle())
+
+        if self.dcs.cluster and self.dcs.cluster.config and self.dcs.cluster.config.data \
+                and self.config.set_dynamic_configuration(self.dcs.cluster.config):
+            self.reload_config()
+
+        if self.postgresql.role != 'uninitialized':
+            self.config.save_cache()
+
+        self.schedule_next_run()
+
+    def _shutdown(self):
+        try:
+            self.api.shutdown()
+        except Exception:
+            logger.exception('Exception during RestApi.shutdown')
+        try:
+            self.ha.shutdown()
+        except Exception:
+            logger.exception('Exception during Ha.shutdown')
+
+
+def patroni_main():
+    from multiprocessing import freeze_support
+    from patroni.validator import schema
+
+    freeze_support()
+    abstract_main(Patroni, schema)
+
+
+def main():
+    if os.getpid() != 1:
+        from patroni import check_psycopg
+
+        check_psycopg()
+        return patroni_main()
+
+    # Patroni started with PID=1, it looks like we are in the container
+    pid = 0
+
+    # Looks like we are in a docker, so we will act like init
+    def sigchld_handler(signo, stack_frame):
+        try:
+            while True:
+                ret = os.waitpid(-1, os.WNOHANG)
+                if ret == (0, 0):
+                    break
+                elif ret[0] != pid:
+                    logger.info('Reaped pid=%s, exit status=%s', *ret)
+        except OSError:
+            pass
+
+    def passtochild(signo, stack_frame):
+        if pid:
+            os.kill(pid, signo)
+
+    if os.name != 'nt':
+        signal.signal(signal.SIGCHLD, sigchld_handler)
+        signal.signal(signal.SIGHUP, passtochild)
+        signal.signal(signal.SIGQUIT, passtochild)
+        signal.signal(signal.SIGUSR1, passtochild)
+        signal.signal(signal.SIGUSR2, passtochild)
+    signal.signal(signal.SIGINT, passtochild)
+    signal.signal(signal.SIGABRT, passtochild)
+    signal.signal(signal.SIGTERM, passtochild)
+
+    import multiprocessing
+    patroni = multiprocessing.Process(target=patroni_main)
+    patroni.start()
+    pid = patroni.pid
+    patroni.join()
 
 
 if __name__ == '__main__':

patroni/api.py

@@ -1,62 +1,72 @@
 import base64
+import hmac
 import json
 import logging
-import psycopg2
 import time
 import traceback
 import dateutil.parser
 import datetime
 import os
+import six
 import socket
+import sys
 
-from patroni.postgresql import PostgresConnectionException
+from ipaddress import ip_address, ip_network as _ip_network
-from patroni.postgresql.misc import postgres_version_to_int, PostgresException
-from patroni.utils import deep_compare, parse_bool, patch_config, Retry, \
-    RetryFailedError, parse_int, split_host_port, tzutc, uri
 from six.moves.BaseHTTPServer import BaseHTTPRequestHandler, HTTPServer
 from six.moves.socketserver import ThreadingMixIn
+from six.moves.urllib_parse import urlparse, parse_qs
 from threading import Thread
 
+from . import psycopg
+from .exceptions import PostgresConnectionException, PostgresException
+from .postgresql.misc import postgres_version_to_int
+from .utils import deep_compare, enable_keepalive, parse_bool, patch_config, Retry, \
+    RetryFailedError, parse_int, split_host_port, tzutc, uri, cluster_as_json
+
 logger = logging.getLogger(__name__)
 
 
-def check_auth(func):
+def ip_network(value):
-    """Decorator function to check authorization header.
+    return _ip_network(value.decode('utf-8') if six.PY2 else value, False)
-
-    Usage example:
-    @check_auth
-    def do_PUT_foo():
-        pass
-    """
-    def wrapper(handler, *args, **kwargs):
-        if handler.check_auth_header():
-            return func(handler, *args, **kwargs)
-    return wrapper
 
 
 class RestApiHandler(BaseHTTPRequestHandler):
 
+    def _write_status_code_only(self, status_code):
+        message = self.responses[status_code][0]
+        self.wfile.write('{0} {1} {2}\r\n\r\n'.format(self.protocol_version, status_code, message).encode('utf-8'))
+        self.log_request(status_code)
+
     def _write_response(self, status_code, body, content_type='text/html', headers=None):
+        # TODO: try-catch ConnectionResetError: [Errno 104] Connection reset by peer and log it in DEBUG level
         self.send_response(status_code)
         headers = headers or {}
         if content_type:
             headers['Content-Type'] = content_type
         for name, value in headers.items():
             self.send_header(name, value)
+        for name, value in self.server.http_extra_headers.items():
+            self.send_header(name, value)
         self.end_headers()
         self.wfile.write(body.encode('utf-8'))
 
     def _write_json_response(self, status_code, response):
-        self._write_response(status_code, json.dumps(response), content_type='application/json')
+        self._write_response(status_code, json.dumps(response, default=str), content_type='application/json')
 
-    def send_auth_request(self, body):
+    def check_access(func):
-        headers = {'WWW-Authenticate': 'Basic realm="' + self.server.patroni.__class__.__name__ + '"'}
+        """Decorator function to check the source ip, authorization header. or client certificates
-        self._write_response(401, body, headers=headers)
 
-    def check_auth_header(self):
+        Usage example:
-        auth_header = self.headers.get('Authorization')
+        @check_access
-        status = self.server.check_auth_header(auth_header)
+        def do_PUT_foo():
-        return not status or self.send_auth_request(status)
+            pass
+        """
+
+        def wrapper(self, *args, **kwargs):
+            if self.server.check_access(self):
+                return func(self, *args, **kwargs)
+
+        return wrapper
 
     def _write_status_response(self, status_code, response):
         patroni = self.server.patroni
@@ -87,59 +97,132 @@ class RestApiHandler(BaseHTTPRequestHandler):
     def do_GET(self, write_status_code_only=False):
         """Default method for processing all GET requests which can not be routed to other methods"""
 
-        time_start = time.time()
+        path = '/primary' if self.path == '/' else self.path
-        request_type = 'OPTIONS' if write_status_code_only else 'GET'
-
-        path = '/master' if self.path == '/' else self.path
         response = self.get_postgresql_status()
 
         patroni = self.server.patroni
         cluster = patroni.dcs.cluster
 
-        if not cluster and patroni.ha.is_paused():
+        leader_optime = cluster and cluster.last_lsn or 0
-            primary_status_code = 200 if response['role'] == 'master' else 503
+        replayed_location = response.get('xlog', {}).get('replayed_location', 0)
-        else:
+        max_replica_lag = parse_int(self.path_query.get('lag', [sys.maxsize])[0], 'B')
-            primary_status_code = 200 if patroni.ha.is_leader() else 503
+        if max_replica_lag is None:
+            max_replica_lag = sys.maxsize
+        is_lagging = leader_optime and leader_optime > replayed_location + max_replica_lag
 
-        replica_status_code = 200 if not patroni.noloadbalance and \
+        replica_status_code = 200 if not patroni.noloadbalance and not is_lagging and \
             response.get('role') == 'replica' and response.get('state') == 'running' else 503
+
+        if not cluster and patroni.ha.is_paused():
+            leader_status_code = 200 if response.get('role') in ('master', 'primary', 'standby_leader') else 503
+            primary_status_code = 200 if response.get('role') in ('master', 'primary') else 503
+            standby_leader_status_code = 200 if response.get('role') == 'standby_leader' else 503
+        elif patroni.ha.is_leader():
+            leader_status_code = 200
+            if patroni.ha.is_standby_cluster():
+                primary_status_code = replica_status_code = 503
+                standby_leader_status_code = 200 if response.get('role') in ('replica', 'standby_leader') else 503
+            else:
+                primary_status_code = 200
+                standby_leader_status_code = 503
+        else:
+            leader_status_code = primary_status_code = standby_leader_status_code = 503
+
         status_code = 503
 
-        if patroni.ha.is_standby_cluster() and ('standby_leader' in path or 'standby-leader' in path):
+        ignore_tags = False
-            status_code = 200 if patroni.ha.is_leader() else 503
+        if 'standby_leader' in path or 'standby-leader' in path:
-        elif 'master' in path or 'leader' in path or 'primary' in path or 'read-write' in path:
+            status_code = standby_leader_status_code
+            ignore_tags = True
+        elif 'leader' in path:
+            status_code = leader_status_code
+            ignore_tags = True
+        elif 'master' in path or 'primary' in path or 'read-write' in path:
             status_code = primary_status_code
+            ignore_tags = True
         elif 'replica' in path:
             status_code = replica_status_code
-        elif 'read-only' in path:
+        elif 'read-only' in path and 'sync' not in path:
-            status_code = 200 if primary_status_code == 200 else replica_status_code
+            status_code = 200 if 200 in (primary_status_code, standby_leader_status_code) else replica_status_code
         elif 'health' in path:
             status_code = 200 if response.get('state') == 'running' else 503
         elif cluster:  # dcs is available
             is_synchronous = cluster.is_synchronous_mode() and cluster.sync \
-                    and cluster.sync.sync_standby == patroni.postgresql.name
+                    and patroni.postgresql.name in cluster.sync.members
             if path in ('/sync', '/synchronous') and is_synchronous:
                 status_code = replica_status_code
             elif path in ('/async', '/asynchronous') and not is_synchronous:
                 status_code = replica_status_code
+            elif path in ('/read-only-sync', '/read-only-synchronous'):
+                if 200 in (primary_status_code, standby_leader_status_code):
+                    status_code = 200
+                elif is_synchronous:
+                    status_code = replica_status_code
+
+        # check for user defined tags in query params
+        if not ignore_tags and status_code == 200:
+            qs_tag_prefix = "tag_"
+            for qs_key, qs_value in self.path_query.items():
+                if not qs_key.startswith(qs_tag_prefix):
+                    continue
+                qs_key = qs_key[len(qs_tag_prefix):]
+                qs_value = qs_value[0]
+                instance_tag_value = patroni.tags.get(qs_key)
+                # tag not registered for instance
+                if instance_tag_value is None:
+                    status_code = 503
+                    break
+                if not isinstance(instance_tag_value, six.string_types):
+                    instance_tag_value = str(instance_tag_value).lower()
+                if instance_tag_value != qs_value:
+                    status_code = 503
+                    break
 
         if write_status_code_only:  # when haproxy sends OPTIONS request it reads only status code and nothing more
-            message = self.responses[status_code][0]
+            self._write_status_code_only(status_code)
-            self.wfile.write('{0} {1} {2}\r\n'.format(self.protocol_version, status_code, message).encode('utf-8'))
         else:
             self._write_status_response(status_code, response)
 
-        time_end = time.time()
-        self.log_message('%s %s %s latency: %s ms', request_type, path,
-                         status_code, (time_end - time_start) * 1000)
-
     def do_OPTIONS(self):
         self.do_GET(write_status_code_only=True)
 
+    def do_HEAD(self):
+        self.do_GET(write_status_code_only=True)
+
+    def do_GET_liveness(self):
+        patroni = self.server.patroni
+        is_primary = patroni.postgresql.role in ('master', 'primary') and patroni.postgresql.is_running()
+        # We can tolerate Patroni problems longer on the replica.
+        # On the primary the liveness probe most likely will start failing only after the leader key expired.
+        # It should not be a big problem because replicas will see that the primary is still alive via REST API call.
+        liveness_threshold = patroni.dcs.ttl * (1 if is_primary else 2)
+
+        # In maintenance mode (pause) we are fine if heartbeat loop stuck.
+        status_code = 200 if patroni.ha.is_paused() or patroni.next_run + liveness_threshold > time.time() else 503
+        self._write_status_code_only(status_code)
+
+    def do_GET_readiness(self):
+        patroni = self.server.patroni
+        if patroni.ha.is_leader():
+            status_code = 200
+        elif patroni.postgresql.state == 'running':
+            status_code = 200 if patroni.dcs.cluster else 503
+        else:
+            status_code = 503
+        self._write_status_code_only(status_code)
+
     def do_GET_patroni(self):
         response = self.get_postgresql_status(True)
         self._write_status_response(200, response)
 
+    def do_GET_cluster(self):
+        cluster = self.server.patroni.dcs.get_cluster(True)
+        self._write_json_response(200, cluster_as_json(cluster))
+
+    def do_GET_history(self):
+        cluster = self.server.patroni.dcs.cluster or self.server.patroni.dcs.get_cluster()
+        self._write_json_response(200, cluster.history and cluster.history.lines or [])
+
     def do_GET_config(self):
         cluster = self.server.patroni.dcs.cluster or self.server.patroni.dcs.get_cluster()
         if cluster.config:
@@ -147,6 +230,108 @@ class RestApiHandler(BaseHTTPRequestHandler):
         else:
             self.send_error(502)
 
+    def do_GET_metrics(self):
+        postgres = self.get_postgresql_status(True)
+        patroni = self.server.patroni
+        epoch = datetime.datetime(1970, 1, 1, tzinfo=tzutc)
+
+        metrics = []
+
+        scope_label = '{{scope="{0}"}}'.format(patroni.postgresql.scope)
+        metrics.append("# HELP patroni_version Patroni semver without periods.")
+        metrics.append("# TYPE patroni_version gauge")
+        padded_semver = ''.join([x.zfill(2) for x in patroni.version.split('.')])  # 2.0.2 => 020002
+        metrics.append("patroni_version{0} {1}".format(scope_label, padded_semver))
+
+        metrics.append("# HELP patroni_postgres_running Value is 1 if Postgres is running, 0 otherwise.")
+        metrics.append("# TYPE patroni_postgres_running gauge")
+        metrics.append("patroni_postgres_running{0} {1}".format(scope_label, int(postgres['state'] == 'running')))
+
+        metrics.append("# HELP patroni_postmaster_start_time Epoch seconds since Postgres started.")
+        metrics.append("# TYPE patroni_postmaster_start_time gauge")
+        postmaster_start_time = postgres.get('postmaster_start_time')
+        postmaster_start_time = (postmaster_start_time - epoch).total_seconds() if postmaster_start_time else 0
+        metrics.append("patroni_postmaster_start_time{0} {1}".format(scope_label, postmaster_start_time))
+
+        metrics.append("# HELP patroni_master Value is 1 if this node is the leader, 0 otherwise.")
+        metrics.append("# TYPE patroni_master gauge")
+        metrics.append("patroni_master{0} {1}".format(scope_label, int(postgres['role'] in ('master', 'primary'))))
+
+        metrics.append("# HELP patroni_primary Value is 1 if this node is the leader, 0 otherwise.")
+        metrics.append("# TYPE patroni_primary gauge")
+        metrics.append("patroni_primary{0} {1}".format(scope_label, int(postgres['role'] in ('master', 'primary'))))
+
+        metrics.append("# HELP patroni_xlog_location Current location of the Postgres"
+                       " transaction log, 0 if this node is not the leader.")
+        metrics.append("# TYPE patroni_xlog_location counter")
+        metrics.append("patroni_xlog_location{0} {1}".format(scope_label, postgres.get('xlog', {}).get('location', 0)))
+
+        metrics.append("# HELP patroni_standby_leader Value is 1 if this node is the standby_leader, 0 otherwise.")
+        metrics.append("# TYPE patroni_standby_leader gauge")
+        metrics.append("patroni_standby_leader{0} {1}".format(scope_label, int(postgres['role'] == 'standby_leader')))
+
+        metrics.append("# HELP patroni_replica Value is 1 if this node is a replica, 0 otherwise.")
+        metrics.append("# TYPE patroni_replica gauge")
+        metrics.append("patroni_replica{0} {1}".format(scope_label, int(postgres['role'] == 'replica')))
+
+        metrics.append("# HELP patroni_xlog_received_location Current location of the received"
+                       " Postgres transaction log, 0 if this node is not a replica.")
+        metrics.append("# TYPE patroni_xlog_received_location counter")
+        metrics.append("patroni_xlog_received_location{0} {1}"
+                       .format(scope_label, postgres.get('xlog', {}).get('received_location', 0)))
+
+        metrics.append("# HELP patroni_xlog_replayed_location Current location of the replayed"
+                       " Postgres transaction log, 0 if this node is not a replica.")
+        metrics.append("# TYPE patroni_xlog_replayed_location counter")
+        metrics.append("patroni_xlog_replayed_location{0} {1}"
+                       .format(scope_label, postgres.get('xlog', {}).get('replayed_location', 0)))
+
+        metrics.append("# HELP patroni_xlog_replayed_timestamp Current timestamp of the replayed"
+                       " Postgres transaction log, 0 if null.")
+        metrics.append("# TYPE patroni_xlog_replayed_timestamp gauge")
+        replayed_timestamp = postgres.get('xlog', {}).get('replayed_timestamp')
+        replayed_timestamp = (replayed_timestamp - epoch).total_seconds() if replayed_timestamp else 0
+        metrics.append("patroni_xlog_replayed_timestamp{0} {1}".format(scope_label, replayed_timestamp))
+
+        metrics.append("# HELP patroni_xlog_paused Value is 1 if the Postgres xlog is paused, 0 otherwise.")
+        metrics.append("# TYPE patroni_xlog_paused gauge")
+        metrics.append("patroni_xlog_paused{0} {1}"
+                       .format(scope_label, int(postgres.get('xlog', {}).get('paused', False) is True)))
+
+        metrics.append("# HELP patroni_postgres_server_version Version of Postgres (if running), 0 otherwise.")
+        metrics.append("# TYPE patroni_postgres_server_version gauge")
+        metrics.append("patroni_postgres_server_version {0} {1}".format(scope_label, postgres.get('server_version', 0)))
+
+        metrics.append("# HELP patroni_cluster_unlocked Value is 1 if the cluster is unlocked, 0 if locked.")
+        metrics.append("# TYPE patroni_cluster_unlocked gauge")
+        metrics.append("patroni_cluster_unlocked{0} {1}".format(scope_label, int(postgres.get('cluster_unlocked', 0))))
+
+        metrics.append("# HELP patroni_failsafe_mode_is_active Value is 1 if the cluster is unlocked, 0 if locked.")
+        metrics.append("# TYPE patroni_failsafe_mode_is_active gauge")
+        metrics.append("patroni_failsafe_mode_is_active{0} {1}"
+                       .format(scope_label, int(postgres.get('failsafe_mode_is_active', 0))))
+
+        metrics.append("# HELP patroni_postgres_timeline Postgres timeline of this node (if running), 0 otherwise.")
+        metrics.append("# TYPE patroni_postgres_timeline counter")
+        metrics.append("patroni_postgres_timeline{0} {1}".format(scope_label, postgres.get('timeline', 0)))
+
+        metrics.append("# HELP patroni_dcs_last_seen Epoch timestamp when DCS was last contacted successfully"
+                       " by Patroni.")
+        metrics.append("# TYPE patroni_dcs_last_seen gauge")
+        metrics.append("patroni_dcs_last_seen{0} {1}".format(scope_label, postgres.get('dcs_last_seen', 0)))
+
+        metrics.append("# HELP patroni_pending_restart Value is 1 if the node needs a restart, 0 otherwise.")
+        metrics.append("# TYPE patroni_pending_restart gauge")
+        metrics.append("patroni_pending_restart{0} {1}"
+                       .format(scope_label, int(patroni.postgresql.pending_restart)))
+
+        metrics.append("# HELP patroni_is_paused Value is 1 if auto failover is disabled, 0 otherwise.")
+        metrics.append("# TYPE patroni_is_paused gauge")
+        metrics.append("patroni_is_paused{0} {1}"
+                       .format(scope_label, int(patroni.ha.is_paused())))
+
+        self._write_response(200, '\n'.join(metrics)+'\n', content_type='text/plain')
+
     def _read_json_content(self, body_is_optional=False):
         if 'content-length' not in self.headers:
             return self.send_error(411) if not body_is_optional else {}
@@ -161,11 +346,13 @@ class RestApiHandler(BaseHTTPRequestHandler):
             logger.exception('Bad request')
         self.send_error(400)
 
-    @check_auth
+    @check_access
     def do_PATCH_config(self):
         request = self._read_json_content()
         if request:
-            cluster = self.server.patroni.dcs.get_cluster()
+            cluster = self.server.patroni.dcs.get_cluster(True)
+            if not (cluster.config and cluster.config.modify_index):
+                return self.send_error(503)
             data = cluster.config.data.copy()
             if patch_config(data, request):
                 value = json.dumps(data, separators=(',', ':'))
@@ -174,7 +361,7 @@ class RestApiHandler(BaseHTTPRequestHandler):
             self.server.patroni.ha.wakeup()
             self._write_json_response(200, data)
 
-    @check_auth
+    @check_access
     def do_PUT_config(self):
         request = self._read_json_content()
         if request:
@@ -185,20 +372,36 @@ class RestApiHandler(BaseHTTPRequestHandler):
                     return self.send_error(502)
             self._write_json_response(200, request)
 
-    @check_auth
+    @check_access
     def do_POST_reload(self):
-        try:
+        self.server.patroni.sighup_handler()
-            if self.server.patroni.config.reload_local_configuration(True):
+        self._write_response(202, 'reload scheduled')
-                status_code = 202
+
-                response = 'reload scheduled'
+    def do_GET_failsafe(self):
-                self.server.patroni.sighup_handler()
+        failsafe = self.server.patroni.dcs.failsafe
-            else:
+        if isinstance(failsafe, dict):
-                status_code = 200
+            self._write_json_response(200, failsafe)
-                response = 'nothing changed'
+        else:
-        except Exception as e:
+            self.send_error(502)
-            status_code = 500
+
-            response = str(e)
+    @check_access
-        self._write_response(status_code, response)
+    def do_POST_failsafe(self):
+        if self.server.patroni.ha.is_failsafe_mode():
+            request = self._read_json_content()
+            if request:
+                message = self.server.patroni.ha.update_failsafe(request) or 'Accepted'
+                code = 200 if message == 'Accepted' else 500
+                self._write_response(code, message)
+        else:
+            self.send_error(502)
+
+    @check_access
+    def do_POST_sigterm(self):
+        """Only for behave testing on windows"""
+
+        if os.name == 'nt' and os.getenv('BEHAVE_DEBUG'):
+            self.server.patroni.api_sigterm()
+        self._write_response(202, 'shutdown scheduled')
 
     @staticmethod
     def parse_schedule(schedule, action):
@@ -221,7 +424,7 @@ class RestApiHandler(BaseHTTPRequestHandler):
             status_code = 422
         return (status_code, error, scheduled_at)
 
-    @check_auth
+    @check_access
     def do_POST_restart(self):
         status_code = 500
         data = 'restart failed'
@@ -244,9 +447,9 @@ class RestApiHandler(BaseHTTPRequestHandler):
                     status_code = _
                     break
             elif k == 'role':
-                if request[k] not in ('master', 'replica'):
+                if request[k] not in ('master', 'primary', 'replica'):
                     status_code = 400
-                    data = "PostgreSQL role should be either master or replica"
+                    data = "PostgreSQL role should be either primary or replica"
                     break
             elif k == 'postgres_version':
                 try:
@@ -282,7 +485,7 @@ class RestApiHandler(BaseHTTPRequestHandler):
                     status_code = 409
         self._write_response(status_code, data)
 
-    @check_auth
+    @check_access
     def do_DELETE_restart(self):
         if self.server.patroni.ha.delete_future_restart():
             data = "scheduled restart deleted"
@@ -292,7 +495,21 @@ class RestApiHandler(BaseHTTPRequestHandler):
             code = 404
         self._write_response(code, data)
 
-    @check_auth
+    @check_access
+    def do_DELETE_switchover(self):
+        failover = self.server.patroni.dcs.get_cluster().failover
+        if failover and failover.scheduled_at:
+            if not self.server.patroni.dcs.manual_failover('', '', index=failover.index):
+                return self.send_error(409)
+            else:
+                data = "scheduled switchover deleted"
+                code = 200
+        else:
+            data = "no switchover is scheduled"
+            code = 404
+        self._write_response(code, data)
+
+    @check_access
     def do_POST_reinitialize(self):
         request = self._read_json_content(body_is_optional=True)
 
@@ -324,20 +541,20 @@ class RestApiHandler(BaseHTTPRequestHandler):
                 if not cluster.failover:
                     return 503, action.title() + ' failed'
             except Exception as e:
-                logger.debug('Exception occured during polling %s result: %s', action, e)
+                logger.debug('Exception occurred during polling %s result: %s', action, e)
         return 503, action.title() + ' status unknown'
 
     def is_failover_possible(self, cluster, leader, candidate, action):
         if leader and (not cluster.leader or cluster.leader.name != leader):
             return 'leader name does not match'
         if candidate:
-            if action == 'switchover' and cluster.is_synchronous_mode() and cluster.sync.sync_standby != candidate:
+            if action == 'switchover' and cluster.is_synchronous_mode() and candidate not in cluster.sync.members:
                 return 'candidate name does not match with sync_standby'
             members = [m for m in cluster.members if m.name == candidate]
             if not members:
                 return 'candidate does not exists'
         elif cluster.is_synchronous_mode():
-            members = [m for m in cluster.members if m.name == cluster.sync.sync_standby]
+            members = [m for m in cluster.members if m.name in cluster.sync.members]
             if not members:
                 return action + ' is not possible: can not find sync_standby'
         else:
@@ -349,7 +566,7 @@ class RestApiHandler(BaseHTTPRequestHandler):
                 return None
         return action + ' is not possible: no good candidates have been found'
 
-    @check_auth
+    @check_access
     def do_POST_failover(self, action='failover'):
         request = self._read_json_content()
         (status_code, data) = (400, '')
@@ -402,6 +619,18 @@ class RestApiHandler(BaseHTTPRequestHandler):
     def do_POST_switchover(self):
         self.do_POST_failover(action='switchover')
 
+    @check_access
+    def do_POST_citus(self):
+        request = self._read_json_content()
+        if not request:
+            return
+
+        patroni = self.server.patroni
+        if patroni.postgresql.citus_handler.is_coordinator() and patroni.ha.is_leader():
+            cluster = patroni.dcs.get_cluster(True)
+            patroni.postgresql.citus_handler.handle_event(cluster, request)
+        self._write_response(200, 'OK')
+
     def parse_request(self):
         """Override parse_request method to enrich basic functionality of `BaseHTTPRequestHandler` class
 
@@ -414,6 +643,9 @@ class RestApiHandler(BaseHTTPRequestHandler):
 
         ret = BaseHTTPRequestHandler.parse_request(self)
         if ret:
+            urlpath = urlparse(self.path)
+            self.path = urlpath.path
+            self.path_query = parse_qs(urlpath.query) or {}
             mname = self.path.lstrip('/').split('/')[0]
             mname = self.command + ('_' + mname if mname else '')
             if hasattr(self, 'do_' + mname):
@@ -427,77 +659,81 @@ class RestApiHandler(BaseHTTPRequestHandler):
         return retry(self.server.query, sql, *params)
 
     def get_postgresql_status(self, retry=False):
+        postgresql = self.server.patroni.postgresql
         try:
             cluster = self.server.patroni.dcs.cluster
 
-            if self.server.patroni.postgresql.state not in ('running', 'restarting', 'starting'):
+            if postgresql.state not in ('running', 'restarting', 'starting'):
                 raise RetryFailedError('')
-            stmt = ("WITH replication_info AS ("
+            stmt = ("SELECT " + postgresql.POSTMASTER_START_TIME + ", " + postgresql.TL_LSN + ","
-                    "SELECT usename, application_name, client_addr, state, sync_state, sync_priority"
+                    " pg_catalog.pg_last_xact_replay_timestamp(),"
-                    " FROM pg_catalog.pg_stat_replication) SELECT"
+                    " pg_catalog.array_to_json(pg_catalog.array_agg(pg_catalog.row_to_json(ri))) "
-                    " pg_catalog.to_char(pg_catalog.pg_postmaster_start_time(), 'YYYY-MM-DD HH24:MI:SS.MS TZ'),"
+                    "FROM (SELECT (SELECT rolname FROM pg_catalog.pg_authid WHERE oid = usesysid) AS usename,"
-                    " CASE WHEN pg_catalog.pg_is_in_recovery() THEN 0"
+                    " application_name, client_addr, w.state, sync_state, sync_priority"
-                    " ELSE ('x' || pg_catalog.substr(pg_catalog.pg_{0}file_name("
+                    " FROM pg_catalog.pg_stat_get_wal_senders() w, pg_catalog.pg_stat_get_activity(pid)) AS ri")
-                    "pg_catalog.pg_current_{0}_{1}()), 1, 8))::bit(32)::int END,"
-                    " CASE WHEN pg_catalog.pg_is_in_recovery() THEN 0"
-                    " ELSE pg_catalog.pg_{0}_{1}_diff(pg_catalog.pg_current_{0}_{1}(), '0/0')::bigint END,"
-                    " pg_catalog.pg_{0}_{1}_diff(COALESCE(pg_catalog.pg_last_{0}_receive_{1}(),"
-                    " pg_catalog.pg_last_{0}_replay_{1}()), '0/0')::bigint,"
-                    " pg_catalog.pg_{0}_{1}_diff(pg_catalog.pg_last_{0}_replay_{1}(), '0/0')::bigint,"
-                    " pg_catalog.to_char(pg_catalog.pg_last_xact_replay_timestamp(), 'YYYY-MM-DD HH24:MI:SS.MS TZ'),"
-                    " pg_catalog.pg_is_in_recovery() AND pg_catalog.pg_is_{0}_replay_paused(), "
-                    "(SELECT pg_catalog.array_to_json(pg_catalog.array_agg("
-                    "pg_catalog.row_to_json(ri))) FROM replication_info ri)")
 
-            row = self.query(stmt.format(self.server.patroni.postgresql.wal_name,
+            row = self.query(stmt.format(postgresql.wal_name, postgresql.lsn_name), retry=retry)[0]
-                                         self.server.patroni.postgresql.lsn_name), retry=retry)[0]
 
             result = {
-                'state': self.server.patroni.postgresql.state,
+                'state': postgresql.state,
                 'postmaster_start_time': row[0],
                 'role': 'replica' if row[1] == 0 else 'master',
-                'server_version': self.server.patroni.postgresql.server_version,
+                'server_version': postgresql.server_version,
-                'cluster_unlocked': bool(not cluster or cluster.is_unlocked()),
                 'xlog': ({
-                    'received_location': row[3],
+                    'received_location': row[4] or row[3],
-                    'replayed_location': row[4],
+                    'replayed_location': row[3],
-                    'replayed_timestamp': row[5],
+                    'replayed_timestamp': row[6],
-                    'paused': row[6]} if row[1] == 0 else {
+                    'paused': row[5]} if row[1] == 0 else {
                     'location': row[2]
                 })
             }
 
             if result['role'] == 'replica' and self.server.patroni.ha.is_standby_cluster():
-                result['role'] = self.server.patroni.postgresql.role
+                result['role'] = postgresql.role
 
             if row[1] > 0:
                 result['timeline'] = row[1]
             else:
                 leader_timeline = None if not cluster or cluster.is_unlocked() else cluster.leader.timeline
-                result['timeline'] = self.server.patroni.postgresql.replica_cached_timeline(leader_timeline)
+                result['timeline'] = postgresql.replica_cached_timeline(leader_timeline)
 
             if row[7]:
                 result['replication'] = row[7]
 
-            return result
+        except (psycopg.Error, RetryFailedError, PostgresConnectionException):
-        except (psycopg2.Error, RetryFailedError, PostgresConnectionException):
+            state = postgresql.state
-            state = self.server.patroni.postgresql.state
             if state == 'running':
                 logger.exception('get_postgresql_status')
                 state = 'unknown'
-            return {'state': state, 'role': self.server.patroni.postgresql.role}
+            result = {'state': state, 'role': postgresql.role}
+
+        if not cluster or cluster.is_unlocked():
+            result['cluster_unlocked'] = True
+        if self.server.patroni.ha.failsafe_is_active():
+            result['failsafe_mode_is_active'] = True
+        result['dcs_last_seen'] = self.server.patroni.dcs.last_seen
+        return result
+
+    def handle_one_request(self):
+        self.__start_time = time.time()
+        BaseHTTPRequestHandler.handle_one_request(self)
 
     def log_message(self, fmt, *args):
-        logger.debug("API thread: %s - - [%s] %s", self.client_address[0], self.log_date_time_string(), fmt % args)
+        latency = 1000.0 * (time.time() - self.__start_time)
+        logger.debug("API thread: %s - - %s latency: %0.3f ms", self.client_address[0], fmt % args, latency)
 
 
 class RestApiServer(ThreadingMixIn, HTTPServer, Thread):
+    # On 3.7+ the `ThreadingMixIn` gathers all non-daemon worker threads in order to join on them at server close.
+    daemon_threads = True  # Make worker threads "fire and forget" to prevent a memory leak.
 
     def __init__(self, patroni, config):
         self.patroni = patroni
         self.__listen = None
-        self.__initialize(config)
+        self.__ssl_options = None
-        self.__set_config_parameters(config)
+        self.__ssl_serial_number = None
+        self._received_new_cert = False
+        self.reload_config(config)
         self.daemon = True
 
     def query(self, sql, *params):
@@ -506,7 +742,7 @@ class RestApiServer(ThreadingMixIn, HTTPServer, Thread):
             with self.patroni.postgresql.connection().cursor() as cursor:
                 cursor.execute(sql, params)
                 return [r for r in cursor]
-        except psycopg2.Error as e:
+        except psycopg.Error as e:
             if cursor and cursor.connection.closed == 0:
                 raise e
             raise PostgresConnectionException('connection problems')
@@ -519,7 +755,7 @@ class RestApiServer(ThreadingMixIn, HTTPServer, Thread):
             fcntl.fcntl(fd, fcntl.F_SETFD, flags | fcntl.FD_CLOEXEC)
 
     def check_basic_auth_key(self, key):
-        return self.__auth_key == key
+        return hmac.compare_digest(self.__auth_key, key.encode('utf-8'))
 
     def check_auth_header(self, auth_header):
         if self.__auth_key:
@@ -529,12 +765,44 @@ class RestApiServer(ThreadingMixIn, HTTPServer, Thread):
                 return 'not authenticated'
 
     @staticmethod
-    def __get_ssl_options(config):
+    def __resolve_ips(host, port):
-        return {option: config[option] for option in ['certfile', 'keyfile'] if option in config}
+        try:
+            for _, _, _, _, sa in socket.getaddrinfo(host, port, 0, socket.SOCK_STREAM, socket.IPPROTO_TCP):
+                yield ip_network(sa[0])
+        except Exception as e:
+            logger.error('Failed to resolve %s: %r', host, e)
 
-    def __set_config_parameters(self, config):
+    def __members_ips(self):
-        self.__auth_key = base64.b64encode(config['auth'].encode('utf-8')).decode('utf-8') if 'auth' in config else None
+        cluster = self.patroni.dcs.cluster
-        self.connection_string = uri(self.__protocol, config.get('connect_address') or self.__listen, 'patroni')
+        if self.__allowlist_include_members and cluster:
+            for cluster in [cluster] + list(cluster.workers.values()):
+                for member in cluster.members:
+                    if member.api_url:
+                        try:
+                            r = urlparse(member.api_url)
+                            host = r.hostname
+                            port = r.port or (443 if r.scheme == 'https' else 80)
+                            for ip in self.__resolve_ips(host, port):
+                                yield ip
+                        except Exception as e:
+                            logger.debug('Failed to parse url %s: %r', member.api_url, e)
+
+    def check_access(self, rh):
+        if self.__allowlist or self.__allowlist_include_members:
+            incoming_ip = rh.client_address[0]
+            incoming_ip = ip_address(incoming_ip.decode('utf-8') if six.PY2 else incoming_ip)
+            if not any(incoming_ip in net for net in self.__allowlist + tuple(self.__members_ips())):
+                return rh._write_response(403, 'Access is denied')
+
+        if not hasattr(rh.request, 'getpeercert') or not rh.request.getpeercert():  # valid client cert isn't present
+            if self.__protocol == 'https' and self.__ssl_options.get('verify_client') in ('required', 'optional'):
+                return rh._write_response(403, 'client certificate required')
+
+        reason = self.check_auth_header(rh.headers.get('Authorization'))
+        if reason:
+            headers = {'WWW-Authenticate': 'Basic realm="' + self.patroni.__class__.__name__ + '"'}
+            return rh._write_response(401, reason, headers=headers)
+        return True
 
     @staticmethod
     def __has_dual_stack():
@@ -553,7 +821,7 @@ class RestApiServer(ThreadingMixIn, HTTPServer, Thread):
 
     def __httpserver_init(self, host, port):
         dual_stack = self.__has_dual_stack()
-        if host == '':
+        if host in ('', '*'):
             host = None
 
         info = socket.getaddrinfo(host, port, socket.AF_UNSPEC, socket.SOCK_STREAM, 0, socket.AI_PASSIVE)
@@ -561,48 +829,123 @@ class RestApiServer(ThreadingMixIn, HTTPServer, Thread):
         info.sort(key=lambda x: x[0] == socket.AF_INET, reverse=not dual_stack)
 
         self.address_family = info[0][0]
-        HTTPServer.__init__(self, info[0][-1][:2], RestApiHandler)
-
-    def __initialize(self, config):
         try:
-            host, port = split_host_port(config['listen'], None)
+            HTTPServer.__init__(self, info[0][-1][:2], RestApiHandler)
+        except socket.error:
+            logger.error(
+                    "Couldn't start a service on '%s:%s', please check your `restapi.listen` configuration", host, port)
+            raise
+
+    def __initialize(self, listen, ssl_options):
+        try:
+            host, port = split_host_port(listen, None)
         except Exception:
             raise ValueError('Invalid "restapi" config: expected <HOST>:<PORT> for "listen", but got "{0}"'
-                             .format(config['listen']))
+                             .format(listen))
 
-        if self.__listen is not None:  # changing config in runtime
+        reloading_config = self.__listen is not None  # changing config in runtime
+        if reloading_config:
             self.shutdown()
+            # Rely on ThreadingMixIn.server_close() to have all requests terminate before we continue
+            self.server_close()
 
-        self.__listen = config['listen']
+        self.__listen = listen
-        self.__ssl_options = self.__get_ssl_options(config)
+        self.__ssl_options = ssl_options
+        self._received_new_cert = False  # reset to False after reload_config()
 
         self.__httpserver_init(host, port)
         Thread.__init__(self, target=self.serve_forever)
         self._set_fd_cloexec(self.socket)
 
-        self.__protocol = 'http'
-
         # wrap socket with ssl if 'certfile' is defined in a config.yaml
         # Sometime it's also needed to pass reference to a 'keyfile'.
+        self.__protocol = 'https' if ssl_options.get('certfile') else 'http'
+        if self.__protocol == 'https':
+            import ssl
+            ctx = ssl.create_default_context(ssl.Purpose.CLIENT_AUTH, cafile=ssl_options.get('cafile'))
+            if ssl_options.get('ciphers'):
+                ctx.set_ciphers(ssl_options['ciphers'])
+            ctx.load_cert_chain(certfile=ssl_options['certfile'], keyfile=ssl_options.get('keyfile'),
+                                password=ssl_options.get('keyfile_password'))
+            verify_client = ssl_options.get('verify_client')
+            if verify_client:
+                modes = {'none': ssl.CERT_NONE, 'optional': ssl.CERT_OPTIONAL, 'required': ssl.CERT_REQUIRED}
+                if verify_client in modes:
+                    ctx.verify_mode = modes[verify_client]
+                else:
+                    logger.error('Bad value in the "restapi.verify_client": %s', verify_client)
+            self.__ssl_serial_number = self.get_certificate_serial_number()
+            self.socket = ctx.wrap_socket(self.socket, server_side=True, do_handshake_on_connect=False)
+        if reloading_config:
+            self.start()
+
+    def process_request_thread(self, request, client_address):
+        enable_keepalive(request, 10, 3)
+        if hasattr(request, 'context'):  # SSLSocket
+            request.do_handshake()
+        super(RestApiServer, self).process_request_thread(request, client_address)
+
+    def shutdown_request(self, request):
+        if hasattr(request, 'context'):  # SSLSocket
+            try:
+                request.unwrap()
+            except Exception as e:
+                logger.debug('Failed to shutdown SSL connection: %r', e)
+        super(RestApiServer, self).shutdown_request(request)
+
+    def get_certificate_serial_number(self):
         if self.__ssl_options.get('certfile'):
             import ssl
-            ctx = ssl.create_default_context(ssl.Purpose.CLIENT_AUTH)
+            try:
-            ctx.load_cert_chain(**self.__ssl_options)
+                crt = ssl._ssl._test_decode_cert(self.__ssl_options['certfile'])
-            self.socket = ctx.wrap_socket(self.socket, server_side=True)
+                return crt.get('serialNumber')
-            self.__protocol = 'https'
+            except ssl.SSLError as e:
-        return True
+                logger.error('Failed to get serial number from certificate %s: %r', self.__ssl_options['certfile'], e)
+
+    def reload_local_certificate(self):
+        if self.__protocol == 'https':
+            on_disk_cert_serial_number = self.get_certificate_serial_number()
+            if on_disk_cert_serial_number != self.__ssl_serial_number:
+                self._received_new_cert = True
+                self.__ssl_serial_number = on_disk_cert_serial_number
+                return True
+
+    def _build_allowlist(self, value):
+        if isinstance(value, list):
+            for v in value:
+                if '/' in v:  # netmask
+                    try:
+                        yield ip_network(v)
+                    except Exception as e:
+                        logger.error('Invalid value "%s" in the allowlist: %r', v, e)
+                else:  # ip or hostname, try to resolve it
+                    for ip in self.__resolve_ips(v, 8080):
+                        yield ip
 
     def reload_config(self, config):
         if 'listen' not in config:  # changing config in runtime
             raise ValueError('Can not find "restapi.listen" config')
 
-        elif (self.__listen != config['listen'] or self.__ssl_options != self.__get_ssl_options(config)) \
+        self.__allowlist = tuple(self._build_allowlist(config.get('allowlist')))
-                and self.__initialize(config):
+        self.__allowlist_include_members = config.get('allowlist_include_members')
-            self.start()
+
-        self.__set_config_parameters(config)
+        ssl_options = {n: config[n] for n in ('certfile', 'keyfile', 'keyfile_password',
+                                              'cafile', 'ciphers') if n in config}
+
+        self.http_extra_headers = config.get('http_extra_headers') or {}
+        self.http_extra_headers.update((config.get('https_extra_headers') or {}) if ssl_options.get('certfile') else {})
+
+        if isinstance(config.get('verify_client'), six.string_types):
+            ssl_options['verify_client'] = config['verify_client'].lower()
+
+        if self.__listen != config['listen'] or self.__ssl_options != ssl_options or self._received_new_cert:
+            self.__initialize(config['listen'], ssl_options)
+
+        self.__auth_key = base64.b64encode(config['auth'].encode('utf-8')) if 'auth' in config else None
+        self.connection_string = uri(self.__protocol, config.get('connect_address') or self.__listen, 'patroni')
 
     @staticmethod
     def handle_error(request, client_address):
-        address, port = client_address
+        logger.warning('Exception happened during processing of request from %s:%s',
-        logger.warning('Exception happened during processing of request from {}:{}'.format(address, port))
+                       client_address[0], client_address[1])
         logger.warning(traceback.format_exc())

patroni/async_executor.py

@@ -110,6 +110,12 @@ class AsyncExecutor(object):
     def run_async(self, func, args=()):
         Thread(target=self.run, args=(func, args)).start()
 
+    def try_run_async(self, action, func, args=()):
+        prev = self.schedule(action)
+        if prev is None:
+            return self.run_async(func, args)
+        return 'Failed to run {0}, {1} is already in progress'.format(action, prev)
+
     def cancel(self):
         with self:
             with self._scheduled_action_lock:

patroni/config.py

@@ -2,19 +2,39 @@ import json
 import logging
 import os
 import shutil
-import sys
+import six
 import tempfile
 import yaml
 
 from collections import defaultdict
 from copy import deepcopy
+from patroni import PATRONI_ENV_PREFIX
+from patroni.exceptions import ConfigParseError
 from patroni.dcs import ClusterConfig
-from patroni.postgresql.config import ConfigHandler
+from patroni.postgresql.config import CaseInsensitiveDict, ConfigHandler
 from patroni.utils import deep_compare, parse_bool, parse_int, patch_config
-from requests.structures import CaseInsensitiveDict
 
 logger = logging.getLogger(__name__)
 
+_AUTH_ALLOWED_PARAMETERS = (
+    'username',
+    'password',
+    'sslmode',
+    'sslcert',
+    'sslkey',
+    'sslpassword',
+    'sslrootcert',
+    'sslcrl',
+    'sslcrldir',
+    'gssencmode',
+    'channel_binding'
+)
+
+
+def default_validator(conf):
+    if not conf:
+        raise ConfigParseError("Config is empty.")
+
 
 class Config(object):
     """
@@ -36,17 +56,24 @@ class Config(object):
          to work with it as with the old `config` object.
     """
 
-    PATRONI_ENV_PREFIX = 'PATRONI_'
     PATRONI_CONFIG_VARIABLE = PATRONI_ENV_PREFIX + 'CONFIGURATION'
 
     __CACHE_FILENAME = 'patroni.dynamic.json'
+    __REMAP_KEYS = {
+        'master_start_timeout': 'primary_start_timeout',
+        'master_stop_timeout': 'primary_stop_timeout'
+    }
     __DEFAULT_CONFIG = {
         'ttl': 30, 'loop_wait': 10, 'retry_timeout': 10,
         'maximum_lag_on_failover': 1048576,
+        'maximum_lag_on_syncnode': -1,
         'check_timeline': False,
-        'master_start_timeout': 300,
+        'primary_start_timeout': 300,
+        'primary_stop_timeout': 0,
         'synchronous_mode': False,
         'synchronous_mode_strict': False,
+        'synchronous_node_count': 1,
+        'failsafe_mode': False,
         'standby_cluster': {
             'create_replica_methods': '',
             'host': '',
@@ -59,34 +86,34 @@ class Config(object):
         'postgresql': {
             'bin_dir': '',
             'use_slots': True,
-            'parameters': CaseInsensitiveDict({p: v[0] for p, v in ConfigHandler.CMDLINE_OPTIONS.items()})
+            'parameters': CaseInsensitiveDict({p: v[0] for p, v in ConfigHandler.CMDLINE_OPTIONS.items()
+                                               if p not in ('wal_keep_segments', 'wal_keep_size')})
         },
         'watchdog': {
             'mode': 'automatic',
         }
     }
 
-    def __init__(self):
+    def __init__(self, configfile, validator=default_validator):
         self._modify_index = -1
         self._dynamic_configuration = {}
 
         self.__environment_configuration = self._build_environment_configuration()
 
         # Patroni reads the configuration from the command-line argument if it exists, otherwise from the environment
-        self._config_file = len(sys.argv) >= 2 and os.path.isfile(sys.argv[1]) and sys.argv[1]
+        self._config_file = configfile and os.path.exists(configfile) and configfile
         if self._config_file:
             self._local_configuration = self._load_config_file()
         else:
             config_env = os.environ.pop(self.PATRONI_CONFIG_VARIABLE, None)
             self._local_configuration = config_env and yaml.safe_load(config_env) or self.__environment_configuration
-            if not self._local_configuration:
+        if validator:
-                print('Usage: {0} config.yml'.format(sys.argv[0]))
+            errors = validator(self._local_configuration)
-                print('\tPatroni may also read the configuration from the {0} environment variable'.
+            if errors:
-                      format(self.PATRONI_CONFIG_VARIABLE))
+                raise ConfigParseError("\n".join(errors))
-                sys.exit(1)
 
         self.__effective_configuration = self._build_effective_configuration({}, self._local_configuration)
-        self._data_dir = self.__effective_configuration['postgresql']['data_dir']
+        self._data_dir = self.__effective_configuration.get('postgresql', {}).get('data_dir', "")
         self._cache_file = os.path.join(self._data_dir, self.__CACHE_FILENAME)
         self._load_cache()
         self._cache_needs_saving = False
@@ -102,12 +129,32 @@ class Config(object):
     def check_mode(self, mode):
         return bool(parse_bool(self._dynamic_configuration.get(mode)))
 
+    def _load_config_path(self, path):
+        """
+        If path is a file, loads the yml file pointed to by path.
+        If path is a directory, loads all yml files in that directory in alphabetical order
+        """
+        if os.path.isfile(path):
+            files = [path]
+        elif os.path.isdir(path):
+            files = [os.path.join(path, f) for f in sorted(os.listdir(path))
+                     if (f.endswith('.yml') or f.endswith('.yaml')) and os.path.isfile(os.path.join(path, f))]
+        else:
+            logger.error('config path %s is neither directory nor file', path)
+            raise ConfigParseError('invalid config path')
+
+        overall_config = {}
+        for fname in files:
+            with open(fname) as f:
+                config = yaml.safe_load(f)
+                patch_config(overall_config, config)
+        return overall_config
+
     def _load_config_file(self):
         """Loads config.yaml from filesystem and applies some values which were set via ENV"""
-        with open(self._config_file) as f:
+        config = self._load_config_path(self._config_file)
-            config = yaml.safe_load(f)
+        patch_config(config, self.__environment_configuration)
-            patch_config(config, self.__environment_configuration)
+        return config
-            return config
 
     def _load_cache(self):
         if os.path.isfile(self._cache_file):
@@ -158,23 +205,19 @@ class Config(object):
             except Exception:
                 logger.exception('Exception when setting dynamic_configuration')
 
-    def reload_local_configuration(self, dry_run=False):
+    def reload_local_configuration(self):
         if self.config_file:
             try:
                 configuration = self._load_config_file()
                 if not deep_compare(self._local_configuration, configuration):
                     new_configuration = self._build_effective_configuration(self._dynamic_configuration, configuration)
-                    if dry_run:
-                        return not deep_compare(new_configuration, self.__effective_configuration)
                     self._local_configuration = configuration
                     self.__effective_configuration = new_configuration
                     return True
                 else:
-                    logger.info('No configuration items changed, nothing to reload.')
+                    logger.info('No local configuration items changed.')
             except Exception:
                 logger.exception('Exception when reloading local configuration from %s', self.config_file)
-                if dry_run:
-                    raise
 
     @staticmethod
     def _process_postgresql_parameters(parameters, is_local=False):
@@ -186,18 +229,22 @@ class Config(object):
         config = deepcopy(self.__DEFAULT_CONFIG)
 
         for name, value in dynamic_configuration.items():
+            # allow copying master_start_timeout->primary_start_timeout when the latter isn't in dynamic_configuration
+            if name in self.__REMAP_KEYS and self.__REMAP_KEYS[name] not in dynamic_configuration:
+                name = self.__REMAP_KEYS[name]
             if name == 'postgresql':
                 for name, value in (value or {}).items():
                     if name == 'parameters':
                         config['postgresql'][name].update(self._process_postgresql_parameters(value))
-                    elif name not in ('connect_address', 'listen', 'data_dir', 'pgpass', 'authentication'):
+                    elif name not in ('connect_address', 'proxy_address', 'listen',
+                                      'config_dir', 'data_dir', 'pgpass', 'authentication'):
                         config['postgresql'][name] = deepcopy(value)
             elif name == 'standby_cluster':
                 for name, value in (value or {}).items():
                     if name in self.__DEFAULT_CONFIG['standby_cluster']:
                         config['standby_cluster'][name] = deepcopy(value)
-            elif name in config:  # only variables present in __DEFAULT_CONFIG allowed to be overriden from DCS
+            elif name in config:  # only variables present in __DEFAULT_CONFIG allowed to be overridden from DCS
-                if name in ('synchronous_mode', 'synchronous_mode_strict'):
+                if name in ('synchronous_mode', 'synchronous_mode_strict', 'failsafe_mode'):
                     config[name] = value
                 else:
                     config[name] = int(value)
@@ -208,7 +255,7 @@ class Config(object):
         ret = defaultdict(dict)
 
         def _popenv(name):
-            return os.environ.pop(Config.PATRONI_ENV_PREFIX + name.upper(), None)
+            return os.environ.pop(PATRONI_ENV_PREFIX + name.upper(), None)
 
         for param in ('name', 'namespace', 'scope'):
             value = _popenv(param)
@@ -217,7 +264,7 @@ class Config(object):
 
         def _fix_log_env(name, oldname):
             value = _popenv(oldname)
-            name = Config.PATRONI_ENV_PREFIX + 'LOG_' + name.upper()
+            name = PATRONI_ENV_PREFIX + 'LOG_' + name.upper()
             if value and name not in os.environ:
                 os.environ[name] = value
 
@@ -230,10 +277,45 @@ class Config(object):
                 if value:
                     ret[section][param] = value
 
-        _set_section_values('restapi', ['listen', 'connect_address', 'certfile', 'keyfile'])
+        _set_section_values('restapi', ['listen', 'connect_address', 'certfile', 'keyfile', 'keyfile_password',
-        _set_section_values('postgresql', ['listen', 'connect_address', 'config_dir', 'data_dir', 'pgpass', 'bin_dir'])
+                                        'cafile', 'ciphers', 'verify_client', 'http_extra_headers',
-        _set_section_values('log', ['level', 'format', 'dateformat', 'max_queue_size',
+                                        'https_extra_headers', 'allowlist', 'allowlist_include_members'])
+        _set_section_values('ctl', ['insecure', 'cacert', 'certfile', 'keyfile', 'keyfile_password'])
+        _set_section_values('postgresql', ['listen', 'connect_address', 'proxy_address',
+                                           'config_dir', 'data_dir', 'pgpass', 'bin_dir'])
+        _set_section_values('log', ['level', 'traceback_level', 'format', 'dateformat', 'max_queue_size',
                                     'dir', 'file_size', 'file_num', 'loggers'])
+        _set_section_values('raft', ['data_dir', 'self_addr', 'partner_addrs', 'password', 'bind_addr'])
+
+        for first, second in (('restapi', 'allowlist_include_members'), ('ctl', 'insecure')):
+            value = ret.get(first, {}).pop(second, None)
+            if value:
+                value = parse_bool(value)
+                if value is not None:
+                    ret[first][second] = value
+
+        for second in ('max_queue_size', 'file_size', 'file_num'):
+            value = ret.get('log', {}).pop(second, None)
+            if value:
+                value = parse_int(value)
+                if value is not None:
+                    ret['log'][second] = value
+
+        def _parse_list(value):
+            if not (value.strip().startswith('-') or '[' in value):
+                value = '[{0}]'.format(value)
+            try:
+                return yaml.safe_load(value)
+            except Exception:
+                logger.exception('Exception when parsing list %s', value)
+                return None
+
+        for first, second in (('raft', 'partner_addrs'), ('restapi', 'allowlist')):
+            value = ret.get(first, {}).pop(second, None)
+            if value:
+                value = _parse_list(value)
+                if value:
+                    ret[first][second] = value
 
         def _parse_dict(value):
             if not value.strip().startswith('{'):
@@ -244,15 +326,17 @@ class Config(object):
                 logger.exception('Exception when parsing dict %s', value)
                 return None
 
-        value = ret.get('log', {}).pop('loggers', None)
+        for first, params in (('restapi', ('http_extra_headers', 'https_extra_headers')), ('log', ('loggers',))):
-        if value:
+            for second in params:
-            value = _parse_dict(value)
+                value = ret.get(first, {}).pop(second, None)
-            if value:
+                if value:
-                ret['log']['loggers'] = value
+                    value = _parse_dict(value)
+                    if value:
+                        ret[first][second] = value
 
-        def _get_auth(name):
+        def _get_auth(name, params=None):
             ret = {}
-            for param in ('username', 'password'):
+            for param in params or _AUTH_ALLOWED_PARAMETERS[:2]:
                 value = _popenv(name + '_' + param)
                 if value:
                     ret[param] = value
@@ -264,47 +348,46 @@ class Config(object):
 
         authentication = {}
         for user_type in ('replication', 'superuser', 'rewind'):
-            entry = _get_auth(user_type)
+            entry = _get_auth(user_type, _AUTH_ALLOWED_PARAMETERS)
             if entry:
                 authentication[user_type] = entry
 
         if authentication:
             ret['postgresql']['authentication'] = authentication
 
-        def _parse_list(value):
-            if not (value.strip().startswith('-') or '[' in value):
-                value = '[{0}]'.format(value)
-            try:
-                return yaml.safe_load(value)
-            except Exception:
-                logger.exception('Exception when parsing list %s', value)
-                return None
-
         for param in list(os.environ.keys()):
-            if param.startswith(Config.PATRONI_ENV_PREFIX):
+            if param.startswith(PATRONI_ENV_PREFIX):
                 # PATRONI_(ETCD|CONSUL|ZOOKEEPER|EXHIBITOR|...)_(HOSTS?|PORT|..)
                 name, suffix = (param[8:].split('_', 1) + [''])[:2]
-                if suffix in ('HOST', 'HOSTS', 'PORT', 'USE_PROXIES', 'PROTOCOL', 'SRV', 'URL', 'PROXY',
+                if suffix in ('HOST', 'HOSTS', 'PORT', 'USE_PROXIES', 'PROTOCOL', 'SRV', 'SRV_SUFFIX', 'URL', 'PROXY',
                               'CACERT', 'CERT', 'KEY', 'VERIFY', 'TOKEN', 'CHECKS', 'DC', 'CONSISTENCY',
-                              'REGISTER_SERVICE', 'SERVICE_CHECK_INTERVAL', 'NAMESPACE', 'CONTEXT',
+                              'REGISTER_SERVICE', 'SERVICE_CHECK_INTERVAL', 'SERVICE_CHECK_TLS_SERVER_NAME',
-                              'USE_ENDPOINTS', 'SCOPE_LABEL', 'ROLE_LABEL', 'POD_IP', 'PORTS', 'LABELS') and name:
+                              'SERVICE_TAGS', 'NAMESPACE', 'CONTEXT', 'USE_ENDPOINTS', 'SCOPE_LABEL', 'ROLE_LABEL',
+                              'POD_IP', 'PORTS', 'LABELS', 'BYPASS_API_SERVICE', 'KEY_PASSWORD', 'USE_SSL', 'SET_ACLS',
+                              'GROUP', 'DATABASE') and name:
                     value = os.environ.pop(param)
-                    if suffix == 'PORT':
+                    if name == 'CITUS':
+                        if suffix == 'GROUP':
+                            value = parse_int(value)
+                        elif suffix != 'DATABASE':
+                            continue
+                    elif suffix == 'PORT':
                         value = value and parse_int(value)
-                    elif suffix in ('HOSTS', 'PORTS', 'CHECKS'):
+                    elif suffix in ('HOSTS', 'PORTS', 'CHECKS', 'SERVICE_TAGS'):
                         value = value and _parse_list(value)
-                    elif suffix == 'LABELS':
+                    elif suffix in ('LABELS', 'SET_ACLS'):
                         value = _parse_dict(value)
-                    elif suffix in ('USE_PROXIES', 'REGISTER_SERVICE'):
+                    elif suffix in ('USE_PROXIES', 'REGISTER_SERVICE', 'USE_ENDPOINTS', 'BYPASS_API_SERVICE', 'VERIFY'):
                         value = parse_bool(value)
-                    if value:
+                    if value is not None:
                         ret[name.lower()][suffix.lower()] = value
-        if 'etcd' in ret:
+        for dcs in ('etcd', 'etcd3'):
-            ret['etcd'].update(_get_auth('etcd'))
+            if dcs in ret:
+                ret[dcs].update(_get_auth(dcs))
 
         users = {}
         for param in list(os.environ.keys()):
-            if param.startswith(Config.PATRONI_ENV_PREFIX):
+            if param.startswith(PATRONI_ENV_PREFIX):
                 name, suffix = (param[8:].rsplit('_', 1) + [''])[:2]
                 # PATRONI_<username>_PASSWORD=<password>, PATRONI_<username>_OPTIONS=<option1,option2,...>
                 # CREATE USER "<username>" WITH <OPTIONS> PASSWORD '<password>'
@@ -324,7 +407,11 @@ class Config(object):
     def _build_effective_configuration(self, dynamic_configuration, local_configuration):
         config = self._safe_copy_dynamic_configuration(dynamic_configuration)
         for name, value in local_configuration.items():
-            if name == 'postgresql':
+            if name == 'citus':  # remove invalid citus configuration
+                if isinstance(value, dict) and isinstance(value.get('group'), six.integer_types)\
+                        and isinstance(value.get('database'), six.string_types):
+                    config[name] = value
+            elif name == 'postgresql':
                 for name, value in (value or {}).items():
                     if name == 'parameters':
                         config['postgresql'][name].update(self._process_postgresql_parameters(value, True))
@@ -332,14 +419,9 @@ class Config(object):
                         config['postgresql'][name] = deepcopy(value)
             elif name not in config or name in ['watchdog']:
                 config[name] = deepcopy(value) if value else {}
-        
-        if os.getenv("PATRONI_DEBUG_MODE"):
-            config['ttl'] = 24 * 60 * 60 # practical infinity for a debugging session
-            config['loop_wait'] = -1
-
 
         # restapi server expects to get restapi.auth = 'username:password'
-        if 'authentication' in config['restapi']:
+        if 'restapi' in config and 'authentication' in config['restapi']:
             config['restapi']['auth'] = '{username}:{password}'.format(**config['restapi']['authentication'])
 
         # special treatment for old config
@@ -358,16 +440,30 @@ class Config(object):
         if 'superuser' not in pg_config['authentication'] and 'pg_rewind' in pg_config:
             pg_config['authentication']['superuser'] = pg_config['pg_rewind']
 
+        # handle setting additional connection parameters that may be available
+        # in the configuration file, such as SSL connection parameters
+        for name, value in pg_config['authentication'].items():
+            pg_config['authentication'][name] = {n: v for n, v in value.items() if n in _AUTH_ALLOWED_PARAMETERS}
+
         # no 'name' in config
         if 'name' not in config and 'name' in pg_config:
             config['name'] = pg_config['name']
 
+        # when bootstrapping the new Citus cluster (coordinator/worker) enable sync replication in global configuration
+        if 'citus' in config:
+            bootstrap = config.setdefault('bootstrap', {})
+            dcs = bootstrap.setdefault('dcs', {})
+            dcs.setdefault('synchronous_mode', True)
+
         updated_fields = (
             'name',
             'scope',
             'retry_timeout',
             'synchronous_mode',
             'synchronous_mode_strict',
+            'synchronous_node_count',
+            'maximum_lag_on_syncnode',
+            'citus'
         )
 
         pg_config.update({p: config[p] for p in updated_fields if p in config})

patroni/daemon.py

@@ -0,0 +1,112 @@
+from __future__ import print_function
+
+import abc
+import os
+import signal
+import six
+import sys
+
+from threading import Lock
+
+
+@six.add_metaclass(abc.ABCMeta)
+class AbstractPatroniDaemon(object):
+
+    def __init__(self, config):
+        from patroni.log import PatroniLogger
+
+        self.setup_signal_handlers()
+
+        self.logger = PatroniLogger()
+        self.config = config
+        AbstractPatroniDaemon.reload_config(self, local=True)
+
+    def sighup_handler(self, *args):
+        self._received_sighup = True
+
+    def api_sigterm(self):
+        with self._sigterm_lock:
+            if not self._received_sigterm:
+                self._received_sigterm = True
+                return True
+
+    def sigterm_handler(self, *args):
+        if self.api_sigterm():
+            sys.exit()
+
+    def setup_signal_handlers(self):
+        self._received_sighup = False
+        self._sigterm_lock = Lock()
+        self._received_sigterm = False
+        if os.name != 'nt':
+            signal.signal(signal.SIGHUP, self.sighup_handler)
+        signal.signal(signal.SIGTERM, self.sigterm_handler)
+
+    @property
+    def received_sigterm(self):
+        with self._sigterm_lock:
+            return self._received_sigterm
+
+    def reload_config(self, sighup=False, local=False):
+        if local:
+            self.logger.reload_config(self.config.get('log', {}))
+
+    @abc.abstractmethod
+    def _run_cycle(self):
+        """_run_cycle"""
+
+    def run(self):
+        self.logger.start()
+        while not self.received_sigterm:
+            if self._received_sighup:
+                self._received_sighup = False
+                self.reload_config(True, self.config.reload_local_configuration())
+
+            self._run_cycle()
+
+    @abc.abstractmethod
+    def _shutdown(self):
+        """_shutdown"""
+
+    def shutdown(self):
+        with self._sigterm_lock:
+            self._received_sigterm = True
+        self._shutdown()
+        self.logger.shutdown()
+
+
+def abstract_main(cls, validator=None):
+    import argparse
+
+    from .config import Config, ConfigParseError
+    from .version import __version__
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--version', action='version', version='%(prog)s {0}'.format(__version__))
+    if validator:
+        parser.add_argument('--validate-config', action='store_true', help='Run config validator and exit')
+    parser.add_argument('configfile', nargs='?', default='',
+                        help='Patroni may also read the configuration from the {0} environment variable'
+                        .format(Config.PATRONI_CONFIG_VARIABLE))
+    args = parser.parse_args()
+    validate_config = validator and args.validate_config
+    try:
+        if validate_config:
+            Config(args.configfile, validator=validator)
+            sys.exit()
+
+        config = Config(args.configfile)
+    except ConfigParseError as e:
+        if e.value:
+            print(e.value, file=sys.stderr)
+        if not validate_config:
+            parser.print_help()
+        sys.exit(1)
+
+    controller = cls(config)
+    try:
+        controller.run()
+    except KeyboardInterrupt:
+        pass
+    finally:
+        controller.shutdown()

patroni/dcs/__init__.py

@@ -1,5 +1,5 @@
 import abc
-import dateutil
+import dateutil.parser
 import importlib
 import inspect
 import json
@@ -13,12 +13,15 @@ import time
 
 from collections import defaultdict, namedtuple
 from copy import deepcopy
-from patroni.exceptions import PatroniException
-from patroni.utils import parse_bool, uri
 from random import randint
 from six.moves.urllib_parse import urlparse, urlunparse, parse_qsl
 from threading import Event, Lock
 
+from ..exceptions import PatroniFatalException
+from ..utils import deep_compare, parse_bool, uri
+
+CITUS_COORDINATOR_GROUP_ID = 0
+citus_group_re = re.compile('^(0|[1-9][0-9]*)$')
 slot_name_re = re.compile('^[a-z0-9_]{1,63}$')
 logger = logging.getLogger(__name__)
 
@@ -66,33 +69,51 @@ def dcs_modules():
     module_prefix = __package__ + '.'
 
     if getattr(sys, 'frozen', False):
-        importer = pkgutil.get_importer(dcs_dirname)
+        toc = set()
-        return [module for module in list(importer.toc) if module.startswith(module_prefix) and module.count('.') == 2]
+        # dcs_dirname may contain a dot, which causes pkgutil.iter_importers()
+        # to misinterpret the path as a package name. This can be avoided
+        # altogether by not passing a path at all, because PyInstaller's
+        # FrozenImporter is a singleton and registered as top-level finder.
+        for importer in pkgutil.iter_importers():
+            if hasattr(importer, 'toc'):
+                toc |= importer.toc
+        return [module for module in toc if module.startswith(module_prefix) and module.count('.') == 2]
     else:
         return [module_prefix + name for _, name, is_pkg in pkgutil.iter_modules([dcs_dirname]) if not is_pkg]
 
 
 def get_dcs(config):
-    available_implementations = set()
+    modules = dcs_modules()
-    for module_name in dcs_modules():
+
-        try:
+    for module_name in modules:
-            module = importlib.import_module(module_name)
+        name = module_name.split('.')[-1]
-            for name in filter(lambda name: not name.startswith('__'), dir(module)):  # iterate through module content
+        if name in config:  # we will try to import only modules which have configuration section in the config file
-                item = getattr(module, name)
+            try:
-                name = name.lower()
+                module = importlib.import_module(module_name)
-                # try to find implementation of AbstractDCS interface, class name must match with module_name
+                for key, item in module.__dict__.items():  # iterate through the module content
-                if inspect.isclass(item) and issubclass(item, AbstractDCS) and __package__ + '.' + name == module_name:
+                    # try to find implementation of AbstractDCS interface, class name must match with module_name
-                    available_implementations.add(name)
+                    if key.lower() == name and inspect.isclass(item) and issubclass(item, AbstractDCS):
-                    if name in config:  # which has configuration section in the config file
                         # propagate some parameters
                         config[name].update({p: config[p] for p in ('namespace', 'name', 'scope', 'loop_wait',
                                              'patronictl', 'ttl', 'retry_timeout') if p in config})
+                        # From citus section we only need "group" parameter, but will propagate everything just in case.
+                        if isinstance(config.get('citus'), dict):
+                            config[name].update(config['citus'])
                         return item(config[name])
+            except ImportError:
+                logger.debug('Failed to import %s', module_name)
+
+    available_implementations = []
+    for module_name in modules:
+        name = module_name.split('.')[-1]
+        try:
+            module = importlib.import_module(module_name)
+            available_implementations.extend(name for key, item in module.__dict__.items() if key.lower() == name
+                                             and inspect.isclass(item) and issubclass(item, AbstractDCS))
         except ImportError:
-            if not config.get('patronictl'):
+            logger.info('Failed to import %s', module_name)
-                logger.info('Failed to import %s', module_name)
+    raise PatroniFatalException("""Can not find suitable configuration of distributed configuration store
-    raise PatroniException("""Can not find suitable configuration of distributed configuration store
+Available implementations: """ + ', '.join(sorted(set(available_implementations))))
-Available implementations: """ + ', '.join(available_implementations))
 
 
 class Member(namedtuple('Member', 'index,name,session,data')):
@@ -122,6 +143,8 @@ class Member(namedtuple('Member', 'index,name,session,data')):
         else:
             try:
                 data = json.loads(data)
+                if not isinstance(data, dict):
+                    data = {}
             except (TypeError, ValueError):
                 data = {}
         return Member(index, name, session, data)
@@ -129,10 +152,10 @@ class Member(namedtuple('Member', 'index,name,session,data')):
     @property
     def conn_url(self):
         conn_url = self.data.get('conn_url')
-        conn_kwargs = self.data.get('conn_kwargs')
         if conn_url:
             return conn_url
 
+        conn_kwargs = self.data.get('conn_kwargs')
         if conn_kwargs:
             conn_url = uri('postgresql', (conn_kwargs.get('host'), conn_kwargs.get('port', 5432)))
             self.data['conn_url'] = conn_url
@@ -140,28 +163,31 @@ class Member(namedtuple('Member', 'index,name,session,data')):
 
     def conn_kwargs(self, auth=None):
         defaults = {
-            "host": "",
+            "host": None,
-            "port": "",
+            "port": None,
-            "database": ""
+            "dbname": None
         }
         ret = self.data.get('conn_kwargs')
         if ret:
             defaults.update(ret)
             ret = defaults
         else:
-            r = urlparse(self.conn_url)
+            conn_url = self.conn_url
+            if not conn_url:
+                return {}  # due to the invalid conn_url we don't care about authentication parameters
+            r = urlparse(conn_url)
             ret = {
                 'host': r.hostname,
                 'port': r.port or 5432,
-                'database': r.path[1:]
+                'dbname': r.path[1:]
             }
             self.data['conn_kwargs'] = ret.copy()
 
+        # apply any remaining authentication parameters
         if auth and isinstance(auth, dict):
+            ret.update({k: v for k, v in auth.items() if v is not None})
             if 'username' in auth:
-                ret['user'] = auth['username']
+                ret['user'] = ret.pop('username')
-            if 'password' in auth:
-                ret['password'] = auth['password']
         return ret
 
     @property
@@ -192,10 +218,18 @@ class Member(namedtuple('Member', 'index,name,session,data')):
     def is_running(self):
         return self.state == 'running'
 
+    @property
+    def version(self):
+        version = self.data.get('version')
+        if version:
+            try:
+                return tuple(map(int, version.split('.')))
+            except Exception:
+                logger.debug('Failed to parse Patroni version %s', version)
+
 
 class RemoteMember(Member):
-    """ Represents a remote master for a standby cluster
+    """Represents a remote member (typically a primary) for a standby cluster"""
-    """
     def __new__(cls, name, data):
         return super(RemoteMember, cls).__new__(cls, None, name, None, data)
 
@@ -232,23 +266,23 @@ class Leader(namedtuple('Leader', 'index,session,member')):
     def conn_url(self):
         return self.member.conn_url
 
+    @property
+    def data(self):
+        return self.member.data
+
     @property
     def timeline(self):
-        return self.member.data.get('timeline')
+        return self.data.get('timeline')
 
     @property
     def checkpoint_after_promote(self):
         """
         >>> Leader(1, '', Member.from_node(1, '', '', '{"version":"z"}')).checkpoint_after_promote
         """
-        version = self.member.data.get('version')
+        version = self.member.version
-        if version:
+        # 1.5.6 is the last version which doesn't expose checkpoint_after_promote: false
-            try:
+        if version and version > (1, 5, 6):
-                # 1.5.6 is the last version which doesn't expose checkpoint_after_promote: false
+            return self.data.get('role') in ('master', 'primary') and 'checkpoint_after_promote' not in self.data
-                if tuple(map(int, version.split('.'))) > (1, 5, 6):
-                    return self.member.data['role'] == 'master' and 'checkpoint_after_promote' not in self.member.data
-            except Exception:
-                logger.debug('Failed to parse Patroni version %s', version)
 
 
 class Failover(namedtuple('Failover', 'index,leader,candidate,scheduled_at')):
@@ -323,13 +357,21 @@ class ClusterConfig(namedtuple('ClusterConfig', 'index,data,modify_index')):
                 self.data.get('permanent_slots') or self.data.get('slots')
         ) or {}
 
+    @property
+    def ignore_slots_matchers(self):
+        return isinstance(self.data, dict) and self.data.get('ignore_slots') or []
+
+    @property
+    def max_timelines_history(self):
+        return self.data.get('max_timelines_history', 0)
+
 
 class SyncState(namedtuple('SyncState', 'index,leader,sync_standby')):
     """Immutable object (namedtuple) which represents last observed synhcronous replication state
 
     :param index: modification index of a synchronization key in a Configuration Store
     :param leader: reference to member that was leader
-    :param sync_standby: standby that was last synchronized to leader
+    :param sync_standby: synchronous standby list (comma delimited) which are last synchronized to leader
     """
 
     @staticmethod
@@ -361,15 +403,22 @@ class SyncState(namedtuple('SyncState', 'index,leader,sync_standby')):
             data = {}
         return SyncState(index, data.get('leader'), data.get('sync_standby'))
 
+    @property
+    def members(self):
+        """ Returns sync_standby in list """
+        return self.sync_standby and self.sync_standby.split(',') or []
+
     def matches(self, name):
         """
         Returns if a node name matches one of the nodes in the sync state
 
-        >>> s = SyncState(1, 'foo', 'bar')
+        >>> s = SyncState(1, 'foo', 'bar,zoo')
         >>> s.matches('foo')
         True
         >>> s.matches('bar')
         True
+        >>> s.matches('zoo')
+        True
         >>> s.matches('baz')
         False
         >>> s.matches(None)
@@ -377,7 +426,7 @@ class SyncState(namedtuple('SyncState', 'index,leader,sync_standby')):
         >>> SyncState(1, None, None).matches('foo')
         False
         """
-        return name is not None and name in (self.leader, self.sync_standby)
+        return name is not None and name in [self.leader] + self.members
 
 
 class TimelineHistory(namedtuple('TimelineHistory', 'index,value,lines')):
@@ -399,23 +448,36 @@ class TimelineHistory(namedtuple('TimelineHistory', 'index,value,lines')):
         return TimelineHistory(index, value, lines)
 
 
-class Cluster(namedtuple('Cluster', 'initialize,config,leader,last_leader_operation,members,failover,sync,history')):
+class Cluster(namedtuple('Cluster', 'initialize,config,leader,last_lsn,members,'
+                                    'failover,sync,history,slots,failsafe,workers')):
 
     """Immutable object (namedtuple) which represents PostgreSQL cluster.
     Consists of the following fields:
     :param initialize: shows whether this cluster has initialization key stored in DC or not.
     :param config: global dynamic configuration, reference to `ClusterConfig` object
     :param leader: `Leader` object which represents current leader of the cluster
-    :param last_leader_operation: int or long object containing position of last known leader operation.
+    :param last_lsn: int or long object containing position of last known leader LSN.
-        This value is stored in `/optime/leader` key
+        This value is stored in the `/status` key or `/optime/leader` (legacy) key
     :param members: list of Member object, all PostgreSQL cluster members including leader
     :param failover: reference to `Failover` object
     :param sync: reference to `SyncState` object, last observed synchronous replication state.
     :param history: reference to `TimelineHistory` object
-    """
+    :param slots: state of permanent logical replication slots on the primary in the format: {"slot_name": int}
+    :param failsafe: failsafe topology. Node is allowed to become the leader only if its name is found in this list.
+    :param workers: workers of the Citus cluster, optional. Format: {int(group): Cluster()}"""
+
+    def __new__(cls, *args):
+        # Make workers argument optional
+        if len(cls._fields) == len(args) + 1:
+            args = args + ({},)
+        return super(Cluster, cls).__new__(cls, *args)
+
+    @property
+    def leader_name(self):
+        return self.leader and self.leader.name
 
     def is_unlocked(self):
-        return not (self.leader and self.leader.name)
+        return not self.leader_name
 
     def has_member(self, member_name):
         return any(m for m in self.members if m.name == member_name)
@@ -437,20 +499,41 @@ class Cluster(namedtuple('Cluster', 'initialize,config,leader,last_leader_operat
     def is_synchronous_mode(self):
         return self.check_mode('synchronous_mode')
 
-    def get_replication_slots(self, name, role):
+    @property
+    def __permanent_slots(self):
+        return self.config and self.config.permanent_slots or {}
+
+    @property
+    def __permanent_physical_slots(self):
+        return {name: value for name, value in self.__permanent_slots.items()
+                if not value or isinstance(value, dict) and value.get('type', 'physical') == 'physical'}
+
+    @property
+    def __permanent_logical_slots(self):
+        return {name: value for name, value in self.__permanent_slots.items() if isinstance(value, dict)
+                and value.get('type', 'logical') == 'logical' and value.get('database') and value.get('plugin')}
+
+    @property
+    def use_slots(self):
+        return self.config and (self.config.data.get('postgresql') or {}).get('use_slots', True)
+
+    def get_replication_slots(self, my_name, role, nofailover, major_version, show_error=False):
         # if the replicatefrom tag is set on the member - we should not create the replication slot for it on
-        # the current master, because that member would replicate from elsewhere. We still create the slot if
+        # the current primary, because that member would replicate from elsewhere. We still create the slot if
         # the replicatefrom destination member is currently not a member of the cluster (fallback to the
-        # master), or if replicatefrom destination member happens to be the current master
+        # primary), or if replicatefrom destination member happens to be the current primary
-        if role in ('master', 'standby_leader'):
+        use_slots = self.use_slots
-            slot_members = [m.name for m in self.members if m.name != name and
+        if role in ('master', 'primary', 'standby_leader'):
-                            (m.replicatefrom is None or m.replicatefrom == name or
+            slot_members = [m.name for m in self.members if use_slots and m.name != my_name and
+                            (m.replicatefrom is None or m.replicatefrom == my_name or
                              not self.has_member(m.replicatefrom))]
-            permanent_slots = (self.config and self.config.permanent_slots or {}).copy()
+            permanent_slots = self.__permanent_slots if use_slots and \
+                role in ('master', 'primary') else self.__permanent_physical_slots
         else:
             # only manage slots for replicas that replicate from this one, except for the leader among them
-            slot_members = [m.name for m in self.members if m.replicatefrom == name and m.name != self.leader.name]
+            slot_members = [m.name for m in self.members if use_slots and
-            permanent_slots = {}
+                            m.replicatefrom == my_name and m.name != self.leader_name]
+            permanent_slots = self.__permanent_logical_slots if use_slots and not nofailover else {}
 
         slots = {slot_name_from_member_name(name): {'type': 'physical'} for name in slot_members}
 
@@ -464,46 +547,82 @@ class Cluster(namedtuple('Cluster', 'initialize,config,leader,last_leader_operat
                                    for k, v in slot_conflicts.items() if len(v) > 1))
 
         # "merge" replication slots for members with permanent_replication_slots
+        disabled_permanent_logical_slots = []
         for name, value in permanent_slots.items():
             if not slot_name_re.match(name):
                 logger.error("Invalid permanent replication slot name '%s'", name)
                 logger.error("Slot name may only contain lower case letters, numbers, and the underscore chars")
                 continue
 
-            if name in slots:
+            value = deepcopy(value) if value else {'type': 'physical'}
-                logger.error("Permanent replication slot {'%s': %s} is conflicting with" +
-                             " physical replication slot for cluster member", name, value)
-                continue
-
-            value = deepcopy(value)
-            if not value:
-                value = {'type': 'physical'}
-
             if isinstance(value, dict):
                 if 'type' not in value:
                     value['type'] = 'logical' if value.get('database') and value.get('plugin') else 'physical'
 
-                if value['type'] == 'physical' or value['type'] == 'logical' \
+                if value['type'] == 'physical':
-                        and value.get('database') and value.get('plugin'):
+                    # Don't try to create permanent physical replication slot for yourself
-                    slots[name] = value
+                    if name != slot_name_from_member_name(my_name):
+                        slots[name] = value
+                    continue
+                elif value['type'] == 'logical' and value.get('database') and value.get('plugin'):
+                    if major_version < 110000:
+                        disabled_permanent_logical_slots.append(name)
+                    elif name in slots:
+                        logger.error("Permanent logical replication slot {'%s': %s} is conflicting with" +
+                                     " physical replication slot for cluster member", name, value)
+                    else:
+                        slots[name] = value
                     continue
 
             logger.error("Bad value for slot '%s' in permanent_slots: %s", name, permanent_slots[name])
 
+        if disabled_permanent_logical_slots and show_error:
+            logger.error("Permanent logical replication slots supported by Patroni only starting from PostgreSQL 11. "
+                         "Following slots will not be created: %s.", disabled_permanent_logical_slots)
+
         return slots
 
-    def has_permanent_logical_slots(self, name):
+    def has_permanent_logical_slots(self, my_name, nofailover, major_version=110000):
-        slots = self.get_replication_slots(name, 'master').values()
+        if major_version < 110000:
+            return False
+        slots = self.get_replication_slots(my_name, 'replica', nofailover, major_version).values()
         return any(v for v in slots if v.get("type") == "logical")
 
+    def should_enforce_hot_standby_feedback(self, my_name, nofailover, major_version):
+        """
+        The hot_standby_feedback must be enabled if the current replica has logical slots
+        or it is working as a cascading replica for the other node that has logical slots.
+        """
+
+        if major_version < 110000:
+            return False
+
+        if self.has_permanent_logical_slots(my_name, nofailover, major_version):
+            return True
+
+        if self.use_slots:
+            members = [m for m in self.members if m.replicatefrom == my_name and m.name != self.leader_name]
+            return any(self.should_enforce_hot_standby_feedback(m.name, m.nofailover, major_version) for m in members)
+        return False
+
+    def get_my_slot_name_on_primary(self, my_name, replicatefrom):
+        """
+        P <-- I <-- L
+        In case of cascading replication we have to check not our physical slot,
+        but slot of the replica that connects us to the primary.
+        """
+
+        m = self.get_member(replicatefrom, False) if replicatefrom else None
+        return self.get_my_slot_name_on_primary(m.name, m.replicatefrom) if m else slot_name_from_member_name(my_name)
+
     @property
     def timeline(self):
         """
-        >>> Cluster(0, 0, 0, 0, 0, 0, 0, 0).timeline
+        >>> Cluster(0, 0, 0, 0, 0, 0, 0, 0, 0, None).timeline
         0
-        >>> Cluster(0, 0, 0, 0, 0, 0, 0, TimelineHistory.from_node(1, '[]')).timeline
+        >>> Cluster(0, 0, 0, 0, 0, 0, 0, TimelineHistory.from_node(1, '[]'), 0, None).timeline
         1
-        >>> Cluster(0, 0, 0, 0, 0, 0, 0, TimelineHistory.from_node(1, '[["a"]]')).timeline
+        >>> Cluster(0, 0, 0, 0, 0, 0, 0, TimelineHistory.from_node(1, '[["a"]]'), 0, None).timeline
         0
         """
         if self.history:
@@ -516,6 +635,24 @@ class Cluster(namedtuple('Cluster', 'initialize,config,leader,last_leader_operat
                 return 1
         return 0
 
+    @property
+    def min_version(self):
+        return next(iter(sorted(filter(lambda v: v, [m.version for m in self.members])) + [None]))
+
+
+class ReturnFalseException(Exception):
+    pass
+
+
+def catch_return_false_exception(func):
+    def wrapper(*args, **kwargs):
+        try:
+            return func(*args, **kwargs)
+        except ReturnFalseException:
+            return False
+
+    return wrapper
+
 
 @six.add_metaclass(abc.ABCMeta)
 class AbstractDCS(object):
@@ -527,8 +664,10 @@ class AbstractDCS(object):
     _HISTORY = 'history'
     _MEMBERS = 'members/'
     _OPTIME = 'optime'
-    _LEADER_OPTIME = _OPTIME + '/' + _LEADER
+    _STATUS = 'status'  # JSON, contains "leader_lsn" and confirmed_flush_lsn of logical "slots" on the leader
+    _LEADER_OPTIME = _OPTIME + '/' + _LEADER  # legacy
     _SYNC = 'sync'
+    _FAILSAFE = 'failsafe'
 
     def __init__(self, config):
         """
@@ -537,17 +676,25 @@ class AbstractDCS(object):
         """
         self._name = config['name']
         self._base_path = re.sub('/+', '/', '/'.join(['', config.get('namespace', 'service'), config['scope']]))
+        self._citus_group = str(config['group']) if isinstance(config.get('group'), six.integer_types) else None
         self._set_loop_wait(config.get('loop_wait', 10))
 
         self._ctl = bool(config.get('patronictl', False))
         self._cluster = None
         self._cluster_valid_till = 0
         self._cluster_thread_lock = Lock()
-        self._last_leader_operation = ''
+        self._last_lsn = ''
+        self._last_seen = 0
+        self._last_status = {}
+        self._last_failsafe = {}
         self.event = Event()
 
     def client_path(self, path):
-        return '/'.join([self._base_path, path.lstrip('/')])
+        components = [self._base_path]
+        if self._citus_group:
+            components.append(self._citus_group)
+        components.append(path.lstrip('/'))
+        return '/'.join(components)
 
     @property
     def initialize_path(self):
@@ -577,6 +724,10 @@ class AbstractDCS(object):
     def history_path(self):
         return self.client_path(self._HISTORY)
 
+    @property
+    def status_path(self):
+        return self.client_path(self._STATUS)
+
     @property
     def leader_optime_path(self):
         return self.client_path(self._LEADER_OPTIME)
@@ -585,6 +736,10 @@ class AbstractDCS(object):
     def sync_path(self):
         return self.client_path(self._SYNC)
 
+    @property
+    def failsafe_path(self):
+        return self.client_path(self._FAILSAFE)
+
     @abc.abstractmethod
     def set_ttl(self, ttl):
         """Set the new ttl value for leader key"""
@@ -609,23 +764,75 @@ class AbstractDCS(object):
     def loop_wait(self):
         return self._loop_wait
 
+    @property
+    def last_seen(self):
+        return self._last_seen
+
     @abc.abstractmethod
-    def _load_cluster(self):
+    def _cluster_loader(self, path):
-        """Internally this method should build  `Cluster` object which
+        """Load and build the `Cluster` object from DCS, which
-           represents current state and topology of the cluster in DCS.
+        represents a single Patroni cluster.
-           this method supposed to be called only by `get_cluster` method.
 
-           raise `~DCSError` in case of communication or other problems with DCS.
+        :param path: the path in DCS where to load Cluster(s) from.
-           If the current node was running as a master and exception raised,
+        :returns: `Cluster`"""
-           instance would be demoted."""
 
-    def get_cluster(self):
+    def _citus_cluster_loader(self, path):
+        """Load and build `Cluster` onjects from DCS that represent all
+        Patroni clusters from a single Citus cluster.
+
+        :param path: the path in DCS where to load Cluster(s) from.
+        :returns: all Citus groups as `dict`, with group ids as keys"""
+
+    @abc.abstractmethod
+    def _load_cluster(self, path, loader):
+        """Internally this method should call the `loader` method that
+        will build `Cluster` object which represents current state and
+        topology of the cluster in DCS. This method supposed to be
+        called only by `get_cluster` method.
+
+        :param path: the path in DCS where to load Cluster(s) from.
+        :param loader: one of `_cluster_loader` or `_citus_cluster_loader`
+        :raise: `~DCSError` in case of communication problems with DCS.
+        If the current node was running as a primary and exception
+        raised, instance would be demoted."""
+
+    def _bypass_caches(self):
+        """Used only in zookeeper"""
+
+    def is_citus_coordinator(self):
+        return self._citus_group == str(CITUS_COORDINATOR_GROUP_ID)
+
+    def get_citus_coordinator(self):
         try:
-            cluster = self._load_cluster()
+            path = '{0}/{1}/'.format(self._base_path, CITUS_COORDINATOR_GROUP_ID)
+            return self._load_cluster(path, self._cluster_loader)
+        except Exception as e:
+            logger.error('Failed to load Citus coordinator cluster from %s: %r', self.__class__.__name__, e)
+
+    def _get_citus_cluster(self):
+        groups = self._load_cluster(self._base_path + '/', self._citus_cluster_loader)
+        if isinstance(groups, Cluster):  # Zookeeper could return a cached version
+            cluster = groups
+        else:
+            cluster = groups.pop(CITUS_COORDINATOR_GROUP_ID,
+                                 Cluster(None, None, None, None, [], None, None, None, None, None))
+            cluster.workers.update(groups)
+        return cluster
+
+    def get_cluster(self, force=False):
+        if force:
+            self._bypass_caches()
+        try:
+            cluster = self._get_citus_cluster() if self.is_citus_coordinator()\
+                else self._load_cluster(self.client_path(''), self._cluster_loader)
         except Exception:
             self.reset_cluster()
             raise
 
+        self._last_seen = int(time.time())
+        self._last_status = {self._OPTIME: cluster.last_lsn, 'slots': cluster.slots}
+        self._last_failsafe = cluster.failsafe
+
         with self._cluster_thread_lock:
             self._cluster = cluster
             self._cluster_valid_till = time.time() + self.ttl
@@ -642,47 +849,88 @@ class AbstractDCS(object):
             self._cluster_valid_till = 0
 
     @abc.abstractmethod
-    def _write_leader_optime(self, last_operation):
+    def _write_leader_optime(self, last_lsn):
-        """write current xlog location into `/optime/leader` key in DCS
+        """write current WAL LSN into `/optime/leader` key in DCS
-        :param last_operation: absolute xlog location in bytes
+
+        :param last_lsn: absolute WAL LSN in bytes
         :returns: `!True` on success."""
 
-    def write_leader_optime(self, last_operation):
+    def write_leader_optime(self, last_lsn):
-        if self._last_leader_operation != last_operation and self._write_leader_optime(last_operation):
+        self.write_status({self._OPTIME: last_lsn})
-            self._last_leader_operation = last_operation
+
+    @abc.abstractmethod
+    def _write_status(self, value):
+        """write current WAL LSN and confirmed_flush_lsn of permanent slots into the `/status` key in DCS
+
+        :param value: status serialized in JSON forman
+        :returns: `!True` on success."""
+
+    def write_status(self, value):
+        if not deep_compare(self._last_status, value) and self._write_status(json.dumps(value, separators=(',', ':'))):
+            self._last_status = value
+        cluster = self.cluster
+        min_version = cluster and cluster.min_version
+        if min_version and min_version < (2, 1, 0) and self._last_lsn != value[self._OPTIME]:
+            self._last_lsn = value[self._OPTIME]
+            self._write_leader_optime(str(value[self._OPTIME]))
+
+    @abc.abstractmethod
+    def _write_failsafe(self, value):
+        """Write current cluster topology to DCS that will be used by failsafe mechanism (if enabled).
+
+        :param value: failsafe topology serialized in JSON format
+        :returns: `!True` on success."""
+
+    def write_failsafe(self, value):
+        if not (isinstance(self._last_failsafe, dict) and deep_compare(self._last_failsafe, value))\
+                and self._write_failsafe(json.dumps(value, separators=(',', ':'))):
+            self._last_failsafe = value
+
+    @property
+    def failsafe(self):
+        return self._last_failsafe
 
     @abc.abstractmethod
     def _update_leader(self):
         """Update leader key (or session) ttl
 
         :returns: `!True` if leader key (or session) has been updated successfully.
-            If not, `!False` must be returned and current instance would be demoted.
 
         You have to use CAS (Compare And Swap) operation in order to update leader key,
-        for example for etcd `prevValue` parameter must be used."""
+        for example for etcd `prevValue` parameter must be used.
+        If update fails due to DCS not being accessible or because it is not able to
+        process requests (hopefuly temporary), the ~DCSError exception should be raised."""
 
-    def update_leader(self, last_operation, access_is_restricted=False):
+    def update_leader(self, last_lsn, slots=None, failsafe=None):
         """Update leader key (or session) ttl and optime/leader
 
-        :param last_operation: absolute xlog location in bytes
+        :param last_lsn: absolute WAL LSN in bytes
-        :returns: `!True` if leader key (or session) has been updated successfully.
+        :param slots: dict with permanent slots confirmed_flush_lsn
-            If not, `!False` must be returned and current instance would be demoted."""
+        :returns: `!True` if leader key (or session) has been updated successfully."""
 
         ret = self._update_leader()
-        if ret and last_operation:
+        if ret and last_lsn:
-            self.write_leader_optime(last_operation)
+            status = {self._OPTIME: last_lsn}
+            if slots:
+                status['slots'] = slots
+            self.write_status(status)
+
+        if ret and failsafe is not None:
+            self.write_failsafe(failsafe)
+
         return ret
 
     @abc.abstractmethod
-    def attempt_to_acquire_leader(self, permanent=False):
+    def attempt_to_acquire_leader(self):
         """Attempt to acquire leader lock
         This method should create `/leader` key with value=`~self._name`
-        :param permanent: if set to `!True`, the leader key will never expire.
-         Used in patronictl for the external master
         :returns: `!True` if key has been created successfully.
 
         Key must be created atomically. In case if key already exists it should not be
-        overwritten and `!False` must be returned"""
+        overwritten and `!False` must be returned.
+
+        If key creation fails due to DCS not being accessible or because it is not able to
+        process requests (hopefuly temporary), the ~DCSError exception should be raised"""
 
     @abc.abstractmethod
     def set_failover_value(self, value, index=None):
@@ -705,15 +953,13 @@ class AbstractDCS(object):
         """Create or update `/config` key"""
 
     @abc.abstractmethod
-    def touch_member(self, data, permanent=False):
+    def touch_member(self, data):
         """Update member key in DCS.
         This method should create or update key with the name = '/members/' + `~self._name`
         and value = data in a given DCS.
 
         :param data: information about instance (including connection strings)
         :param ttl: ttl for member key, optional parameter. If it is None `~self.member_ttl will be used`
-        :param permanent: if set to `!True`, the member key will never expire.
-         Used in patronictl for the external master.
         :returns: `!True` on success otherwise `!False`
         """
 
@@ -735,10 +981,19 @@ class AbstractDCS(object):
         otherwise it should return `!False`"""
 
     @abc.abstractmethod
-    def delete_leader(self):
+    def _delete_leader(self):
-        """Voluntarily remove leader key from DCS
+        """Remove leader key from DCS.
         This method should remove leader key if current instance is the leader"""
 
+    def delete_leader(self, last_lsn=None):
+        """Update optime/leader and voluntarily remove leader key from DCS.
+        This method should remove leader key if current instance is the leader.
+        :param last_lsn: latest checkpoint location in bytes"""
+
+        if last_lsn:
+            self.write_status({self._OPTIME: last_lsn})
+        return self._delete_leader()
+
     @abc.abstractmethod
     def cancel_initialization(self):
         """ Removes the initialize key for a cluster """
@@ -749,8 +1004,10 @@ class AbstractDCS(object):
 
     @staticmethod
     def sync_state(leader, sync_standby):
-        """Build sync_state dict"""
+        """Build sync_state dict
-        return {'leader': leader, 'sync_standby': sync_standby}
+           sync_standby dictionary key being kept for backward compatibility
+        """
+        return {'leader': leader, 'sync_standby': sync_standby and ','.join(sorted(sync_standby)) or None}
 
     def write_sync_state(self, leader, sync_standby, index=None):
         sync_value = self.sync_state(leader, sync_standby)
@@ -769,7 +1026,7 @@ class AbstractDCS(object):
         """"""
 
     def watch(self, leader_index, timeout):
-        """If the current node is a master it should just sleep.
+        """If the current node is a leader it should just sleep.
         Any other node should watch for changes of leader key with a given timeout
 
         :param leader_index: index of a leader key
@@ -777,4 +1034,4 @@ class AbstractDCS(object):
         :returns: `!True` if you would like to reschedule the next run of ha cycle"""
 
         self.event.wait(timeout)
-        return self.event.isSet()
+        return self.event.is_set()

patroni/dcs/consul.py

@@ -8,14 +8,17 @@ import ssl
 import time
 import urllib3
 
+from collections import defaultdict, namedtuple
 from consul import ConsulException, NotFound, base
-from patroni.dcs import AbstractDCS, ClusterConfig, Cluster, Failover, Leader, Member, SyncState, TimelineHistory
-from patroni.exceptions import DCSError
-from patroni.utils import deep_compare, parse_bool, Retry, RetryFailedError, split_host_port, uri
 from urllib3.exceptions import HTTPError
 from six.moves.urllib.parse import urlencode, urlparse, quote
 from six.moves.http_client import HTTPException
 
+from . import AbstractDCS, Cluster, ClusterConfig, Failover, Leader, Member, SyncState,\
+        TimelineHistory, ReturnFalseException, catch_return_false_exception, citus_group_re
+from ..exceptions import DCSError
+from ..utils import deep_compare, parse_bool, Retry, RetryFailedError, split_host_port, uri, USER_AGENT
+
 logger = logging.getLogger(__name__)
 
 
@@ -35,6 +38,9 @@ class InvalidSession(ConsulException):
     """invalid session"""
 
 
+Response = namedtuple('Response', 'code,headers,body,content')
+
+
 class HTTPClient(object):
 
     def __init__(self, host='127.0.0.1', port=8500, token=None, scheme='http', verify=True, cert=None, ca_cert=None):
@@ -52,9 +58,8 @@ class HTTPClient(object):
                 kwargs['cert_file'] = cert
         if ca_cert:
             kwargs['ca_certs'] = ca_cert
-        if verify or ca_cert:
+        kwargs['cert_reqs'] = ssl.CERT_REQUIRED if verify or ca_cert else ssl.CERT_NONE
-            kwargs['cert_reqs'] = ssl.CERT_REQUIRED
+        self.http = urllib3.PoolManager(num_pools=10, maxsize=10, **kwargs)
-        self.http = urllib3.PoolManager(num_pools=10, **kwargs)
         self._ttl = None
 
     def set_read_timeout(self, timeout):
@@ -71,16 +76,17 @@ class HTTPClient(object):
 
     @staticmethod
     def response(response):
-        data = response.data.decode('utf-8')
+        content = response.data
+        body = content.decode('utf-8')
         if response.status == 500:
-            msg = '{0} {1}'.format(response.status, data)
+            msg = '{0} {1}'.format(response.status, body)
-            if data.startswith('Invalid Session TTL'):
+            if body.startswith('Invalid Session TTL'):
                 raise InvalidSessionTTL(msg)
-            elif data.startswith('invalid session'):
+            elif body.startswith('invalid session'):
                 raise InvalidSession(msg)
             else:
                 raise ConsulInternalError(msg)
-        return base.Response(response.status, response.headers, data)
+        return Response(response.status, response.headers, body, content)
 
     def uri(self, path, params=None):
         return '{0}{1}{2}'.format(self.base_uri, path, params and '?' + urlencode(params) or '')
@@ -89,7 +95,7 @@ class HTTPClient(object):
         if method not in ('get', 'post', 'put', 'delete'):
             raise AttributeError("HTTPClient instance has no attribute '{0}'".format(method))
 
-        def wrapper(callback, path, params=None, data=''):
+        def wrapper(callback, path, params=None, data='', headers=None):
             # python-consul doesn't allow to specify ttl smaller then 10 seconds
             # because session_ttl_min defaults to 10s, so we have to do this ugly dirty hack...
             if method == 'put' and path == '/v1/session/create':
@@ -106,13 +112,15 @@ class HTTPClient(object):
                 # According to the documentation a small random amount of additional wait time is added to the
                 # supplied maximum wait time to spread out the wake up time of any concurrent requests. This adds
                 # up to wait / 16 additional time to the maximum duration. Since our goal is actually getting a
-                # response rather read timeout we will add to the timeout a sligtly bigger value.
+                # response rather read timeout we will add to the timeout a slightly bigger value.
                 kwargs['timeout'] = timeout + max(timeout/15.0, 1)
             else:
                 kwargs['timeout'] = self._read_timeout
+            kwargs['headers'] = (headers or {}).copy()
+            kwargs['headers'].update(urllib3.make_headers(user_agent=USER_AGENT))
             token = params.pop('token', self.token) if isinstance(params, dict) else self.token
             if token:
-                kwargs['headers'] = {'X-Consul-Token': token}
+                kwargs['headers']['X-Consul-Token'] = token
             return callback(self.response(self.http.request(method.upper(), self.uri(path, params), **kwargs)))
         return wrapper
 
@@ -125,7 +133,7 @@ class ConsulClient(base.Consul):
         self.token = kwargs.get('token')
         super(ConsulClient, self).__init__(*args, **kwargs)
 
-    def connect(self, *args, **kwargs):
+    def http_connect(self, *args, **kwargs):
         kwargs.update(dict(zip(['host', 'port', 'scheme', 'verify'], args)))
         if self._cert:
             kwargs['cert'] = self._cert
@@ -135,6 +143,9 @@ class ConsulClient(base.Consul):
             kwargs['token'] = self.token
         return HTTPClient(**kwargs)
 
+    def connect(self, *args, **kwargs):
+        return self.http_connect(*args, **kwargs)
+
     def reload_config(self, config):
         self.http.token = self.token = config.get('token')
         self.consistency = config.get('consistency', 'default')
@@ -179,6 +190,7 @@ class Consul(AbstractDCS):
 
     def __init__(self, config):
         super(Consul, self).__init__(config)
+        self._base_path = self._base_path[1:]
         self._scope = config['scope']
         self._session = None
         self.__do_not_watch = False
@@ -215,16 +227,18 @@ class Consul(AbstractDCS):
         self.set_retry_timeout(config['retry_timeout'])
         self.set_ttl(config.get('ttl') or 30)
         self._last_session_refresh = 0
-        self.__session_checks = config.get('checks')
+        self.__session_checks = config.get('checks', [])
         self._register_service = config.get('register_service', False)
+        self._previous_loop_register_service = self._register_service
+        self._service_tags = sorted(config.get('service_tags', []))
+        self._previous_loop_service_tags = self._service_tags
         if self._register_service:
-            self._service_name = service_name_from_scope_name(self._scope)
+            self._set_service_name()
-            if self._scope != self._service_name:
-                logger.warning('Using %s as consul service name instead of scope name %s', self._service_name,
-                               self._scope)
         self._service_check_interval = config.get('service_check_interval', '5s')
+        self._service_check_tls_server_name = config.get('service_check_tls_server_name', None)
         if not self._ctl:
             self.create_session()
+        self._previous_loop_token = self._client.token
 
     def retry(self, *args, **kwargs):
         return self._retry.copy()(*args, **kwargs)
@@ -239,7 +253,18 @@ class Consul(AbstractDCS):
 
     def reload_config(self, config):
         super(Consul, self).reload_config(config)
-        self._client.reload_config(config.get('consul', {}))
+
+        consul_config = config.get('consul', {})
+        self._client.reload_config(consul_config)
+        self._previous_loop_service_tags = self._service_tags
+        self._service_tags = sorted(consul_config.get('service_tags', []))
+
+        should_register_service = consul_config.get('register_service', False)
+        if should_register_service and not self._register_service:
+            self._set_service_name()
+
+        self._previous_loop_register_service = self._register_service
+        self._register_service = should_register_service
 
     def set_ttl(self, ttl):
         if self._client.http.set_ttl(ttl/2.0):  # Consul multiplies the TTL by 2x
@@ -248,7 +273,7 @@ class Consul(AbstractDCS):
 
     @property
     def ttl(self):
-        return self._client.http.ttl
+        return self._client.http.ttl * 2  # we multiply the value by 2 because it was divided in the `set_ttl()` method
 
     def set_retry_timeout(self, retry_timeout):
         self._retry.deadline = retry_timeout
@@ -263,9 +288,9 @@ class Consul(AbstractDCS):
         except Exception:
             logger.exception('adjust_ttl')
 
-    def _do_refresh_session(self):
+    def _do_refresh_session(self, force=False):
         """:returns: `!True` if it had to create new session"""
-        if self._session and self._last_session_refresh + self._loop_wait > time.time():
+        if not force and self._session and self._last_session_refresh + self._loop_wait > time.time():
             return False
 
         if self._session:
@@ -294,92 +319,126 @@ class Consul(AbstractDCS):
             logger.exception('refresh_session')
         raise ConsulError('Failed to renew/create session')
 
-    def client_path(self, path):
-        return super(Consul, self).client_path(path)[1:]
-
     @staticmethod
     def member(node):
         return Member.from_node(node['ModifyIndex'], os.path.basename(node['Key']), node.get('Session'), node['Value'])
 
-    def _load_cluster(self):
+    def _cluster_from_nodes(self, nodes):
+        # get initialize flag
+        initialize = nodes.get(self._INITIALIZE)
+        initialize = initialize and initialize['Value']
+
+        # get global dynamic configuration
+        config = nodes.get(self._CONFIG)
+        config = config and ClusterConfig.from_node(config['ModifyIndex'], config['Value'])
+
+        # get timeline history
+        history = nodes.get(self._HISTORY)
+        history = history and TimelineHistory.from_node(history['ModifyIndex'], history['Value'])
+
+        # get last known leader lsn and slots
+        status = nodes.get(self._STATUS)
+        if status:
+            try:
+                status = json.loads(status['Value'])
+                last_lsn = status.get(self._OPTIME)
+                slots = status.get('slots')
+            except Exception:
+                slots = last_lsn = None
+        else:
+            last_lsn = nodes.get(self._LEADER_OPTIME)
+            last_lsn = last_lsn and last_lsn['Value']
+            slots = None
+
         try:
-            path = self.client_path('/')
+            last_lsn = int(last_lsn)
-            _, results = self.retry(self._client.kv.get, path, recurse=True)
+        except Exception:
+            last_lsn = 0
 
-            if results is None:
+        # get list of members
-                raise NotFound
+        members = [self.member(n) for k, n in nodes.items() if k.startswith(self._MEMBERS) and k.count('/') == 1]
 
-            nodes = {}
+        # get leader
-            for node in results:
+        leader = nodes.get(self._LEADER)
+
+        if leader:
+            member = Member(-1, leader['Value'], None, {})
+            member = ([m for m in members if m.name == leader['Value']] or [member])[0]
+            leader = Leader(leader['ModifyIndex'], leader.get('Session'), member)
+
+        # failover key
+        failover = nodes.get(self._FAILOVER)
+        if failover:
+            failover = Failover.from_node(failover['ModifyIndex'], failover['Value'])
+
+        # get synchronization state
+        sync = nodes.get(self._SYNC)
+        sync = SyncState.from_node(sync and sync['ModifyIndex'], sync and sync['Value'])
+
+        # get failsafe topology
+        failsafe = nodes.get(self._FAILSAFE)
+        try:
+            failsafe = json.loads(failsafe['Value']) if failsafe else None
+        except Exception:
+            failsafe = None
+
+        return Cluster(initialize, config, leader, last_lsn, members, failover, sync, history, slots, failsafe)
+
+    def _cluster_loader(self, path):
+        _, results = self.retry(self._client.kv.get, path, recurse=True)
+        if results is None:
+            raise NotFound
+        nodes = {}
+        for node in results:
+            node['Value'] = (node['Value'] or b'').decode('utf-8')
+            nodes[node['Key'][len(path):]] = node
+
+        return self._cluster_from_nodes(nodes)
+
+    def _citus_cluster_loader(self, path):
+        _, results = self.retry(self._client.kv.get, path, recurse=True)
+        clusters = defaultdict(dict)
+        for node in results or []:
+            key = node['Key'][len(path):].split('/', 1)
+            if len(key) == 2 and citus_group_re.match(key[0]):
                 node['Value'] = (node['Value'] or b'').decode('utf-8')
-                nodes[node['Key'][len(path):].lstrip('/')] = node
+                clusters[int(key[0])][key[1]] = node
+        return {group: self._cluster_from_nodes(nodes) for group, nodes in clusters.items()}
 
-            # get initialize flag
+    def _load_cluster(self, path, loader):
-            initialize = nodes.get(self._INITIALIZE)
+        try:
-            initialize = initialize and initialize['Value']
+            return loader(path)
-
-            # get global dynamic configuration
-            config = nodes.get(self._CONFIG)
-            config = config and ClusterConfig.from_node(config['ModifyIndex'], config['Value'])
-
-            # get timeline history
-            history = nodes.get(self._HISTORY)
-            history = history and TimelineHistory.from_node(history['ModifyIndex'], history['Value'])
-
-            # get last leader operation
-            last_leader_operation = nodes.get(self._LEADER_OPTIME)
-            last_leader_operation = 0 if last_leader_operation is None else int(last_leader_operation['Value'])
-
-            # get list of members
-            members = [self.member(n) for k, n in nodes.items() if k.startswith(self._MEMBERS) and k.count('/') == 1]
-
-            # get leader
-            leader = nodes.get(self._LEADER)
-            if not self._ctl and leader and leader['Value'] == self._name \
-                    and self._session != leader.get('Session', 'x'):
-                logger.info('I am leader but not owner of the session. Removing leader node')
-                self._client.kv.delete(self.leader_path, cas=leader['ModifyIndex'])
-                leader = None
-
-            if leader:
-                member = Member(-1, leader['Value'], None, {})
-                member = ([m for m in members if m.name == leader['Value']] or [member])[0]
-                leader = Leader(leader['ModifyIndex'], leader.get('Session'), member)
-
-            # failover key
-            failover = nodes.get(self._FAILOVER)
-            if failover:
-                failover = Failover.from_node(failover['ModifyIndex'], failover['Value'])
-
-            # get synchronization state
-            sync = nodes.get(self._SYNC)
-            sync = SyncState.from_node(sync and sync['ModifyIndex'], sync and sync['Value'])
-
-            return Cluster(initialize, config, leader, last_leader_operation, members, failover, sync, history)
         except NotFound:
-            return Cluster(None, None, None, None, [], None, None, None)
+            return Cluster(None, None, None, None, [], None, None, None, None, None)
         except Exception:
             logger.exception('get_cluster')
             raise ConsulError('Consul is not responding properly')
 
     @catch_consul_errors
-    def touch_member(self, data, permanent=False):
+    def touch_member(self, data):
         cluster = self.cluster
         member = cluster and cluster.get_member(self._name, fallback_to_leader=False)
-        create_member = not permanent and self.refresh_session()
+
+        try:
+            create_member = self.refresh_session()
+        except DCSError:
+            return False
 
         if member and (create_member or member.session != self._session):
             self._client.kv.delete(self.member_path)
             create_member = True
 
+        if self._register_service or self._previous_loop_register_service:
+            try:
+                self.update_service(not create_member and member and member.data or {}, data)
+            except Exception:
+                logger.exception('update_service')
+
         if not create_member and member and deep_compare(data, member.data):
             return True
 
         try:
-            args = {} if permanent else {'acquire': self._session}
+            self._client.kv.put(self.member_path, json.dumps(data, separators=(',', ':')), acquire=self._session)
-            self._client.kv.put(self.member_path, json.dumps(data, separators=(',', ':')), **args)
-            if self._register_service:
-                self.update_service(not create_member and member and member.data or {}, data)
             return True
         except InvalidSession:
             self._session = None
@@ -388,6 +447,11 @@ class Consul(AbstractDCS):
             logger.exception('touch_member')
         return False
 
+    def _set_service_name(self):
+        self._service_name = service_name_from_scope_name(self._scope)
+        if self._scope != self._service_name:
+            logger.warning('Using %s as consul service name instead of scope name %s', self._service_name, self._scope)
+
     @catch_consul_errors
     def register_service(self, service_name, **kwargs):
         logger.info('Register service %s, params %s', service_name, kwargs)
@@ -409,18 +473,32 @@ class Consul(AbstractDCS):
         conn_parts = urlparse(data['conn_url'])
         check = base.Check.http(api_parts.geturl(), self._service_check_interval,
                                 deregister='{0}s'.format(self._client.http.ttl * 10))
+        if self._service_check_tls_server_name is not None:
+            check['TLSServerName'] = self._service_check_tls_server_name
+        tags = self._service_tags[:]
+        tags.append(role)
+        if role == 'master':
+            tags.append('primary')
+        elif role == 'primary':
+            tags.append('master')
+        self._previous_loop_service_tags = self._service_tags
+        self._previous_loop_token = self._client.token
+
         params = {
             'service_id': '{0}/{1}'.format(self._scope, self._name),
             'address': conn_parts.hostname,
             'port': conn_parts.port,
             'check': check,
-            'tags': [role]
+            'tags': tags,
+            'enable_tag_override': True,
         }
 
-        if state == 'stopped':
+        if state == 'stopped' or (not self._register_service and self._previous_loop_register_service):
+            self._previous_loop_register_service = self._register_service
             return self.deregister_service(params['service_id'])
 
-        if role in ['master', 'replica', 'standby-leader']:
+        self._previous_loop_register_service = self._register_service
+        if role in ['master', 'primary', 'replica', 'standby-leader']:
             if state != 'running':
                 return
             return self.register_service(service_name, **params)
@@ -438,25 +516,39 @@ class Consul(AbstractDCS):
             if old_data.get(key) != new_data[key]:
                 update = True
 
-        if force or update:
+        if (
+            force or update or self._register_service != self._previous_loop_register_service
+            or self._service_tags != self._previous_loop_service_tags
+            or self._client.token != self._previous_loop_token
+        ):
             return self._update_service(new_data)
 
-    @catch_consul_errors
+    def _do_attempt_to_acquire_leader(self, retry):
-    def _do_attempt_to_acquire_leader(self, permanent):
         try:
-            kwargs = {} if permanent else {'acquire': self._session}
+            return retry(self._client.kv.put, self.leader_path, self._name, acquire=self._session)
-            return self.retry(self._client.kv.put, self.leader_path, self._name, **kwargs)
         except InvalidSession:
-            self._session = None
             logger.error('Our session disappeared from Consul. Will try to get a new one and retry attempt')
-            self.refresh_session()
+            self._session = None
-            return self.retry(self._client.kv.put, self.leader_path, self._name, acquire=self._session)
+            retry.deadline = retry.stoptime - time.time()
 
-    def attempt_to_acquire_leader(self, permanent=False):
+            retry(self._do_refresh_session)
-        if not self._session and not permanent:
-            self.refresh_session()
 
-        ret = self._do_attempt_to_acquire_leader(permanent)
+            retry.deadline = retry.stoptime - time.time()
+            if retry.deadline < 1:
+                raise ConsulError('_do_attempt_to_acquire_leader timeout')
+
+            return retry(self._client.kv.put, self.leader_path, self._name, acquire=self._session)
+
+    @catch_return_false_exception
+    def attempt_to_acquire_leader(self):
+        retry = self._retry.copy()
+        self._run_and_handle_exceptions(self._do_refresh_session, retry=retry)
+
+        retry.deadline = retry.stoptime - time.time()
+        if retry.deadline < 1:
+            raise ConsulError('attempt_to_acquire_leader timeout')
+
+        ret = self._run_and_handle_exceptions(self._do_attempt_to_acquire_leader, retry, retry=None)
         if not ret:
             logger.info('Could not take out TTL lock')
 
@@ -474,14 +566,50 @@ class Consul(AbstractDCS):
         return self._client.kv.put(self.config_path, value, cas=index)
 
     @catch_consul_errors
-    def _write_leader_optime(self, last_operation):
+    def _write_leader_optime(self, last_lsn):
-        return self._client.kv.put(self.leader_optime_path, last_operation)
+        return self._client.kv.put(self.leader_optime_path, last_lsn)
 
     @catch_consul_errors
+    def _write_status(self, value):
+        return self._client.kv.put(self.status_path, value)
+
+    @catch_consul_errors
+    def _write_failsafe(self, value):
+        return self._client.kv.put(self.failsafe_path, value)
+
+    @staticmethod
+    def _run_and_handle_exceptions(method, *args, **kwargs):
+        retry = kwargs.pop('retry', None)
+        try:
+            return retry(method, *args, **kwargs) if retry else method(*args, **kwargs)
+        except (RetryFailedError, InvalidSession, HTTPException, HTTPError, socket.error, socket.timeout) as e:
+            raise ConsulError(e)
+        except ConsulException:
+            raise ReturnFalseException
+
+    @catch_return_false_exception
     def _update_leader(self):
+        retry = self._retry.copy()
+
+        self._run_and_handle_exceptions(self._do_refresh_session, True, retry=retry)
+
         if self._session:
-            self.retry(self._client.session.renew, self._session)
+            cluster = self.cluster
-            self._last_session_refresh = time.time()
+            leader_session = cluster and isinstance(cluster.leader, Leader) and cluster.leader.session
+            if leader_session != self._session:
+                retry.deadline = retry.stoptime - time.time()
+                if retry.deadline < 1:
+                    raise ConsulError('update_leader timeout')
+                logger.warning('Recreating the leader key due to session mismatch')
+                if cluster.leader:
+                    self._run_and_handle_exceptions(self._client.kv.delete, self.leader_path, cas=cluster.leader.index)
+
+                retry.deadline = retry.stoptime - time.time()
+                if retry.deadline < 0.5:
+                    raise ConsulError('update_leader timeout')
+                self._run_and_handle_exceptions(self._client.kv.put, self.leader_path,
+                                                self._name, acquire=self._session)
+
         return bool(self._session)
 
     @catch_consul_errors
@@ -502,7 +630,7 @@ class Consul(AbstractDCS):
         return self._client.kv.put(self.history_path, value)
 
     @catch_consul_errors
-    def delete_leader(self):
+    def _delete_leader(self):
         cluster = self.cluster
         if cluster and isinstance(cluster.leader, Leader) and cluster.leader.name == self._name:
             return self._client.kv.delete(self.leader_path, cas=cluster.leader.index)

patroni/dcs/etcd.py

@@ -1,30 +1,39 @@
 from __future__ import absolute_import
+import abc
 import etcd
 import json
 import logging
 import os
 import urllib3.util.connection
 import random
-import requests
 import six
 import socket
 import time
 
+from collections import defaultdict
+from copy import deepcopy
 from dns.exception import DNSException
 from dns import resolver
-from patroni.dcs import AbstractDCS, ClusterConfig, Cluster, Failover, Leader, Member, SyncState, TimelineHistory
+from urllib3 import Timeout
-from patroni.exceptions import DCSError
+from urllib3.exceptions import HTTPError, ReadTimeoutError, ProtocolError
-from patroni.utils import Retry, RetryFailedError, split_host_port, uri
-from urllib3.exceptions import HTTPError, ReadTimeoutError
-from requests.exceptions import RequestException
 from six.moves.queue import Queue
 from six.moves.http_client import HTTPException
 from six.moves.urllib_parse import urlparse
 from threading import Thread
 
+from . import AbstractDCS, Cluster, ClusterConfig, Failover, Leader, Member, SyncState,\
+        TimelineHistory, ReturnFalseException, catch_return_false_exception, citus_group_re
+from ..exceptions import DCSError
+from ..request import get as requests_get
+from ..utils import Retry, RetryFailedError, split_host_port, uri, USER_AGENT
+
 logger = logging.getLogger(__name__)
 
 
+class EtcdRaftInternal(etcd.EtcdException):
+    """Raft Internal Error"""
+
+
 class EtcdError(DCSError):
     pass
 
@@ -65,16 +74,20 @@ class DnsCachingResolver(Thread):
     def resolve_async(self, host, port, attempt=0):
         self._resolve_queue.put(((host, port), attempt))
 
+    def remove(self, host, port):
+        self._cache.pop((host, port), None)
+
     @staticmethod
     def _do_resolve(host, port):
         try:
             return socket.getaddrinfo(host, port, 0, socket.SOCK_STREAM, socket.IPPROTO_TCP)
-        except socket.gaierror:
+        except Exception as e:
-            logger.warning('failed to resolve host %s', host)
+            logger.warning('failed to resolve host %s: %s', host, e)
             return []
 
 
-class Client(etcd.Client):
+@six.add_metaclass(abc.ABCMeta)
+class AbstractEtcdClientWithFailover(etcd.Client):
 
     def __init__(self, config, dns_resolver, cache_ttl=300):
         self._dns_resolver = dns_resolver
@@ -82,34 +95,81 @@ class Client(etcd.Client):
         self._machines_cache_updated = 0
         args = {p: config.get(p) for p in ('host', 'port', 'protocol', 'use_proxies', 'username', 'password',
                                            'cert', 'ca_cert') if config.get(p)}
-        super(Client, self).__init__(read_timeout=config['retry_timeout'], **args)
+        super(AbstractEtcdClientWithFailover, self).__init__(read_timeout=config['retry_timeout'], **args)
         # For some reason python3-etcd on debian and ubuntu are not based on the latest version
         # Workaround for the case when https://github.com/jplana/python-etcd/pull/196 is not applied
         self.http.connection_pool_kw.pop('ssl_version', None)
         self._config = config
+        self._initial_machines_cache = []
         self._load_machines_cache()
         self._allow_reconnect = True
+        # allow passing retry argument to api_execute in params
+        self._comparison_conditions.add('retry')
+        self._read_options.add('retry')
+        self._del_conditions.add('retry')
 
-    def _build_request_parameters(self):
+    def _calculate_timeouts(self, etcd_nodes, timeout=None):
-        kwargs = {'headers': self._get_headers(), 'redirect': self.allow_redirect}
+        """Calculate a request timeout and number of retries per single etcd node.
+        In case if the timeout per node is too small (less than one second) we will reduce the number of nodes.
+        For the cluster with only one node we will try to do 2 retries.
+        For clusters with 2 nodes we will try to do 1 retry for every node.
+        No retries for clusters with 3 or more nodes. We better rely on switching to a different node."""
 
-        # calculate the number of retries and timeout *per node*
+        per_node_timeout = timeout = float(timeout or self.read_timeout)
-        # actual number of retries depends on the number of nodes
-        etcd_nodes = len(self._machines_cache) + 1
-        kwargs['retries'] = 0 if etcd_nodes > 3 else (1 if etcd_nodes > 1 else 2)
 
-        # if etcd_nodes > 3:
+        max_retries = 4 - min(etcd_nodes, 3)
-        #     kwargs.update({'retries': 0, 'timeout': float(self.read_timeout)/etcd_nodes})
+        per_node_retries = 1
-        # elif etcd_nodes > 1:
+        min_timeout = 1.0
-        #     kwargs.update({'retries': 1, 'timeout': self.read_timeout/2.0/etcd_nodes})
+
-        # else:
+        while etcd_nodes > 0:
-        #     kwargs.update({'retries': 2, 'timeout': self.read_timeout/3.0})
+            per_node_timeout = float(timeout) / etcd_nodes
-        kwargs['timeout'] = self.read_timeout/float(kwargs['retries'] + 1)/etcd_nodes
+            if per_node_timeout >= min_timeout:
+                # for small clusters we will try to do more than on try on every node
+                while per_node_retries < max_retries and per_node_timeout / (per_node_retries + 1) >= min_timeout:
+                    per_node_retries += 1
+                per_node_timeout /= per_node_retries
+                break
+            # if the timeout per one node is to small try to reduce number of nodes
+            etcd_nodes -= 1
+            max_retries = 1
+
+        return etcd_nodes, per_node_timeout, per_node_retries - 1
+
+    def reload_config(self, config):
+        self.username = config.get('username')
+        self.password = config.get('password')
+
+    def _get_headers(self):
+        basic_auth = ':'.join((self.username, self.password)) if self.username and self.password else None
+        return urllib3.make_headers(basic_auth=basic_auth, user_agent=USER_AGENT)
+
+    def _prepare_common_parameters(self, etcd_nodes, timeout=None):
+        kwargs = {'headers': self._get_headers(), 'redirect': self.allow_redirect, 'preload_content': False}
+
+        if timeout is not None:
+            kwargs.update(retries=0, timeout=timeout)
+        else:
+            _, per_node_timeout, per_node_retries = self._calculate_timeouts(etcd_nodes)
+            connect_timeout = max(1, per_node_timeout/2)
+            kwargs.update(timeout=Timeout(connect=connect_timeout, total=per_node_timeout), retries=per_node_retries)
         return kwargs
 
     def set_machines_cache_ttl(self, cache_ttl):
         self._machines_cache_ttl = cache_ttl
 
+    @abc.abstractmethod
+    def _prepare_get_members(self, etcd_nodes):
+        """returns: request parameters"""
+
+    @abc.abstractmethod
+    def _get_members(self, base_uri, **kwargs):
+        """returns: list of clientURLs"""
+
+    @property
+    def machines_cache(self):
+        base_uri, cache = self._base_uri, self._machines_cache
+        return ([base_uri] if base_uri in cache else []) + [machine for machine in cache if machine != base_uri]
+
     @property
     def machines(self):
         """Original `machines` method(property) of `etcd.Client` class raise exception
@@ -122,100 +182,106 @@ class Client(etcd.Client):
         Also this method implements the same timeout-retry logic as `api_execute`, because
         the original method was retrying 2 times with the `read_timeout` on each node."""
 
-        kwargs = self._build_request_parameters()
+        machines_cache = self.machines_cache
+        kwargs = self._prepare_get_members(len(machines_cache))
 
-        while True:
+        for base_uri in machines_cache:
             try:
-                response = self.http.request(self._MGET, self._base_uri + self.version_prefix + '/machines', **kwargs)
+                machines = list(set(self._get_members(base_uri, **kwargs)))
-                data = self._handle_server_response(response).data.decode('utf-8')
-                machines = [m.strip() for m in data.split(',') if m.strip()]
                 logger.debug("Retrieved list of machines: %s", machines)
-                if not machines:
+                if machines:
-                    raise etcd.EtcdException
+                    random.shuffle(machines)
-                random.shuffle(machines)
+                    if not self._use_proxies:
-                for url in machines:
+                        self._update_dns_cache(self._dns_resolver.resolve_async, machines)
-                    r = urlparse(url)
+                    return machines
-                    port = r.port or (443 if r.scheme == 'https' else 80)
-                    self._dns_resolver.resolve_async(r.hostname, port)
-                return machines
             except Exception as e:
-                # We can't get the list of machines, if one server is in the
+                self.http.clear()
-                # machines cache, try on it
+                logger.error("Failed to get list of machines from %s%s: %r", base_uri, self.version_prefix, e)
-                logger.error("Failed to get list of machines from %s%s: %r", self._base_uri, self.version_prefix, e)
+
-                if self._machines_cache:
+        raise etcd.EtcdConnectionFailed('No more machines in the cluster')
-                    self._base_uri = self._machines_cache.pop(0)
-                    logger.info("Retrying on %s", self._base_uri)
-                elif self._update_machines_cache:
-                    raise etcd.EtcdException("Could not get the list of servers, "
-                                             "maybe you provided the wrong "
-                                             "host(s) to connect to?")
-                else:
-                    return []
 
     def set_read_timeout(self, timeout):
         self._read_timeout = timeout
 
-    def _do_http_request(self, request_executor, method, url, fields=None, **kwargs):
+    def _do_http_request(self, retry, machines_cache, request_executor, method, path, fields=None, **kwargs):
-        try:
+        if fields is not None:
-            response = request_executor(method, url, fields=fields, **kwargs)
+            kwargs['fields'] = fields
-            response.data.decode('utf-8')
+        some_request_failed = False
-            self._check_cluster_id(response)
+        for i, base_uri in enumerate(machines_cache):
-        except (HTTPError, HTTPException, socket.error, socket.timeout) as e:
+            if i > 0:
-            if (isinstance(fields, dict) and fields.get("wait") == "true" and
+                logger.info("Retrying on %s", base_uri)
-                    isinstance(e, ReadTimeoutError)):
+            try:
-                logger.debug("Watch timed out.")
+                response = request_executor(method, base_uri + path, **kwargs)
-                raise etcd.EtcdWatchTimedOut("Watch timed out: {0}".format(e), cause=e)
+                response.data.decode('utf-8')
-            logger.error("Request to server %s failed: %r", self._base_uri, e)
+                if some_request_failed:
-            logger.info("Reconnection allowed, looking for another server.")
+                    self.set_base_uri(base_uri)
-            self._base_uri = self._next_server(cause=e)
+                    self._refresh_machines_cache()
-            response = False
+                return response
-        return response
+            except (HTTPError, HTTPException, socket.error, socket.timeout) as e:
+                self.http.clear()
+                if not retry:
+                    if len(machines_cache) == 1:
+                        self.set_base_uri(self._base_uri)  # trigger Etcd3 watcher restart
+                    # switch to the next etcd node because we don't know exactly what happened,
+                    # whether the key didn't received an update or there is a network problem.
+                    elif i + 1 < len(machines_cache):
+                        self.set_base_uri(machines_cache[i + 1])
+                if (isinstance(fields, dict) and fields.get("wait") == "true" and
+                        isinstance(e, (ReadTimeoutError, ProtocolError))):
+                    logger.debug("Watch timed out.")
+                    raise etcd.EtcdWatchTimedOut("Watch timed out: {0}".format(e), cause=e)
+                logger.error("Request to server %s failed: %r", base_uri, e)
+                logger.info("Reconnection allowed, looking for another server.")
+                if not retry:
+                    raise etcd.EtcdException('{0} {1} request failed'.format(method, path))
+                some_request_failed = True
+
+        raise etcd.EtcdConnectionFailed('No more machines in the cluster')
+
+    @abc.abstractmethod
+    def _prepare_request(self, kwargs, params=None, method=None):
+        """returns: request_executor"""
 
     def api_execute(self, path, method, params=None, timeout=None):
-        if not path.startswith('/'):
+        retry = params.pop('retry', None) if isinstance(params, dict) else None
-            raise ValueError('Path does not start with /')
-
-        kwargs = {'fields': params, 'preload_content': False}
-
-        if method in [self._MGET, self._MDELETE]:
-            request_executor = self.http.request
-        elif method in [self._MPUT, self._MPOST]:
-            request_executor = self.http.request_encode_body
-            kwargs['encode_multipart'] = False
-        else:
-            raise etcd.EtcdException('HTTP method {0} not supported'.format(method))
 
         # Update machines_cache if previous attempt of update has failed
         if self._update_machines_cache:
             self._load_machines_cache()
         elif not self._use_proxies and time.time() - self._machines_cache_updated > self._machines_cache_ttl:
             self._refresh_machines_cache()
-            self._machines_cache_updated = time.time()
 
-        kwargs.update(self._build_request_parameters())
+        machines_cache = self.machines_cache
+        etcd_nodes = len(machines_cache)
 
-        if timeout is not None:
+        kwargs = self._prepare_common_parameters(etcd_nodes, timeout)
-            kwargs.update({'retries': 0, 'timeout': timeout})
+        request_executor = self._prepare_request(kwargs, params, method)
 
-        response = False
+        while True:
-
+            try:
-        try:
+                response = self._do_http_request(retry, machines_cache, request_executor, method, path, **kwargs)
-            some_request_failed = False
+                return self._handle_server_response(response)
-            while not response:
+            except etcd.EtcdWatchTimedOut:
-                response = self._do_http_request(request_executor, method, self._base_uri + path, **kwargs)
-
-                if response is False:
-                    some_request_failed = True
-            if some_request_failed:
-                self._refresh_machines_cache()
-        except etcd.EtcdConnectionFailed as e:
-            if isinstance(e, etcd.EtcdWatchTimedOut) and self._machines_cache:
-                self._base_uri = self._next_server()
-            else:
-                self._update_machines_cache = True
-            if not response:
                 raise
-        return self._handle_server_response(response)
+            except etcd.EtcdConnectionFailed as ex:
+                try:
+                    if self._load_machines_cache():
+                        machines_cache = self.machines_cache
+                        etcd_nodes = len(machines_cache)
+                except Exception as e:
+                    logger.debug('Failed to update list of etcd nodes: %r', e)
+                sleeptime = retry.sleeptime
+                remaining_time = retry.stoptime - sleeptime - time.time()
+                nodes, timeout, retries = self._calculate_timeouts(etcd_nodes, remaining_time)
+                if nodes == 0:
+                    self._update_machines_cache = True
+                    self.set_base_uri(self._base_uri)  # trigger Etcd3 watcher restart
+                    raise ex
+                retry.sleep_func(sleeptime)
+                retry.update_delay()
+                # We still have some time left. Partially reduce `machines_cache` and retry request
+                kwargs.update(timeout=Timeout(connect=max(1, timeout/2), total=timeout), retries=retries)
+                machines_cache = machines_cache[:nodes]
 
     @staticmethod
     def get_srv_record(host):
@@ -224,25 +290,26 @@ class Client(etcd.Client):
         except DNSException:
             return []
 
-    def _get_machines_cache_from_srv(self, srv):
+    def _get_machines_cache_from_srv(self, srv, srv_suffix=None):
         """Fetch list of etcd-cluster member by resolving _etcd-server._tcp. SRV record.
         This record should contain list of host and peer ports which could be used to run
         'GET http://{host}:{port}/members' request (peer protocol)"""
 
         ret = []
         for r in ['-client-ssl', '-client', '-ssl', '', '-server-ssl', '-server']:
+            r = '{0}-{1}'.format(r, srv_suffix) if srv_suffix else r
             protocol = 'https' if '-ssl' in r else 'http'
             endpoint = '/members' if '-server' in r else ''
             for host, port in self.get_srv_record('_etcd{0}._tcp.{1}'.format(r, srv)):
                 url = uri(protocol, (host, port), endpoint)
                 if endpoint:
                     try:
-                        response = requests.get(url, timeout=self.read_timeout, verify=False)
+                        response = requests_get(url, timeout=self.read_timeout, verify=False)
-                        if response.ok:
+                        if response.status < 400:
-                            for member in response.json():
+                            for member in json.loads(response.data.decode('utf-8')):
                                 ret.extend(member['clientURLs'])
                             break
-                    except RequestException:
+                    except Exception:
                         logger.exception('GET %s', url)
                 else:
                     ret.append(url)
@@ -267,7 +334,7 @@ class Client(etcd.Client):
 
         machines_cache = []
         if 'srv' in self._config:
-            machines_cache = self._get_machines_cache_from_srv(self._config['srv'])
+            machines_cache = self._get_machines_cache_from_srv(self._config['srv'], self._config.get('srv_suffix'))
 
         if not machines_cache and 'hosts' in self._config:
             machines_cache = list(self._config['hosts'])
@@ -276,6 +343,13 @@ class Client(etcd.Client):
             machines_cache = self._get_machines_cache_from_dns(self._config['host'], self._config['port'])
         return machines_cache
 
+    @staticmethod
+    def _update_dns_cache(func, machines):
+        for url in machines:
+            r = urlparse(url)
+            port = r.port or (443 if r.scheme == 'https' else 80)
+            func(r.hostname, port)
+
     def _load_machines_cache(self):
         """This method should fill up `_machines_cache` from scratch.
         It could happen only in two cases:
@@ -287,40 +361,96 @@ class Client(etcd.Client):
         if 'srv' not in self._config and 'host' not in self._config and 'hosts' not in self._config:
             raise Exception('Neither srv, hosts, host nor url are defined in etcd section of config')
 
-        self._machines_cache = self._get_machines_cache_from_config()
+        machines_cache = self._get_machines_cache_from_config()
-
         # Can not bootstrap list of etcd-cluster members, giving up
-        if not self._machines_cache:
+        if not machines_cache:
             raise etcd.EtcdException
 
-        # After filling up initial list of machines_cache we should ask etcd-cluster about actual list
+        # enforce resolving dns name,they might get new ips
-        self._base_uri = self._next_server()
+        self._update_dns_cache(self._dns_resolver.remove, machines_cache)
-        self._refresh_machines_cache()
+
+        # The etcd cluster could change its topology over time and depending on how we resolve the initial
+        # topology (list of hosts in the Patroni config or DNS records, A or SRV) we might get into the situation
+        # the the real topology doesn't match anymore with the topology resolved from the configuration file.
+        # In case if the "initial" topology is the same as before we will not override the `_machines_cache`.
+        ret = set(machines_cache) != set(self._initial_machines_cache)
+        if ret:
+            self._initial_machines_cache = self._machines_cache = machines_cache
+
+            # After filling up the initial list of machines_cache we should ask etcd-cluster about actual list
+            self._refresh_machines_cache(True)
 
         self._update_machines_cache = False
+        return ret
+
+    def _refresh_machines_cache(self, updating_cache=False):
+        if self._use_proxies:
+            self._machines_cache = self._get_machines_cache_from_config()
+        else:
+            try:
+                self._machines_cache = self.machines
+            except etcd.EtcdConnectionFailed:
+                if updating_cache:
+                    raise etcd.EtcdException("Could not get the list of servers, "
+                                             "maybe you provided the wrong "
+                                             "host(s) to connect to?")
+                return
+
+        if self._base_uri not in self._machines_cache:
+            self.set_base_uri(self._machines_cache[0])
         self._machines_cache_updated = time.time()
 
-    def _refresh_machines_cache(self):
+    def set_base_uri(self, value):
-        self._machines_cache = self._get_machines_cache_from_config() if self._use_proxies else self.machines
+        if self._base_uri != value:
-        if self._base_uri in self._machines_cache:
+            logger.info('Selected new etcd server %s', value)
-            self._machines_cache.remove(self._base_uri)
+            self._base_uri = value
 
 
-class Etcd(AbstractDCS):
+class EtcdClient(AbstractEtcdClientWithFailover):
 
-    def __init__(self, config):
+    ERROR_CLS = EtcdError
-        super(Etcd, self).__init__(config)
+
-        self._ttl = int(config.get('ttl') or 30)
+    def __del__(self):
+        if self.http is not None:
+            try:
+                self.http.clear()
+            except (ReferenceError, TypeError, AttributeError):
+                pass
+
+    def _prepare_get_members(self, etcd_nodes):
+        return self._prepare_common_parameters(etcd_nodes)
+
+    def _get_members(self, base_uri, **kwargs):
+        response = self.http.request(self._MGET, base_uri + self.version_prefix + '/machines', **kwargs)
+        data = self._handle_server_response(response).data.decode('utf-8')
+        return [m.strip() for m in data.split(',') if m.strip()]
+
+    def _prepare_request(self, kwargs, params=None, method=None):
+        kwargs['fields'] = params
+        if method in (self._MPOST, self._MPUT):
+            kwargs['encode_multipart'] = False
+        return self.http.request
+
+
+class AbstractEtcd(AbstractDCS):
+
+    def __init__(self, config, client_cls, retry_errors_cls):
+        super(AbstractEtcd, self).__init__(config)
         self._retry = Retry(deadline=config['retry_timeout'], max_delay=1, max_tries=-1,
-                            retry_exceptions=(etcd.EtcdLeaderElectionInProgress,
+                            retry_exceptions=retry_errors_cls)
-                                              etcd.EtcdWatcherCleared,
+        self._ttl = int(config.get('ttl') or 30)
-                                              etcd.EtcdEventIndexCleared))
+        self._client = self.get_etcd_client(config, client_cls)
-        self._client = self.get_etcd_client(config)
         self.__do_not_watch = False
         self._has_failed = False
 
+    def reload_config(self, config):
+        super(AbstractEtcd, self).reload_config(config)
+        self._client.reload_config(config.get(self.__class__.__name__.lower(), {}))
+
     def retry(self, *args, **kwargs):
-        return self._retry.copy()(*args, **kwargs)
+        retry = self._retry.copy()
+        kwargs['retry'] = retry
+        return retry(*args, **kwargs)
 
     def _handle_exception(self, e, name='', do_sleep=False, raise_ex=None):
         if not self._has_failed:
@@ -333,22 +463,26 @@ class Etcd(AbstractDCS):
         if isinstance(raise_ex, Exception):
             raise raise_ex
 
-    def catch_etcd_errors(func):
+    def _run_and_handle_exceptions(self, method, *args, **kwargs):
-        def wrapper(self, *args, **kwargs):
+        retry = kwargs.pop('retry', self.retry)
-            try:
+        try:
-                retval = func(self, *args, **kwargs) is not None
+            return retry(method, *args, **kwargs) if retry else method(*args, **kwargs)
-                self._has_failed = False
+        except (RetryFailedError, etcd.EtcdConnectionFailed) as e:
-                return retval
+            raise self._client.ERROR_CLS(e)
-            except (RetryFailedError, etcd.EtcdException) as e:
+        except etcd.EtcdException as e:
-                self._handle_exception(e)
+            self._handle_exception(e)
-                return False
+            raise ReturnFalseException
-            except Exception as e:
+        except Exception as e:
-                self._handle_exception(e, raise_ex=EtcdError('unexpected error'))
+            self._handle_exception(e, raise_ex=self._client.ERROR_CLS('unexpected error'))
-
-        return wrapper
 
     @staticmethod
-    def get_etcd_client(config):
+    def set_socket_options(sock, socket_options):
+        if socket_options:
+            for opt in socket_options:
+                sock.setsockopt(*opt)
+
+    def get_etcd_client(self, config, client_cls):
+        config = deepcopy(config)
         if 'proxy' in config:
             config['use_proxies'] = True
             config['url'] = config['proxy']
@@ -368,7 +502,7 @@ class Etcd(AbstractDCS):
             config['hosts'] = []
             for value in hosts:
                 if isinstance(value, six.string_types):
-                    config['hosts'].append(uri(protocol, split_host_port(value, default_port)))
+                    config['hosts'].append(uri(protocol, split_host_port(value.strip(), default_port)))
         elif 'host' in config:
             host, port = split_host_port(config['host'], 2379)
             config['host'] = host
@@ -397,9 +531,7 @@ class Etcd(AbstractDCS):
                 sock = None
                 try:
                     sock = socket.socket(af, socktype, proto)
-                    if socket_options:
+                    self.set_socket_options(sock, socket_options)
-                        for opt in socket_options:
-                            sock.setsockopt(*opt)
                     if timeout is not socket._GLOBAL_DEFAULT_TIMEOUT:
                         sock.settimeout(timeout)
                     if source_address:
@@ -423,7 +555,7 @@ class Etcd(AbstractDCS):
         client = None
         while not client:
             try:
-                client = Client(config, dns_resolver)
+                client = client_cls(config, dns_resolver)
                 if 'use_proxies' in config and not client.machines:
                     raise etcd.EtcdException
             except etcd.EtcdException:
@@ -433,9 +565,10 @@ class Etcd(AbstractDCS):
 
     def set_ttl(self, ttl):
         ttl = int(ttl)
-        self.__do_not_watch = self._ttl != ttl
+        ret = self._ttl != ttl
         self._ttl = ttl
         self._client.set_machines_cache_ttl(ttl*10)
+        return ret
 
     @property
     def ttl(self):
@@ -445,81 +578,140 @@ class Etcd(AbstractDCS):
         self._retry.deadline = retry_timeout
         self._client.set_read_timeout(retry_timeout)
 
+
+def catch_etcd_errors(func):
+    def wrapper(self, *args, **kwargs):
+        try:
+            retval = func(self, *args, **kwargs) is not None
+            self._has_failed = False
+            return retval
+        except (RetryFailedError, etcd.EtcdException) as e:
+            self._handle_exception(e)
+            return False
+        except Exception as e:
+            self._handle_exception(e, raise_ex=self._client.ERROR_CLS('unexpected error'))
+
+    return wrapper
+
+
+class Etcd(AbstractEtcd):
+
+    def __init__(self, config):
+        super(Etcd, self).__init__(config, EtcdClient, (etcd.EtcdLeaderElectionInProgress, EtcdRaftInternal))
+        self.__do_not_watch = False
+
+    def set_ttl(self, ttl):
+        self.__do_not_watch = super(Etcd, self).set_ttl(ttl)
+
     @staticmethod
     def member(node):
         return Member.from_node(node.modifiedIndex, os.path.basename(node.key), node.ttl, node.value)
 
-    def _load_cluster(self):
+    def _cluster_from_nodes(self, etcd_index, nodes):
+        # get initialize flag
+        initialize = nodes.get(self._INITIALIZE)
+        initialize = initialize and initialize.value
+
+        # get global dynamic configuration
+        config = nodes.get(self._CONFIG)
+        config = config and ClusterConfig.from_node(config.modifiedIndex, config.value)
+
+        # get timeline history
+        history = nodes.get(self._HISTORY)
+        history = history and TimelineHistory.from_node(history.modifiedIndex, history.value)
+
+        # get last know leader lsn and slots
+        status = nodes.get(self._STATUS)
+        if status:
+            try:
+                status = json.loads(status.value)
+                last_lsn = status.get(self._OPTIME)
+                slots = status.get('slots')
+            except Exception:
+                slots = last_lsn = None
+        else:
+            last_lsn = nodes.get(self._LEADER_OPTIME)
+            last_lsn = last_lsn and last_lsn.value
+            slots = None
+
+        try:
+            last_lsn = int(last_lsn)
+        except Exception:
+            last_lsn = 0
+
+        # get list of members
+        members = [self.member(n) for k, n in nodes.items() if k.startswith(self._MEMBERS) and k.count('/') == 1]
+
+        # get leader
+        leader = nodes.get(self._LEADER)
+        if leader:
+            member = Member(-1, leader.value, None, {})
+            member = ([m for m in members if m.name == leader.value] or [member])[0]
+            index = etcd_index if etcd_index > leader.modifiedIndex else leader.modifiedIndex + 1
+            leader = Leader(index, leader.ttl, member)
+
+        # failover key
+        failover = nodes.get(self._FAILOVER)
+        if failover:
+            failover = Failover.from_node(failover.modifiedIndex, failover.value)
+
+        # get synchronization state
+        sync = nodes.get(self._SYNC)
+        sync = SyncState.from_node(sync and sync.modifiedIndex, sync and sync.value)
+
+        # get failsafe topology
+        failsafe = nodes.get(self._FAILSAFE)
+        try:
+            failsafe = json.loads(failsafe.value) if failsafe else None
+        except Exception:
+            failsafe = None
+
+        return Cluster(initialize, config, leader, last_lsn, members, failover, sync, history, slots, failsafe)
+
+    def _cluster_loader(self, path):
+        result = self.retry(self._client.read, path, recursive=True)
+        nodes = {node.key[len(result.key):].lstrip('/'): node for node in result.leaves}
+        return self._cluster_from_nodes(result.etcd_index, nodes)
+
+    def _citus_cluster_loader(self, path):
+        clusters = defaultdict(dict)
+        result = self.retry(self._client.read, path, recursive=True)
+        for node in result.leaves:
+            key = node.key[len(result.key):].lstrip('/').split('/', 1)
+            if len(key) == 2 and citus_group_re.match(key[0]):
+                clusters[int(key[0])][key[1]] = node
+        return {group: self._cluster_from_nodes(result.etcd_index, nodes) for group, nodes in clusters.items()}
+
+    def _load_cluster(self, path, loader):
         cluster = None
         try:
-            result = self.retry(self._client.read, self.client_path(''), recursive=True)
+            cluster = loader(path)
-            nodes = {node.key[len(result.key):].lstrip('/'): node for node in result.leaves}
-
-            # get initialize flag
-            initialize = nodes.get(self._INITIALIZE)
-            initialize = initialize and initialize.value
-
-            # get global dynamic configuration
-            config = nodes.get(self._CONFIG)
-            config = config and ClusterConfig.from_node(config.modifiedIndex, config.value)
-
-            # get timeline history
-            history = nodes.get(self._HISTORY)
-            history = history and TimelineHistory.from_node(history.modifiedIndex, history.value)
-
-            # get last leader operation
-            last_leader_operation = nodes.get(self._LEADER_OPTIME)
-            last_leader_operation = 0 if last_leader_operation is None else int(last_leader_operation.value)
-
-            # get list of members
-            members = [self.member(n) for k, n in nodes.items() if k.startswith(self._MEMBERS) and k.count('/') == 1]
-
-            # get leader
-            leader = nodes.get(self._LEADER)
-            if leader:
-                member = Member(-1, leader.value, None, {})
-                member = ([m for m in members if m.name == leader.value] or [member])[0]
-                index = result.etcd_index if result.etcd_index > leader.modifiedIndex else leader.modifiedIndex + 1
-                leader = Leader(index, leader.ttl, member)
-
-            # failover key
-            failover = nodes.get(self._FAILOVER)
-            if failover:
-                failover = Failover.from_node(failover.modifiedIndex, failover.value)
-
-            # get synchronization state
-            sync = nodes.get(self._SYNC)
-            sync = SyncState.from_node(sync and sync.modifiedIndex, sync and sync.value)
-
-            cluster = Cluster(initialize, config, leader, last_leader_operation, members, failover, sync, history)
         except etcd.EtcdKeyNotFound:
-            cluster = Cluster(None, None, None, None, [], None, None, None)
+            cluster = Cluster(None, None, None, None, [], None, None, None, None, None)
         except Exception as e:
             self._handle_exception(e, 'get_cluster', raise_ex=EtcdError('Etcd is not responding properly'))
         self._has_failed = False
         return cluster
 
     @catch_etcd_errors
-    def touch_member(self, data, permanent=False):
+    def touch_member(self, data):
         data = json.dumps(data, separators=(',', ':'))
-        return self._client.set(self.member_path, data, None if permanent else self._ttl)
+        return self._client.set(self.member_path, data, self._ttl)
 
     @catch_etcd_errors
     def take_leader(self):
-        return self.retry(self._client.set, self.leader_path, self._name, self._ttl)
+        return self.retry(self._client.write, self.leader_path, self._name, ttl=self._ttl)
 
-    def attempt_to_acquire_leader(self, permanent=False):
+    def _do_attempt_to_acquire_leader(self):
         try:
-            return bool(self.retry(self._client.write,
+            return bool(self.retry(self._client.write, self.leader_path, self._name, ttl=self._ttl, prevExist=False))
-                                   self.leader_path,
-                                   self._name,
-                                   ttl=None if permanent else self._ttl,
-                                   prevExist=False))
         except etcd.EtcdAlreadyExist:
             logger.info('Could not take out TTL lock')
-        except (RetryFailedError, etcd.EtcdException):
+            return False
-            pass
+
-        return False
+    @catch_return_false_exception
+    def attempt_to_acquire_leader(self):
+        return self._run_and_handle_exceptions(self._do_attempt_to_acquire_leader, retry=None)
 
     @catch_etcd_errors
     def set_failover_value(self, value, index=None):
@@ -530,19 +722,34 @@ class Etcd(AbstractDCS):
         return self._client.write(self.config_path, value, prevIndex=index or 0)
 
     @catch_etcd_errors
-    def _write_leader_optime(self, last_operation):
+    def _write_leader_optime(self, last_lsn):
-        return self._client.set(self.leader_optime_path, last_operation)
+        return self._client.set(self.leader_optime_path, last_lsn)
 
     @catch_etcd_errors
+    def _write_status(self, value):
+        return self._client.set(self.status_path, value)
+
+    def _do_update_leader(self):
+        try:
+            return self.retry(self._client.write, self.leader_path, self._name,
+                              prevValue=self._name, ttl=self._ttl) is not None
+        except etcd.EtcdKeyNotFound:
+            return self._do_attempt_to_acquire_leader()
+
+    @catch_etcd_errors
+    def _write_failsafe(self, value):
+        return self._client.set(self.failsafe_path, value)
+
+    @catch_return_false_exception
     def _update_leader(self):
-        return self.retry(self._client.test_and_set, self.leader_path, self._name, self._name, self._ttl)
+        return self._run_and_handle_exceptions(self._do_update_leader, retry=None)
 
     @catch_etcd_errors
     def initialize(self, create_new=True, sysid=""):
         return self.retry(self._client.write, self.initialize_path, sysid, prevExist=(not create_new))
 
     @catch_etcd_errors
-    def delete_leader(self):
+    def _delete_leader(self):
         return self._client.delete(self.leader_path, prevValue=self._name)
 
     @catch_etcd_errors
@@ -575,13 +782,14 @@ class Etcd(AbstractDCS):
 
             while timeout >= 1:  # when timeout is too small urllib3 doesn't have enough time to connect
                 try:
-                    self._client.watch(self.leader_path, index=leader_index, timeout=timeout + 0.5)
+                    result = self._client.watch(self.leader_path, index=leader_index, timeout=timeout + 0.5)
                     self._has_failed = False
+                    if result.action == 'compareAndSwap':
+                        time.sleep(0.01)
                     # Synchronous work of all cluster members with etcd is less expensive
                     # than reestablishing http connection every time from every replica.
                     return True
                 except etcd.EtcdWatchTimedOut:
-                    self._client.http.clear()
                     self._has_failed = False
                     return False
                 except (etcd.EtcdEventIndexCleared, etcd.EtcdWatcherCleared):  # Watch failed
@@ -596,3 +804,6 @@ class Etcd(AbstractDCS):
             return super(Etcd, self).watch(None, timeout)
         finally:
             self.event.clear()
+
+
+etcd.EtcdError.error_exceptions[300] = EtcdRaftInternal

patroni/dcs/etcd3.py

@@ -0,0 +1,891 @@
+from __future__ import absolute_import
+import base64
+import etcd
+import json
+import logging
+import os
+import six
+import socket
+import sys
+import time
+import urllib3
+
+from collections import defaultdict
+from threading import Condition, Lock, Thread
+from urllib3.exceptions import ReadTimeoutError, ProtocolError
+
+from . import ClusterConfig, Cluster, Failover, Leader, Member, SyncState,\
+        TimelineHistory, ReturnFalseException, catch_return_false_exception, citus_group_re
+from .etcd import AbstractEtcdClientWithFailover, AbstractEtcd, catch_etcd_errors
+from ..exceptions import DCSError, PatroniException
+from ..utils import deep_compare, enable_keepalive, iter_response_objects, RetryFailedError, USER_AGENT
+
+logger = logging.getLogger(__name__)
+
+
+class Etcd3Error(DCSError):
+    pass
+
+
+class UnsupportedEtcdVersion(PatroniException):
+    pass
+
+
+# google.golang.org/grpc/codes
+GRPCCode = type('Enum', (), {'OK': 0, 'Canceled': 1, 'Unknown': 2, 'InvalidArgument': 3, 'DeadlineExceeded': 4,
+                             'NotFound': 5, 'AlreadyExists': 6, 'PermissionDenied': 7, 'ResourceExhausted': 8,
+                             'FailedPrecondition': 9, 'Aborted': 10, 'OutOfRange': 11, 'Unimplemented': 12,
+                             'Internal': 13, 'Unavailable': 14, 'DataLoss': 15, 'Unauthenticated': 16})
+GRPCcodeToText = {v: k for k, v in GRPCCode.__dict__.items() if not k.startswith('__') and isinstance(v, int)}
+
+
+class Etcd3Exception(etcd.EtcdException):
+    pass
+
+
+class Etcd3ClientError(Etcd3Exception):
+
+    def __init__(self, code=None, error=None, status=None):
+        if not hasattr(self, 'error'):
+            self.error = error and error.strip()
+        self.codeText = GRPCcodeToText.get(code)
+        self.status = status
+
+    def __repr__(self):
+        return "<{0} error: '{1}', code: {2}>".format(self.__class__.__name__, self.error, self.code)
+
+    __str__ = __repr__
+
+    def as_dict(self):
+        return {'error': self.error, 'code': self.code, 'codeText': self.codeText, 'status': self.status}
+
+    @classmethod
+    def get_subclasses(cls):
+        for subclass in cls.__subclasses__():
+            for subsubclass in subclass.get_subclasses():
+                yield subsubclass
+            yield subclass
+
+
+class Unknown(Etcd3ClientError):
+    code = GRPCCode.Unknown
+
+
+class InvalidArgument(Etcd3ClientError):
+    code = GRPCCode.InvalidArgument
+
+
+class DeadlineExceeded(Etcd3ClientError):
+    code = GRPCCode.DeadlineExceeded
+    error = "context deadline exceeded"
+
+
+class NotFound(Etcd3ClientError):
+    code = GRPCCode.NotFound
+
+
+class FailedPrecondition(Etcd3ClientError):
+    code = GRPCCode.FailedPrecondition
+
+
+class Unavailable(Etcd3ClientError):
+    code = GRPCCode.Unavailable
+
+
+# https://github.com/etcd-io/etcd/commits/main/api/v3rpc/rpctypes/error.go
+class LeaseNotFound(NotFound):
+    error = "etcdserver: requested lease not found"
+
+
+class UserEmpty(InvalidArgument):
+    error = "etcdserver: user name is empty"
+
+
+class AuthFailed(InvalidArgument):
+    error = "etcdserver: authentication failed, invalid user ID or password"
+
+
+class PermissionDenied(Etcd3ClientError):
+    code = GRPCCode.PermissionDenied
+    error = "etcdserver: permission denied"
+
+
+class AuthNotEnabled(FailedPrecondition):
+    error = "etcdserver: authentication is not enabled"
+
+
+class InvalidAuthToken(Etcd3ClientError):
+    code = GRPCCode.Unauthenticated
+    error = "etcdserver: invalid auth token"
+
+
+errStringToClientError = {s.error: s for s in Etcd3ClientError.get_subclasses() if hasattr(s, 'error')}
+errCodeToClientError = {s.code: s for s in Etcd3ClientError.__subclasses__()}
+
+
+def _raise_for_data(data, status_code=None):
+    try:
+        error = data.get('error') or data.get('Error')
+        if isinstance(error, dict):  # streaming response
+            status_code = error.get('http_code')
+            code = error['grpc_code']
+            error = error['message']
+        else:
+            code = data.get('code') or data.get('Code')
+    except Exception:
+        error = str(data)
+        code = GRPCCode.Unknown
+    err = errStringToClientError.get(error) or errCodeToClientError.get(code) or Unknown
+    raise err(code, error, status_code)
+
+
+def to_bytes(v):
+    return v if isinstance(v, bytes) else v.encode('utf-8')
+
+
+def prefix_range_end(v):
+    v = bytearray(to_bytes(v))
+    for i in range(len(v) - 1, -1, -1):
+        if v[i] < 0xff:
+            v[i] += 1
+            break
+    return bytes(v)
+
+
+def base64_encode(v):
+    return base64.b64encode(to_bytes(v)).decode('utf-8')
+
+
+def base64_decode(v):
+    return base64.b64decode(v).decode('utf-8')
+
+
+def build_range_request(key, range_end=None):
+    fields = {'key': base64_encode(key)}
+    if range_end:
+        fields['range_end'] = base64_encode(range_end)
+    return fields
+
+
+class Etcd3Client(AbstractEtcdClientWithFailover):
+
+    ERROR_CLS = Etcd3Error
+
+    def __init__(self, config, dns_resolver, cache_ttl=300):
+        self._token = None
+        self._cluster_version = None
+        self.version_prefix = '/v3beta'
+        super(Etcd3Client, self).__init__(config, dns_resolver, cache_ttl)
+
+        if six.PY2:  # pragma: no cover
+            # Old grpc-gateway sometimes sends double 'transfer-encoding: chunked' headers,
+            # what breaks the old (python2.7) httplib.HTTPConnection (it closes the socket).
+            def dedup_addheader(httpm, key, value):
+                prev = httpm.dict.get(key)
+                if prev is None:
+                    httpm.dict[key] = value
+                elif key != 'transfer-encoding' or prev != value:
+                    combined = ", ".join((prev, value))
+                    httpm.dict[key] = combined
+
+            import httplib
+            httplib.HTTPMessage.addheader = dedup_addheader
+
+        try:
+            self.authenticate()
+        except AuthFailed as e:
+            logger.fatal('Etcd3 authentication failed: %r', e)
+            sys.exit(1)
+
+    def _get_headers(self):
+        headers = urllib3.make_headers(user_agent=USER_AGENT)
+        if self._token and self._cluster_version >= (3, 3, 0):
+            headers['authorization'] = self._token
+        return headers
+
+    def _prepare_request(self, kwargs, params=None, method=None):
+        if params is not None:
+            kwargs['body'] = json.dumps(params)
+            kwargs['headers']['Content-Type'] = 'application/json'
+        return self.http.urlopen
+
+    @staticmethod
+    def _handle_server_response(response):
+        data = response.data
+        try:
+            data = data.decode('utf-8')
+            data = json.loads(data)
+        except (TypeError, ValueError, UnicodeError) as e:
+            if response.status < 400:
+                raise etcd.EtcdException('Server response was not valid JSON: %r' % e)
+        if response.status < 400:
+            return data
+        _raise_for_data(data, response.status)
+
+    def _ensure_version_prefix(self, base_uri, **kwargs):
+        if self.version_prefix != '/v3':
+            response = self.http.urlopen(self._MGET, base_uri + '/version', **kwargs)
+            response = self._handle_server_response(response)
+
+            server_version_str = response['etcdserver']
+            server_version = tuple(int(x) for x in server_version_str.split('.'))
+            cluster_version_str = response['etcdcluster']
+            self._cluster_version = tuple(int(x) for x in cluster_version_str.split('.'))
+
+            if self._cluster_version < (3, 0) or server_version < (3, 0, 4):
+                raise UnsupportedEtcdVersion('Detected Etcd version {0} is lower than 3.0.4'.format(server_version_str))
+
+            if self._cluster_version < (3, 3):
+                if self.version_prefix != '/v3alpha':
+                    if self._cluster_version < (3, 1):
+                        logger.warning('Detected Etcd version %s is lower than 3.1.0, watches are not supported',
+                                       cluster_version_str)
+                    if self.username and self.password:
+                        logger.warning('Detected Etcd version %s is lower than 3.3.0, authentication is not supported',
+                                       cluster_version_str)
+                    self.version_prefix = '/v3alpha'
+            elif self._cluster_version < (3, 4):
+                self.version_prefix = '/v3beta'
+            else:
+                self.version_prefix = '/v3'
+
+    def _prepare_get_members(self, etcd_nodes):
+        kwargs = self._prepare_common_parameters(etcd_nodes)
+        self._prepare_request(kwargs, {})
+        return kwargs
+
+    def _get_members(self, base_uri, **kwargs):
+        self._ensure_version_prefix(base_uri, **kwargs)
+        resp = self.http.urlopen(self._MPOST, base_uri + self.version_prefix + '/cluster/member/list', **kwargs)
+        members = self._handle_server_response(resp)['members']
+        return set(url for member in members for url in member.get('clientURLs', []))
+
+    def call_rpc(self, method, fields, retry=None):
+        fields['retry'] = retry
+        return self.api_execute(self.version_prefix + method, self._MPOST, fields)
+
+    def authenticate(self):
+        if self._use_proxies and self._cluster_version is None:
+            kwargs = self._prepare_common_parameters(1)
+            self._ensure_version_prefix(self._base_uri, **kwargs)
+        if self._cluster_version >= (3, 3) and self.username and self.password:
+            logger.info('Trying to authenticate on Etcd...')
+            old_token, self._token = self._token, None
+            try:
+                response = self.call_rpc('/auth/authenticate', {'name': self.username, 'password': self.password})
+            except AuthNotEnabled:
+                logger.info('Etcd authentication is not enabled')
+                self._token = None
+            except Exception:
+                self._token = old_token
+                raise
+            else:
+                self._token = response.get('token')
+            return old_token != self._token
+
+    def _handle_auth_errors(func):
+        def wrapper(self, *args, **kwargs):
+            def retry(ex):
+                if self.username and self.password:
+                    self.authenticate()
+                    return func(self, *args, **kwargs)
+                else:
+                    logger.fatal('Username or password not set, authentication is not possible')
+                    raise ex
+
+            try:
+                return func(self, *args, **kwargs)
+            except (UserEmpty, PermissionDenied) as e:  # no token provided
+                # PermissionDenied is raised on 3.0 and 3.1
+                if self._cluster_version < (3, 3) and (not isinstance(e, PermissionDenied)
+                                                       or self._cluster_version < (3, 2)):
+                    raise UnsupportedEtcdVersion('Authentication is required by Etcd cluster but not '
+                                                 'supported on version lower than 3.3.0. Cluster version: '
+                                                 '{0}'.format('.'.join(map(str, self._cluster_version))))
+                return retry(e)
+            except InvalidAuthToken as e:
+                logger.error('Invalid auth token: %s', self._token)
+                return retry(e)
+
+        return wrapper
+
+    @_handle_auth_errors
+    def range(self, key, range_end=None, retry=None):
+        params = build_range_request(key, range_end)
+        params['serializable'] = True  # For better performance. We can tolerate stale reads.
+        return self.call_rpc('/kv/range', params, retry)
+
+    def prefix(self, key, retry=None):
+        return self.range(key, prefix_range_end(key), retry)
+
+    @_handle_auth_errors
+    def lease_grant(self, ttl, retry=None):
+        return self.call_rpc('/lease/grant', {'TTL': ttl}, retry)['ID']
+
+    def lease_keepalive(self, ID, retry=None):
+        return self.call_rpc('/lease/keepalive', {'ID': ID}, retry).get('result', {}).get('TTL')
+
+    def txn(self, compare, success, retry=None):
+        return self.call_rpc('/kv/txn', {'compare': [compare], 'success': [success]}, retry).get('succeeded')
+
+    @_handle_auth_errors
+    def put(self, key, value, lease=None, create_revision=None, mod_revision=None, retry=None):
+        fields = {'key': base64_encode(key), 'value': base64_encode(value)}
+        if lease:
+            fields['lease'] = lease
+        if create_revision is not None:
+            compare = {'target': 'CREATE', 'create_revision': create_revision}
+        elif mod_revision is not None:
+            compare = {'target': 'MOD', 'mod_revision': mod_revision}
+        else:
+            return self.call_rpc('/kv/put', fields, retry)
+        compare['key'] = fields['key']
+        return self.txn(compare, {'request_put': fields}, retry)
+
+    @_handle_auth_errors
+    def deleterange(self, key, range_end=None, mod_revision=None, retry=None):
+        fields = build_range_request(key, range_end)
+        if mod_revision is None:
+            return self.call_rpc('/kv/deleterange', fields, retry)
+        compare = {'target': 'MOD', 'mod_revision': mod_revision, 'key': fields['key']}
+        return self.txn(compare, {'request_delete_range': fields}, retry)
+
+    def deleteprefix(self, key, retry=None):
+        return self.deleterange(key, prefix_range_end(key), retry=retry)
+
+    def watchrange(self, key, range_end=None, start_revision=None, filters=None, read_timeout=None):
+        """returns: response object"""
+        params = build_range_request(key, range_end)
+        if start_revision is not None:
+            params['start_revision'] = start_revision
+        params['filters'] = filters or []
+        kwargs = self._prepare_common_parameters(1, self.read_timeout)
+        request_executor = self._prepare_request(kwargs, {'create_request': params})
+        kwargs.update(timeout=urllib3.Timeout(connect=kwargs['timeout'], read=read_timeout), retries=0)
+        return request_executor(self._MPOST, self._base_uri + self.version_prefix + '/watch', **kwargs)
+
+    def watchprefix(self, key, start_revision=None, filters=None, read_timeout=None):
+        return self.watchrange(key, prefix_range_end(key), start_revision, filters, read_timeout)
+
+
+class KVCache(Thread):
+
+    def __init__(self, dcs, client):
+        Thread.__init__(self)
+        self.daemon = True
+        self._dcs = dcs
+        self._client = client
+        self.condition = Condition()
+        self._config_key = base64_encode(dcs.config_path)
+        self._leader_key = base64_encode(dcs.leader_path)
+        self._optime_key = base64_encode(dcs.leader_optime_path)
+        self._status_key = base64_encode(dcs.status_path)
+        self._name = base64_encode(dcs._name)
+        self._is_ready = False
+        self._response = None
+        self._response_lock = Lock()
+        self._object_cache = {}
+        self._object_cache_lock = Lock()
+        self.start()
+
+    def set(self, value, overwrite=False):
+        with self._object_cache_lock:
+            name = value['key']
+            old_value = self._object_cache.get(name)
+            ret = not old_value or int(old_value['mod_revision']) < int(value['mod_revision'])
+            if ret or overwrite and old_value['mod_revision'] == value['mod_revision']:
+                self._object_cache[name] = value
+        return ret, old_value
+
+    def delete(self, name, mod_revision):
+        with self._object_cache_lock:
+            old_value = self._object_cache.get(name)
+            ret = old_value and int(old_value['mod_revision']) < int(mod_revision)
+            if ret:
+                del self._object_cache[name]
+        return not old_value or ret, old_value
+
+    def copy(self):
+        with self._object_cache_lock:
+            return [v.copy() for v in self._object_cache.values()]
+
+    def get(self, name):
+        with self._object_cache_lock:
+            return self._object_cache.get(name)
+
+    def _process_event(self, event):
+        kv = event['kv']
+        key = kv['key']
+        if event.get('type') == 'DELETE':
+            success, old_value = self.delete(key, kv['mod_revision'])
+        else:
+            success, old_value = self.set(kv, True)
+
+        if success:
+            old_value = old_value and old_value.get('value')
+            new_value = kv.get('value')
+
+            value_changed = old_value != new_value and \
+                (key == self._leader_key or key in (self._optime_key, self._status_key) and new_value is not None or
+                 key == self._config_key and old_value is not None and new_value is not None)
+
+            if value_changed:
+                logger.debug('%s changed from %s to %s', key, old_value, new_value)
+
+            # We also want to wake up HA loop on replicas if leader optime (or status key) was updated
+            if value_changed and (key not in (self._optime_key, self._status_key) or
+                                  (self.get(self._leader_key) or {}).get('value') != self._name):
+                self._dcs.event.set()
+
+    def _process_message(self, message):
+        logger.debug('Received message: %s', message)
+        if 'error' in message:
+            _raise_for_data(message)
+        for event in message.get('result', {}).get('events', []):
+            self._process_event(event)
+
+    @staticmethod
+    def _finish_response(response):
+        try:
+            response.close()
+        finally:
+            response.release_conn()
+
+    def _do_watch(self, revision):
+        with self._response_lock:
+            self._response = None
+        # We do most of requests with timeouts. The only exception /watch requests to Etcd v3.
+        # In order to interrupt the /watch request we do socket.shutdown() from the main thread,
+        # which doesn't work on Windows. Therefore we want to use the last resort, `read_timeout`.
+        # Setting it to TTL will help to partially mitigate the problem.
+        # Setting it to lower value is not nice because for idling clusters it will increase
+        # the numbers of interrupts and reconnects.
+        read_timeout = self._dcs.ttl if os.name == 'nt' else None
+        response = self._client.watchprefix(self._dcs.cluster_prefix, revision, read_timeout=read_timeout)
+        with self._response_lock:
+            if self._response is None:
+                self._response = response
+
+        if not self._response:
+            return self._finish_response(response)
+
+        for message in iter_response_objects(response):
+            self._process_message(message)
+
+    def _build_cache(self):
+        result = self._dcs.retry(self._client.prefix, self._dcs.cluster_prefix)
+        with self._object_cache_lock:
+            self._object_cache = {node['key']: node for node in result.get('kvs', [])}
+        with self.condition:
+            self._is_ready = True
+            self.condition.notify()
+
+        try:
+            self._do_watch(result['header']['revision'])
+        except Exception as e:
+            # Following exceptions are expected on Windows because the /watch request  is done with `read_timeout`
+            if not (os.name == 'nt' and isinstance(e, (ReadTimeoutError, ProtocolError))):
+                logger.error('watchprefix failed: %r', e)
+        finally:
+            with self.condition:
+                self._is_ready = False
+            with self._response_lock:
+                response, self._response = self._response, None
+            if response:
+                self._finish_response(response)
+
+    def run(self):
+        while True:
+            try:
+                self._build_cache()
+            except Exception as e:
+                logger.error('KVCache.run %r', e)
+                time.sleep(1)
+
+    def kill_stream(self):
+        sock = None
+        with self._response_lock:
+            if self._response:
+                try:
+                    sock = self._response.connection.sock
+                except Exception:
+                    sock = None
+            else:
+                self._response = False
+        if sock:
+            try:
+                sock.shutdown(socket.SHUT_RDWR)
+                sock.close()
+            except Exception as e:
+                logger.debug('Error on socket.shutdown: %r', e)
+
+    def is_ready(self):
+        """Must be called only when holding the lock on `condition`"""
+        return self._is_ready
+
+
+class PatroniEtcd3Client(Etcd3Client):
+
+    def __init__(self, *args, **kwargs):
+        self._kv_cache = None
+        super(PatroniEtcd3Client, self).__init__(*args, **kwargs)
+
+    def configure(self, etcd3):
+        self._etcd3 = etcd3
+
+    def start_watcher(self):
+        if self._cluster_version >= (3, 1):
+            self._kv_cache = KVCache(self._etcd3, self)
+
+    def _restart_watcher(self):
+        if self._kv_cache:
+            self._kv_cache.kill_stream()
+
+    def set_base_uri(self, value):
+        super(PatroniEtcd3Client, self).set_base_uri(value)
+        self._restart_watcher()
+
+    def authenticate(self):
+        ret = super(PatroniEtcd3Client, self).authenticate()
+        if ret:
+            self._restart_watcher()
+        return ret
+
+    def _wait_cache(self, timeout):
+        stop_time = time.time() + timeout
+        while not self._kv_cache.is_ready():
+            timeout = stop_time - time.time()
+            if timeout <= 0:
+                raise RetryFailedError('Exceeded retry deadline')
+            self._kv_cache.condition.wait(timeout)
+
+    def get_cluster(self, path):
+        if self._kv_cache and path.startswith(self._etcd3.cluster_prefix):
+            with self._kv_cache.condition:
+                self._wait_cache(self._etcd3._retry.deadline)
+                ret = self._kv_cache.copy()
+        else:
+            ret = self._etcd3.retry(self.prefix, path).get('kvs', [])
+        for node in ret:
+            node.update({'key': base64_decode(node['key']),
+                         'value': base64_decode(node.get('value', '')),
+                         'lease': node.get('lease')})
+        return ret
+
+    def call_rpc(self, method, fields, retry=None):
+        ret = super(PatroniEtcd3Client, self).call_rpc(method, fields, retry)
+
+        if self._kv_cache:
+            value = delete = None
+            if method == '/kv/txn' and ret.get('succeeded'):
+                on_success = fields['success'][0]
+                value = on_success.get('request_put')
+                delete = on_success.get('request_delete_range')
+            elif method == '/kv/put' and ret:
+                value = fields
+            elif method == '/kv/deleterange' and ret:
+                delete = fields
+
+            if value:
+                value['mod_revision'] = ret['header']['revision']
+                self._kv_cache.set(value)
+            elif delete and 'range_end' not in delete:
+                self._kv_cache.delete(delete['key'], ret['header']['revision'])
+
+        return ret
+
+
+class Etcd3(AbstractEtcd):
+
+    def __init__(self, config):
+        super(Etcd3, self).__init__(config, PatroniEtcd3Client, (DeadlineExceeded, Unavailable, FailedPrecondition))
+        self.__do_not_watch = False
+        self._lease = None
+        self._last_lease_refresh = 0
+
+        self._client.configure(self)
+        if not self._ctl:
+            self._client.start_watcher()
+            self.create_lease()
+
+    def set_socket_options(self, sock, socket_options):
+        enable_keepalive(sock, self.ttl, int(self.loop_wait + self._retry.deadline))
+
+    def set_ttl(self, ttl):
+        self.__do_not_watch = super(Etcd3, self).set_ttl(ttl)
+        if self.__do_not_watch:
+            self._lease = None
+
+    def _do_refresh_lease(self, force=False, retry=None):
+        if not force and self._lease and self._last_lease_refresh + self._loop_wait > time.time():
+            return False
+
+        if self._lease and not self._client.lease_keepalive(self._lease, retry):
+            self._lease = None
+
+        ret = not self._lease
+        if ret:
+            self._lease = self._client.lease_grant(self._ttl, retry)
+
+        self._last_lease_refresh = time.time()
+        return ret
+
+    def refresh_lease(self):
+        try:
+            return self.retry(self._do_refresh_lease)
+        except (Etcd3ClientError, RetryFailedError):
+            logger.exception('refresh_lease')
+        raise Etcd3Error('Failed to keepalive/grant lease')
+
+    def create_lease(self):
+        while not self._lease:
+            try:
+                self.refresh_lease()
+            except Etcd3Error:
+                logger.info('waiting on etcd')
+                time.sleep(5)
+
+    @property
+    def cluster_prefix(self):
+        return self._base_path + '/' if self.is_citus_coordinator() else self.client_path('')
+
+    @staticmethod
+    def member(node):
+        return Member.from_node(node['mod_revision'], os.path.basename(node['key']), node['lease'], node['value'])
+
+    def _cluster_from_nodes(self, nodes):
+        # get initialize flag
+        initialize = nodes.get(self._INITIALIZE)
+        initialize = initialize and initialize['value']
+
+        # get global dynamic configuration
+        config = nodes.get(self._CONFIG)
+        config = config and ClusterConfig.from_node(config['mod_revision'], config['value'])
+
+        # get timeline history
+        history = nodes.get(self._HISTORY)
+        history = history and TimelineHistory.from_node(history['mod_revision'], history['value'])
+
+        # get last know leader lsn and slots
+        status = nodes.get(self._STATUS)
+        if status:
+            try:
+                status = json.loads(status['value'])
+                last_lsn = status.get(self._OPTIME)
+                slots = status.get('slots')
+            except Exception:
+                slots = last_lsn = None
+        else:
+            last_lsn = nodes.get(self._LEADER_OPTIME)
+            last_lsn = last_lsn and last_lsn['value']
+            slots = None
+
+        try:
+            last_lsn = int(last_lsn)
+        except Exception:
+            last_lsn = 0
+
+        # get list of members
+        members = [self.member(n) for k, n in nodes.items() if k.startswith(self._MEMBERS) and k.count('/') == 1]
+
+        # get leader
+        leader = nodes.get(self._LEADER)
+        if not self._ctl and leader and leader['value'] == self._name and self._lease != leader.get('lease'):
+            logger.warning('I am the leader but not owner of the lease')
+
+        if leader:
+            member = Member(-1, leader['value'], None, {})
+            member = ([m for m in members if m.name == leader['value']] or [member])[0]
+            leader = Leader(leader['mod_revision'], leader['lease'], member)
+
+        # failover key
+        failover = nodes.get(self._FAILOVER)
+        if failover:
+            failover = Failover.from_node(failover['mod_revision'], failover['value'])
+
+        # get synchronization state
+        sync = nodes.get(self._SYNC)
+        sync = SyncState.from_node(sync and sync['mod_revision'], sync and sync['value'])
+
+        # get failsafe topology
+        failsafe = nodes.get(self._FAILSAFE)
+        try:
+            failsafe = json.loads(failsafe['value']) if failsafe else None
+        except Exception:
+            failsafe = None
+
+        return Cluster(initialize, config, leader, last_lsn, members, failover, sync, history, slots, failsafe)
+
+    def _cluster_loader(self, path):
+        nodes = {node['key'][len(path):]: node
+                 for node in self._client.get_cluster(path)
+                 if node['key'].startswith(path)}
+        return self._cluster_from_nodes(nodes)
+
+    def _citus_cluster_loader(self, path):
+        clusters = defaultdict(dict)
+        path = self._base_path + '/'
+        for node in self._client.get_cluster(path):
+            key = node['key'][len(path):].split('/', 1)
+            if len(key) == 2 and citus_group_re.match(key[0]):
+                clusters[int(key[0])][key[1]] = node
+        return {group: self._cluster_from_nodes(nodes) for group, nodes in clusters.items()}
+
+    def _load_cluster(self, path, loader):
+        cluster = None
+        try:
+            cluster = loader(path)
+        except UnsupportedEtcdVersion:
+            raise
+        except Exception as e:
+            self._handle_exception(e, 'get_cluster', raise_ex=Etcd3Error('Etcd is not responding properly'))
+        self._has_failed = False
+        return cluster
+
+    @catch_etcd_errors
+    def touch_member(self, data):
+        try:
+            self.refresh_lease()
+        except Etcd3Error:
+            return False
+
+        cluster = self.cluster
+        member = cluster and cluster.get_member(self._name, fallback_to_leader=False)
+
+        if member and member.session == self._lease and deep_compare(data, member.data):
+            return True
+
+        data = json.dumps(data, separators=(',', ':'))
+        try:
+            return self._client.put(self.member_path, data, self._lease)
+        except LeaseNotFound:
+            self._lease = None
+            logger.error('Our lease disappeared from Etcd, can not "touch_member"')
+
+    @catch_etcd_errors
+    def take_leader(self):
+        return self.retry(self._client.put, self.leader_path, self._name, self._lease)
+
+    def _do_attempt_to_acquire_leader(self, retry):
+        def _retry(*args, **kwargs):
+            kwargs['retry'] = retry
+            return retry(*args, **kwargs)
+
+        try:
+            return _retry(self._client.put, self.leader_path, self._name, self._lease, 0)
+        except LeaseNotFound:
+            logger.error('Our lease disappeared from Etcd. Will try to get a new one and retry attempt')
+            self._lease = None
+            retry.deadline = retry.stoptime - time.time()
+
+            _retry(self._do_refresh_lease)
+
+            retry.deadline = retry.stoptime - time.time()
+            if retry.deadline < 1:
+                raise Etcd3Error('_do_attempt_to_acquire_leader timeout')
+
+            return _retry(self._client.put, self.leader_path, self._name, self._lease, 0)
+
+    @catch_return_false_exception
+    def attempt_to_acquire_leader(self):
+        retry = self._retry.copy()
+
+        def _retry(*args, **kwargs):
+            kwargs['retry'] = retry
+            return retry(*args, **kwargs)
+
+        self._run_and_handle_exceptions(self._do_refresh_lease, retry=_retry)
+
+        retry.deadline = retry.stoptime - time.time()
+        if retry.deadline < 1:
+            raise Etcd3Error('attempt_to_acquire_leader timeout')
+
+        ret = self._run_and_handle_exceptions(self._do_attempt_to_acquire_leader, retry, retry=None)
+        if not ret:
+            logger.info('Could not take out TTL lock')
+        return ret
+
+    @catch_etcd_errors
+    def set_failover_value(self, value, index=None):
+        return self._client.put(self.failover_path, value, mod_revision=index)
+
+    @catch_etcd_errors
+    def set_config_value(self, value, index=None):
+        return self._client.put(self.config_path, value, mod_revision=index)
+
+    @catch_etcd_errors
+    def _write_leader_optime(self, last_lsn):
+        return self._client.put(self.leader_optime_path, last_lsn)
+
+    @catch_etcd_errors
+    def _write_status(self, value):
+        return self._client.put(self.status_path, value)
+
+    @catch_etcd_errors
+    def _write_failsafe(self, value):
+        return self._client.put(self.failsafe_path, value)
+
+    @catch_return_false_exception
+    def _update_leader(self):
+        retry = self._retry.copy()
+
+        def _retry(*args, **kwargs):
+            kwargs['retry'] = retry
+            return retry(*args, **kwargs)
+
+        self._run_and_handle_exceptions(self._do_refresh_lease, True, retry=_retry)
+
+        if self._lease:
+            cluster = self.cluster
+            leader_lease = cluster and isinstance(cluster.leader, Leader) and cluster.leader.session
+            if leader_lease != self._lease:
+                retry.deadline = retry.stoptime - time.time()
+                if retry.deadline < 1:
+                    raise Etcd3Error('update_leader timeout')
+
+                try:
+                    self._run_and_handle_exceptions(self._client.put, self.leader_path,
+                                                    self._name, self._lease, retry=_retry)
+                except ReturnFalseException:
+                    pass
+        return bool(self._lease)
+
+    @catch_etcd_errors
+    def initialize(self, create_new=True, sysid=""):
+        return self.retry(self._client.put, self.initialize_path, sysid, None, 0 if create_new else None)
+
+    @catch_etcd_errors
+    def _delete_leader(self):
+        cluster = self.cluster
+        if cluster and isinstance(cluster.leader, Leader) and cluster.leader.name == self._name:
+            return self._client.deleterange(self.leader_path, mod_revision=cluster.leader.index)
+
+    @catch_etcd_errors
+    def cancel_initialization(self):
+        return self.retry(self._client.deleterange, self.initialize_path)
+
+    @catch_etcd_errors
+    def delete_cluster(self):
+        return self.retry(self._client.deleteprefix, self.client_path(''))
+
+    @catch_etcd_errors
+    def set_history_value(self, value):
+        return self._client.put(self.history_path, value)
+
+    @catch_etcd_errors
+    def set_sync_state_value(self, value, index=None):
+        return self.retry(self._client.put, self.sync_path, value, mod_revision=index)
+
+    @catch_etcd_errors
+    def delete_sync_state(self, index=None):
+        return self.retry(self._client.deleterange, self.sync_path, mod_revision=index)
+
+    def watch(self, leader_index, timeout):
+        if self.__do_not_watch:
+            self.__do_not_watch = False
+            return True
+
+        try:
+            return super(Etcd3, self).watch(None, timeout)
+        finally:
+            self.event.clear()

patroni/dcs/exhibitor.py

@@ -1,11 +1,11 @@
+import json
 import logging
 import random
-import requests
 import time
 
 from patroni.dcs.zookeeper import ZooKeeper
+from patroni.request import get as requests_get
 from patroni.utils import uri
-from requests.exceptions import RequestException
 
 logger = logging.getLogger(__name__)
 
@@ -19,7 +19,7 @@ class ExhibitorEnsembleProvider(object):
         self._uri_path = uri_path
         self._poll_interval = poll_interval
         self._exhibitors = hosts
-        self._master_exhibitors = hosts
+        self._boot_exhibitors = hosts
         self._zookeeper_hosts = ''
         self._next_poll = None
         while not self.poll():
@@ -32,7 +32,7 @@ class ExhibitorEnsembleProvider(object):
 
         json = self._query_exhibitors(self._exhibitors)
         if not json:
-            json = self._query_exhibitors(self._master_exhibitors)
+            json = self._query_exhibitors(self._boot_exhibitors)
 
         if isinstance(json, dict) and 'servers' in json and 'port' in json:
             self._next_poll = time.time() + self._poll_interval
@@ -48,10 +48,10 @@ class ExhibitorEnsembleProvider(object):
         random.shuffle(exhibitors)
         for host in exhibitors:
             try:
-                response = requests.get(uri('http', (host, self._exhibitor_port), self._uri_path), timeout=self.TIMEOUT)
+                response = requests_get(uri('http', (host, self._exhibitor_port), self._uri_path), timeout=self.TIMEOUT)
-                return response.json()
+                return json.loads(response.data.decode('utf-8'))
-            except RequestException:
+            except Exception:
-                pass
+                logging.debug('Request to %s failed', host)
         return None
 
     @property
@@ -68,7 +68,7 @@ class Exhibitor(ZooKeeper):
         config['hosts'] = self._ensemble_provider.zookeeper_hosts
         super(Exhibitor, self).__init__(config)
 
-    def _load_cluster(self):
+    def _load_cluster(self, path, loader):
         if self._ensemble_provider.poll():
             self._client.set_hosts(self._ensemble_provider.zookeeper_hosts)
-        return super(Exhibitor, self)._load_cluster()
+        return super(Exhibitor, self)._load_cluster(path, loader)

patroni/dcs/raft.py

@@ -0,0 +1,459 @@
+import json
+import logging
+import os
+import threading
+import time
+
+from collections import defaultdict
+from pysyncobj import SyncObj, SyncObjConf, replicated, FAIL_REASON
+from pysyncobj.dns_resolver import globalDnsResolver
+from pysyncobj.node import TCPNode
+from pysyncobj.transport import TCPTransport, CONNECTION_STATE
+from pysyncobj.utility import TcpUtility
+
+from . import AbstractDCS, ClusterConfig, Cluster, Failover, Leader, Member, SyncState, TimelineHistory, citus_group_re
+from ..exceptions import DCSError
+from ..utils import validate_directory
+
+logger = logging.getLogger(__name__)
+
+
+class RaftError(DCSError):
+    pass
+
+
+class _TCPTransport(TCPTransport):
+
+    def __init__(self, syncObj, selfNode, otherNodes):
+        super(_TCPTransport, self).__init__(syncObj, selfNode, otherNodes)
+        self.setOnUtilityMessageCallback('members', syncObj.getMembers)
+
+    def _connectIfNecessarySingle(self, node):
+        try:
+            return super(_TCPTransport, self)._connectIfNecessarySingle(node)
+        except Exception as e:
+            logger.debug('Connection to %s failed: %r', node, e)
+            return False
+
+
+def resolve_host(self):
+    return globalDnsResolver().resolve(self.host)
+
+
+setattr(TCPNode, 'ip', property(resolve_host))
+
+
+class SyncObjUtility(object):
+
+    def __init__(self, otherNodes, conf, retry_timeout=10):
+        self._nodes = otherNodes
+        self._utility = TcpUtility(conf.password, retry_timeout/max(1, len(otherNodes)))
+
+    def executeCommand(self, command):
+        try:
+            return self._utility.executeCommand(self.__node, command)
+        except Exception:
+            return None
+
+    def getMembers(self):
+        for self.__node in self._nodes:
+            response = self.executeCommand(['members'])
+            if response:
+                return [member['addr'] for member in response]
+
+
+class DynMemberSyncObj(SyncObj):
+
+    def __init__(self, selfAddress, partnerAddrs, conf, retry_timeout=10):
+        self.__early_apply_local_log = selfAddress is not None
+        self.applied_local_log = False
+
+        utility = SyncObjUtility(partnerAddrs, conf, retry_timeout)
+        members = utility.getMembers()
+        add_self = members and selfAddress not in members
+
+        partnerAddrs = [member for member in (members or partnerAddrs) if member != selfAddress]
+
+        super(DynMemberSyncObj, self).__init__(selfAddress, partnerAddrs, conf, transportClass=_TCPTransport)
+
+        if add_self:
+            thread = threading.Thread(target=utility.executeCommand, args=(['add', selfAddress],))
+            thread.daemon = True
+            thread.start()
+
+    def getMembers(self, args, callback):
+        callback([{'addr': node.id, 'leader': node == self._getLeader(), 'status': CONNECTION_STATE.CONNECTED
+                   if self.isNodeConnected(node) else CONNECTION_STATE.DISCONNECTED} for node in self.otherNodes] +
+                 [{'addr': self.selfNode.id, 'leader': self._isLeader(), 'status': CONNECTION_STATE.CONNECTED}], None)
+
+    def _onTick(self, timeToWait=0.0):
+        super(DynMemberSyncObj, self)._onTick(timeToWait)
+
+        # The SyncObj calls onReady callback only when cluster got the leader and is ready for writes.
+        # In some cases for us it is safe to "signal" the Raft object when the local log is fully applied.
+        # We are using the `applied_local_log` property for that, but not calling the callback function.
+        if self.__early_apply_local_log and not self.applied_local_log and self.raftLastApplied == self.raftCommitIndex:
+            self.applied_local_log = True
+
+
+class KVStoreTTL(DynMemberSyncObj):
+
+    def __init__(self, on_ready, on_set, on_delete, **config):
+        self.__thread = None
+        self.__on_set = on_set
+        self.__on_delete = on_delete
+        self.__limb = {}
+        self.set_retry_timeout(int(config.get('retry_timeout') or 10))
+
+        self_addr = config.get('self_addr')
+        partner_addrs = set(config.get('partner_addrs', []))
+        if config.get('patronictl'):
+            if self_addr:
+                partner_addrs.add(self_addr)
+            self_addr = None
+
+        # Create raft data_dir if necessary
+        raft_data_dir = config.get('data_dir', '')
+        if raft_data_dir != '':
+            validate_directory(raft_data_dir)
+
+        file_template = (self_addr or '')
+        file_template = file_template.replace(':', '_') if os.name == 'nt' else file_template
+        file_template = os.path.join(raft_data_dir, file_template)
+        conf = SyncObjConf(password=config.get('password'), autoTick=False, appendEntriesUseBatch=False,
+                           bindAddress=config.get('bind_addr'), dnsFailCacheTime=(config.get('loop_wait') or 10),
+                           dnsCacheTime=(config.get('ttl') or 30), commandsWaitLeader=config.get('commandsWaitLeader'),
+                           fullDumpFile=(file_template + '.dump' if self_addr else None),
+                           journalFile=(file_template + '.journal' if self_addr else None),
+                           onReady=on_ready, dynamicMembershipChange=True)
+
+        super(KVStoreTTL, self).__init__(self_addr, partner_addrs, conf, self.__retry_timeout)
+        self.__data = {}
+
+    @staticmethod
+    def __check_requirements(old_value, **kwargs):
+        return ('prevExist' not in kwargs or bool(kwargs['prevExist']) == bool(old_value)) and \
+            ('prevValue' not in kwargs or old_value and old_value['value'] == kwargs['prevValue']) and \
+            (not kwargs.get('prevIndex') or old_value and old_value['index'] == kwargs['prevIndex'])
+
+    def set_retry_timeout(self, retry_timeout):
+        self.__retry_timeout = retry_timeout
+
+    def retry(self, func, *args, **kwargs):
+        event = threading.Event()
+        ret = {'result': None, 'error': -1}
+
+        def callback(result, error):
+            ret.update(result=result, error=error)
+            event.set()
+
+        kwargs['callback'] = callback
+        timeout = kwargs.pop('timeout', None) or self.__retry_timeout
+        deadline = timeout and time.time() + timeout
+
+        while True:
+            event.clear()
+            func(*args, **kwargs)
+            event.wait(timeout)
+            if ret['error'] == FAIL_REASON.SUCCESS:
+                return ret['result']
+            elif ret['error'] == FAIL_REASON.REQUEST_DENIED:
+                break
+            elif deadline:
+                timeout = deadline - time.time()
+                if timeout <= 0:
+                    raise RaftError('timeout')
+            time.sleep(1)
+        return False
+
+    @replicated
+    def _set(self, key, value, **kwargs):
+        old_value = self.__data.get(key, {})
+        if not self.__check_requirements(old_value, **kwargs):
+            return False
+
+        if old_value and old_value['created'] != value['created']:
+            value['created'] = value['updated']
+        value['index'] = self.raftLastApplied + 1
+
+        self.__data[key] = value
+        if self.__on_set:
+            self.__on_set(key, value)
+        return True
+
+    def set(self, key, value, ttl=None, handle_raft_error=True, **kwargs):
+        old_value = self.__data.get(key, {})
+        if not self.__check_requirements(old_value, **kwargs):
+            return False
+
+        value = {'value': value, 'updated': time.time()}
+        value['created'] = old_value.get('created', value['updated'])
+        if ttl:
+            value['expire'] = value['updated'] + ttl
+        try:
+            return self.retry(self._set, key, value, **kwargs)
+        except RaftError:
+            if not handle_raft_error:
+                raise
+            return False
+
+    def __pop(self, key):
+        self.__data.pop(key)
+        if self.__on_delete:
+            self.__on_delete(key)
+
+    @replicated
+    def _delete(self, key, recursive=False, **kwargs):
+        if recursive:
+            for k in list(self.__data.keys()):
+                if k.startswith(key):
+                    self.__pop(k)
+        elif not self.__check_requirements(self.__data.get(key, {}), **kwargs):
+            return False
+        else:
+            self.__pop(key)
+        return True
+
+    def delete(self, key, recursive=False, **kwargs):
+        if not recursive and not self.__check_requirements(self.__data.get(key, {}), **kwargs):
+            return False
+        try:
+            return self.retry(self._delete, key, recursive=recursive, **kwargs)
+        except RaftError:
+            return False
+
+    @staticmethod
+    def __values_match(old, new):
+        return all(old.get(n) == new.get(n) for n in ('created', 'updated', 'expire', 'value'))
+
+    @replicated
+    def _expire(self, key, value, callback=None):
+        current = self.__data.get(key)
+        if current and self.__values_match(current, value):
+            self.__pop(key)
+
+    def __expire_keys(self):
+        for key, value in self.__data.items():
+            if value and 'expire' in value and value['expire'] <= time.time() and \
+                    not (key in self.__limb and self.__values_match(self.__limb[key], value)):
+                self.__limb[key] = value
+
+                def callback(*args):
+                    if key in self.__limb and self.__values_match(self.__limb[key], value):
+                        self.__limb.pop(key)
+                self._expire(key, value, callback=callback)
+
+    def get(self, key, recursive=False):
+        if not recursive:
+            return self.__data.get(key)
+        return {k: v for k, v in self.__data.items() if k.startswith(key)}
+
+    def _onTick(self, timeToWait=0.0):
+        super(KVStoreTTL, self)._onTick(timeToWait)
+
+        if self._isLeader():
+            self.__expire_keys()
+        else:
+            self.__limb.clear()
+
+    def _autoTickThread(self):
+        self.__destroying = False
+        while not self.__destroying:
+            self.doTick(self.conf.autoTickPeriod)
+
+    def startAutoTick(self):
+        self.__thread = threading.Thread(target=self._autoTickThread)
+        self.__thread.daemon = True
+        self.__thread.start()
+
+    def destroy(self):
+        if self.__thread:
+            self.__destroying = True
+            self.__thread.join()
+        super(KVStoreTTL, self).destroy()
+
+
+class Raft(AbstractDCS):
+
+    def __init__(self, config):
+        super(Raft, self).__init__(config)
+        self._ttl = int(config.get('ttl') or 30)
+
+        ready_event = threading.Event()
+        self._sync_obj = KVStoreTTL(ready_event.set, self._on_set, self._on_delete, commandsWaitLeader=False, **config)
+        self._sync_obj.startAutoTick()
+
+        while True:
+            ready_event.wait(5)
+            if ready_event.is_set() or self._sync_obj.applied_local_log:
+                break
+            else:
+                logger.info('waiting on raft')
+
+    def _on_set(self, key, value):
+        leader = (self._sync_obj.get(self.leader_path) or {}).get('value')
+        if key == value['created'] == value['updated'] and \
+                (key.startswith(self.members_path) or key == self.leader_path and leader != self._name) or \
+                key in (self.leader_optime_path, self.status_path) and leader != self._name or \
+                key in (self.config_path, self.sync_path):
+            self.event.set()
+
+    def _on_delete(self, key):
+        if key == self.leader_path:
+            self.event.set()
+
+    def set_ttl(self, ttl):
+        self._ttl = ttl
+
+    @property
+    def ttl(self):
+        return self._ttl
+
+    def set_retry_timeout(self, retry_timeout):
+        self._sync_obj.set_retry_timeout(retry_timeout)
+
+    def reload_config(self, config):
+        super(Raft, self).reload_config(config)
+        globalDnsResolver().setTimeouts(self.ttl, self.loop_wait)
+
+    @staticmethod
+    def member(key, value):
+        return Member.from_node(value['index'], os.path.basename(key), None, value['value'])
+
+    def _cluster_from_nodes(self, nodes):
+        # get initialize flag
+        initialize = nodes.get(self._INITIALIZE)
+        initialize = initialize and initialize['value']
+
+        # get global dynamic configuration
+        config = nodes.get(self._CONFIG)
+        config = config and ClusterConfig.from_node(config['index'], config['value'])
+
+        # get timeline history
+        history = nodes.get(self._HISTORY)
+        history = history and TimelineHistory.from_node(history['index'], history['value'])
+
+        # get last know leader lsn and slots
+        status = nodes.get(self._STATUS)
+        if status:
+            try:
+                status = json.loads(status['value'])
+                last_lsn = status.get(self._OPTIME)
+                slots = status.get('slots')
+            except Exception:
+                slots = last_lsn = None
+        else:
+            last_lsn = nodes.get(self._LEADER_OPTIME)
+            last_lsn = last_lsn and last_lsn['value']
+            slots = None
+
+        try:
+            last_lsn = int(last_lsn)
+        except Exception:
+            last_lsn = 0
+
+        # get list of members
+        members = [self.member(k, n) for k, n in nodes.items() if k.startswith(self._MEMBERS) and k.count('/') == 1]
+
+        # get leader
+        leader = nodes.get(self._LEADER)
+        if leader:
+            member = Member(-1, leader['value'], None, {})
+            member = ([m for m in members if m.name == leader['value']] or [member])[0]
+            leader = Leader(leader['index'], None, member)
+
+        # failover key
+        failover = nodes.get(self._FAILOVER)
+        if failover:
+            failover = Failover.from_node(failover['index'], failover['value'])
+
+        # get synchronization state
+        sync = nodes.get(self._SYNC)
+        sync = SyncState.from_node(sync and sync['index'], sync and sync['value'])
+
+        # get failsafe topology
+        failsafe = nodes.get(self._FAILSAFE)
+        try:
+            failsafe = json.loads(failsafe['value']) if failsafe else None
+        except Exception:
+            failsafe = None
+
+        return Cluster(initialize, config, leader, last_lsn, members, failover, sync, history, slots, failsafe)
+
+    def _cluster_loader(self, path):
+        response = self._sync_obj.get(path, recursive=True)
+        if not response:
+            return Cluster(None, None, None, None, [], None, None, None, None, None)
+        nodes = {key[len(path):]: value for key, value in response.items()}
+        return self._cluster_from_nodes(nodes)
+
+    def _citus_cluster_loader(self, path):
+        clusters = defaultdict(dict)
+        response = self._sync_obj.get(path, recursive=True)
+        for key, value in response.items():
+            key = key[len(path):].split('/', 1)
+            if len(key) == 2 and citus_group_re.match(key[0]):
+                clusters[int(key[0])][key[1]] = value
+        return {group: self._cluster_from_nodes(nodes) for group, nodes in clusters.items()}
+
+    def _load_cluster(self, path, loader):
+        return loader(path)
+
+    def _write_leader_optime(self, last_lsn):
+        return self._sync_obj.set(self.leader_optime_path, last_lsn, timeout=1)
+
+    def _write_status(self, value):
+        return self._sync_obj.set(self.status_path, value, timeout=1)
+
+    def _write_failsafe(self, value):
+        return self._sync_obj.set(self.failsafe_path, value, timeout=1)
+
+    def _update_leader(self):
+        ret = self._sync_obj.set(self.leader_path, self._name, ttl=self._ttl,
+                                 handle_raft_error=False, prevValue=self._name)
+        if not ret and self._sync_obj.get(self.leader_path) is None:
+            ret = self.attempt_to_acquire_leader()
+        return ret
+
+    def attempt_to_acquire_leader(self):
+        return self._sync_obj.set(self.leader_path, self._name, ttl=self._ttl, handle_raft_error=False, prevExist=False)
+
+    def set_failover_value(self, value, index=None):
+        return self._sync_obj.set(self.failover_path, value, prevIndex=index)
+
+    def set_config_value(self, value, index=None):
+        return self._sync_obj.set(self.config_path, value, prevIndex=index)
+
+    def touch_member(self, data):
+        data = json.dumps(data, separators=(',', ':'))
+        return self._sync_obj.set(self.member_path, data, self._ttl, timeout=2)
+
+    def take_leader(self):
+        return self._sync_obj.set(self.leader_path, self._name, ttl=self._ttl)
+
+    def initialize(self, create_new=True, sysid=''):
+        return self._sync_obj.set(self.initialize_path, sysid, prevExist=(not create_new))
+
+    def _delete_leader(self):
+        return self._sync_obj.delete(self.leader_path, prevValue=self._name, timeout=1)
+
+    def cancel_initialization(self):
+        return self._sync_obj.delete(self.initialize_path)
+
+    def delete_cluster(self):
+        return self._sync_obj.delete(self.client_path(''), recursive=True)
+
+    def set_history_value(self, value):
+        return self._sync_obj.set(self.history_path, value)
+
+    def set_sync_state_value(self, value, index=None):
+        return self._sync_obj.set(self.sync_path, value, prevIndex=index)
+
+    def delete_sync_state(self, index=None):
+        return self._sync_obj.delete(self.sync_path, prevIndex=index)
+
+    def watch(self, leader_index, timeout):
+        try:
+            return super(Raft, self).watch(leader_index, timeout)
+        finally:
+            self.event.clear()

patroni/dcs/zookeeper.py

@@ -1,14 +1,19 @@
 import json
 import logging
 import select
+import six
 import time
 
 from kazoo.client import KazooClient, KazooState, KazooRetry
-from kazoo.exceptions import NoNodeError, NodeExistsError
+from kazoo.exceptions import ConnectionClosedError, NoNodeError, NodeExistsError, SessionExpiredError
 from kazoo.handlers.threading import SequentialThreadingHandler
-from patroni.dcs import AbstractDCS, ClusterConfig, Cluster, Failover, Leader, Member, SyncState, TimelineHistory
+from kazoo.protocol.states import KeeperState
-from patroni.exceptions import DCSError
+from kazoo.retry import RetryFailedError
-from patroni.utils import deep_compare
+from kazoo.security import make_acl
+
+from . import AbstractDCS, ClusterConfig, Cluster, Failover, Leader, Member, SyncState, TimelineHistory, citus_group_re
+from ..exceptions import DCSError
+from ..utils import deep_compare
 
 logger = logging.getLogger(__name__)
 
@@ -47,11 +52,35 @@ class PatroniSequentialThreadingHandler(SequentialThreadingHandler):
         return super(PatroniSequentialThreadingHandler, self).create_connection(*args, **kwargs)
 
     def select(self, *args, **kwargs):
-        """Python3 raises `ValueError` if socket is closed, because fd == -1"""
+        """
+        Python 3.XY may raise following exceptions if select/poll are called with an invalid socket:
+        - `ValueError`: because fd == -1
+        - `TypeError`: Invalid file descriptor: -1 (starting from kazoo 2.9)
+        Python 2.7 may raise the `IOError` instead of `socket.error` (starting from kazoo 2.9)
+
+        When it is appropriate we map these exceptions to `socket.error`.
+        """
+
         try:
             return super(PatroniSequentialThreadingHandler, self).select(*args, **kwargs)
-        except ValueError as e:
+        except IOError as e:
-            raise select.error(9, str(e))
+            raise (select.error(e.errno, e.strerror) if six.PY2 else e)
+        except (TypeError, ValueError) as e:
+            raise (e if six.PY2 and isinstance(e, TypeError) else select.error(9, str(e)))
+
+
+class PatroniKazooClient(KazooClient):
+
+    def _call(self, request, async_object):
+        # Before kazoo==2.7.0 it wasn't possible to send requests to zookeeper if
+        # the connection is in the SUSPENDED state and Patroni was strongly relying on it.
+        # The https://github.com/python-zk/kazoo/pull/588 changed it, and now such requests are queued.
+        # We override the `_call()` method in order to keep the old behavior.
+
+        if self._state == KeeperState.CONNECTING:
+            async_object.set_exception(SessionExpiredError())
+            return False
+        return super(PatroniKazooClient, self)._call(request, async_object)
 
 
 class ZooKeeper(AbstractDCS):
@@ -63,20 +92,39 @@ class ZooKeeper(AbstractDCS):
         if isinstance(hosts, list):
             hosts = ','.join(hosts)
 
-        self._client = KazooClient(hosts, handler=PatroniSequentialThreadingHandler(config['retry_timeout']),
+        mapping = {'use_ssl': 'use_ssl', 'verify': 'verify_certs', 'cacert': 'ca',
-                                   timeout=config['ttl'], connection_retry=KazooRetry(max_delay=1, max_tries=-1,
+                   'cert': 'certfile', 'key': 'keyfile', 'key_password': 'keyfile_password'}
-                                   sleep_func=time.sleep), command_retry=KazooRetry(deadline=config['retry_timeout'],
+        kwargs = {v: config[k] for k, v in mapping.items() if k in config}
-                                   max_delay=1, max_tries=-1, sleep_func=time.sleep))
+
+        if 'set_acls' in config:
+            kwargs['default_acl'] = []
+            for principal, permissions in config['set_acls'].items():
+                normalizedPermissions = [p.upper() for p in permissions]
+                kwargs['default_acl'].append(make_acl(scheme='x509',
+                                                      credential=principal,
+                                                      read='READ' in normalizedPermissions,
+                                                      write='WRITE' in normalizedPermissions,
+                                                      create='CREATE' in normalizedPermissions,
+                                                      delete='DELETE' in normalizedPermissions,
+                                                      admin='ADMIN' in normalizedPermissions,
+                                                      all='ALL' in normalizedPermissions))
+
+        self._client = PatroniKazooClient(hosts, handler=PatroniSequentialThreadingHandler(config['retry_timeout']),
+                                          timeout=config['ttl'], connection_retry=KazooRetry(max_delay=1, max_tries=-1,
+                                          sleep_func=time.sleep), command_retry=KazooRetry(max_delay=1, max_tries=-1,
+                                          deadline=config['retry_timeout'], sleep_func=time.sleep), **kwargs)
         self._client.add_listener(self.session_listener)
 
         self._fetch_cluster = True
+        self._fetch_status = True
+        self.__last_member_data = None
 
         self._orig_kazoo_connect = self._client._connection._connect
         self._client._connection._connect = self._kazoo_connect
 
         self._client.start()
 
-    def _kazoo_connect(self, host, port):
+    def _kazoo_connect(self, *args):
         """Kazoo is using Ping's to determine health of connection to zookeeper. If there is no
         response on Ping after Ping interval (1/2 from read_timeout) it will consider current
         connection dead and try to connect to another node. Without this "magic" it was taking
@@ -88,16 +136,24 @@ class ZooKeeper(AbstractDCS):
         than loop_wait, because we can spend up to 2 seconds when calling `touch_member()` and
         `write_leader_optime()` methods, which also may hang..."""
 
-        ret = self._orig_kazoo_connect(host, port)
+        ret = self._orig_kazoo_connect(*args)
         return max(self.loop_wait - 2, 2)*1000, ret[1]
 
     def session_listener(self, state):
         if state in [KazooState.SUSPENDED, KazooState.LOST]:
             self.cluster_watcher(None)
 
+    def status_watcher(self, event):
+        self._fetch_status = True
+        self.event.set()
+
     def cluster_watcher(self, event):
         self._fetch_cluster = True
-        self.event.set()
+        if not event or event.state != KazooState.CONNECTED or event.path.startswith(self.client_path('')):
+            self.status_watcher(event)
+
+    def members_watcher(self, event):
+        self._fetch_cluster = True
 
     def reload_config(self, config):
         self.set_retry_timeout(config['retry_timeout'])
@@ -114,13 +170,14 @@ class ZooKeeper(AbstractDCS):
         # the same time, set_ttl method will reestablish connection and return
         # `!True`, otherwise we will close existing connection and let kazoo
         # open the new one.
-        if not self.set_ttl(int(config['ttl'] * 1000)) and loop_wait_changed:
+        if not self.set_ttl(config['ttl']) and loop_wait_changed:
             self._client._connection._socket.close()
 
     def set_ttl(self, ttl):
         """It is not possible to change ttl (session_timeout) in zookeeper without
         destroying old session and creating the new one. This method returns `!True`
         if session_timeout has been changed (`restart()` has been called)."""
+        ttl = int(ttl * 1000)
         if self._client._session_timeout != ttl:
             self._client._session_timeout = ttl
             self._client.restart()
@@ -128,7 +185,7 @@ class ZooKeeper(AbstractDCS):
 
     @property
     def ttl(self):
-        return self._client._session_timeout
+        return self._client._session_timeout / 1000.0
 
     def set_retry_timeout(self, retry_timeout):
         retry = self._client.retry if isinstance(self._client.retry, KazooRetry) else self._client._retry
@@ -141,6 +198,30 @@ class ZooKeeper(AbstractDCS):
         except NoNodeError:
             return None
 
+    def get_status(self, path, leader):
+        watch = self.status_watcher if not leader or leader.name != self._name else None
+
+        status = self.get_node(path + self._STATUS, watch)
+        if status:
+            try:
+                status = json.loads(status[0])
+                last_lsn = status.get(self._OPTIME)
+                slots = status.get('slots')
+            except Exception:
+                slots = last_lsn = None
+        else:
+            last_lsn = self.get_node(path + self._LEADER_OPTIME, watch)
+            last_lsn = last_lsn and last_lsn[0]
+            slots = None
+
+        try:
+            last_lsn = int(last_lsn)
+        except Exception:
+            last_lsn = 0
+
+        self._fetch_status = False
+        return last_lsn, slots
+
     @staticmethod
     def member(name, value, znode):
         return Member.from_node(znode.version, name, znode.ephemeralOwner, value)
@@ -151,78 +232,103 @@ class ZooKeeper(AbstractDCS):
         except NoNodeError:
             return []
 
-    def load_members(self, sync_standby):
+    def load_members(self, path):
         members = []
-        for member in self.get_children(self.members_path, self.cluster_watcher):
+        for member in self.get_children(path + self._MEMBERS, self.cluster_watcher):
-            watch = member == sync_standby and self.cluster_watcher or None
+            data = self.get_node(path + self._MEMBERS + member)
-            data = self.get_node(self.members_path + member, watch)
             if data is not None:
                 members.append(self.member(member, *data))
         return members
 
-    def _inner_load_cluster(self):
+    def _cluster_loader(self, path):
         self._fetch_cluster = False
         self.event.clear()
-        nodes = set(self.get_children(self.client_path(''), self.cluster_watcher))
+        nodes = set(self.get_children(path, self.cluster_watcher))
         if not nodes:
             self._fetch_cluster = True
 
         # get initialize flag
-        initialize = (self.get_node(self.initialize_path) or [None])[0] if self._INITIALIZE in nodes else None
+        initialize = (self.get_node(path + self._INITIALIZE) or [None])[0] if self._INITIALIZE in nodes else None
 
         # get global dynamic configuration
-        config = self.get_node(self.config_path, watch=self.cluster_watcher) if self._CONFIG in nodes else None
+        config = self.get_node(path + self._CONFIG, watch=self.cluster_watcher) if self._CONFIG in nodes else None
         config = config and ClusterConfig.from_node(config[1].version, config[0], config[1].mzxid)
 
         # get timeline history
-        history = self.get_node(self.history_path, watch=self.cluster_watcher) if self._HISTORY in nodes else None
+        history = self.get_node(path + self._HISTORY, watch=self.cluster_watcher) if self._HISTORY in nodes else None
         history = history and TimelineHistory.from_node(history[1].mzxid, history[0])
 
-        # get last leader operation
-        last_leader_operation = self._OPTIME in nodes and self._fetch_cluster and self.get_node(self.leader_optime_path)
-        last_leader_operation = last_leader_operation and int(last_leader_operation[0]) or 0
-
         # get synchronization state
-        sync = self.get_node(self.sync_path, watch=self.cluster_watcher) if self._SYNC in nodes else None
+        sync = self.get_node(path + self._SYNC, watch=self.cluster_watcher) if self._SYNC in nodes else None
         sync = SyncState.from_node(sync and sync[1].version, sync and sync[0])
 
         # get list of members
-        sync_standby = sync.leader == self._name and sync.sync_standby or None
+        members = self.load_members(path) if self._MEMBERS[:-1] in nodes else []
-        members = self.load_members(sync_standby) if self._MEMBERS[:-1] in nodes else []
 
         # get leader
-        leader = self.get_node(self.leader_path) if self._LEADER in nodes else None
+        leader = self.get_node(path + self._LEADER) if self._LEADER in nodes else None
         if leader:
-            client_id = self._client.client_id
+            member = Member(-1, leader[0], None, {})
-            if not self._ctl and leader[0] == self._name and client_id is not None \
+            member = ([m for m in members if m.name == leader[0]] or [member])[0]
-                    and client_id[0] != leader[1].ephemeralOwner:
+            leader = Leader(leader[1].version, leader[1].ephemeralOwner, member)
-                logger.info('I am leader but not owner of the session. Removing leader node')
+            self._fetch_cluster = member.index == -1
-                self._client.delete(self.leader_path)
-                leader = None
 
-            if leader:
+        # get last known leader lsn and slots
-                member = Member(-1, leader[0], None, {})
+        last_lsn, slots = self.get_status(path, leader)
-                member = ([m for m in members if m.name == leader[0]] or [member])[0]
-                leader = Leader(leader[1].version, leader[1].ephemeralOwner, member)
-                self._fetch_cluster = member.index == -1
 
         # failover key
-        failover = self.get_node(self.failover_path, watch=self.cluster_watcher) if self._FAILOVER in nodes else None
+        failover = self.get_node(path + self._FAILOVER, watch=self.cluster_watcher) if self._FAILOVER in nodes else None
         failover = failover and Failover.from_node(failover[1].version, failover[0])
 
-        return Cluster(initialize, config, leader, last_leader_operation, members, failover, sync, history)
+        # get failsafe topology
+        failsafe = self.get_node(path + self._FAILSAFE, watch=self.cluster_watcher) if self._FAILSAFE in nodes else None
+        try:
+            failsafe = json.loads(failsafe[0]) if failsafe else None
+        except Exception:
+            failsafe = None
 
-    def _load_cluster(self):
+        return Cluster(initialize, config, leader, last_lsn, members, failover, sync, history, slots, failsafe)
-        cluster = self.cluster
+
+    def _citus_cluster_loader(self, path):
+        fetch_cluster = False
+        ret = {}
+        for node in self.get_children(path, self.cluster_watcher):
+            if citus_group_re.match(node):
+                ret[int(node)] = self._cluster_loader(path + node + '/')
+                fetch_cluster = fetch_cluster or self._fetch_cluster
+        self._fetch_cluster = fetch_cluster
+        return ret
+
+    def _load_cluster(self, path, loader):
+        cluster = self.cluster if path == self._base_path + '/' else None
         if self._fetch_cluster or cluster is None:
             try:
-                cluster = self._client.retry(self._inner_load_cluster)
+                cluster = self._client.retry(loader, path)
             except Exception:
                 logger.exception('get_cluster')
                 self.cluster_watcher(None)
                 raise ZooKeeperError('ZooKeeper in not responding properly')
+        # The /status ZNode was updated or doesn't exist
+        elif self._fetch_status and not self._fetch_cluster or not cluster.last_lsn \
+                or cluster.has_permanent_logical_slots(self._name, False) and not cluster.slots:
+            # If current node is the leader just clear the event without fetching anything (we are updating the /status)
+            if cluster.leader and cluster.leader.name == self._name:
+                self.event.clear()
+            else:
+                try:
+                    last_lsn, slots = self.get_status(self.client_path(''), cluster.leader)
+                    self.event.clear()
+                    cluster = list(cluster)
+                    cluster[3] = last_lsn
+                    cluster[8] = slots
+                    cluster = Cluster(*cluster)
+                except Exception:
+                    pass
         return cluster
 
+    def _bypass_caches(self):
+        self._fetch_cluster = True
+
     def _create(self, path, value, retry=False, ephemeral=False):
         try:
             if retry:
@@ -234,11 +340,18 @@ class ZooKeeper(AbstractDCS):
             logger.exception('Failed to create %s', path)
         return False
 
-    def attempt_to_acquire_leader(self, permanent=False):
+    def attempt_to_acquire_leader(self):
-        ret = self._create(self.leader_path, self._name.encode('utf-8'), retry=True, ephemeral=not permanent)
+        try:
-        if not ret:
+            self._client.retry(self._client.create, self.leader_path, self._name.encode('utf-8'),
-            logger.info('Could not take out TTL lock')
+                               makepath=True, ephemeral=True)
-        return ret
+            return True
+        except (ConnectionClosedError, RetryFailedError) as e:
+            raise ZooKeeperError(e)
+        except Exception as e:
+            if not isinstance(e, NodeExistsError):
+                logger.error('Failed to create %s: %r', self.leader_path, e)
+        logger.info('Could not take out TTL lock')
+        return False
 
     def _set_or_create(self, key, value, index=None, retry=False, do_not_create_empty=False):
         value = value.encode('utf-8')
@@ -270,14 +383,17 @@ class ZooKeeper(AbstractDCS):
         return self._create(self.initialize_path, sysid, retry=True) if create_new \
             else self._client.retry(self._client.set, self.initialize_path, sysid)
 
-    def touch_member(self, data, permanent=False):
+    def touch_member(self, data):
         cluster = self.cluster
         member = cluster and cluster.get_member(self._name, fallback_to_leader=False)
-        encoded_data = json.dumps(data, separators=(',', ':')).encode('utf-8')
+        member_data = self.__last_member_data or member and member.data
+        #  We want to notify leader if some important fields in the member key changed by removing ZNode
         if member and (self._client.client_id is not None and member.session != self._client.client_id[0] or
-                       not (deep_compare(member.data.get('tags', {}), data.get('tags', {})) and
+                       not (deep_compare(member_data.get('tags', {}), data.get('tags', {})) and
-                            member.data.get('version') == data.get('version') and
+                            (member_data.get('state') == data.get('state') or
-                            member.data.get('checkpoint_after_promote') == data.get('checkpoint_after_promote'))):
+                                'running' not in (member_data.get('state'), data.get('state'))) and
+                            member_data.get('version') == data.get('version') and
+                            member_data.get('checkpoint_after_promote') == data.get('checkpoint_after_promote'))):
             try:
                 self._client.delete_async(self.member_path).get(timeout=1)
             except NoNodeError:
@@ -286,13 +402,14 @@ class ZooKeeper(AbstractDCS):
                 return False
             member = None
 
+        encoded_data = json.dumps(data, separators=(',', ':')).encode('utf-8')
         if member:
-            if deep_compare(data, member.data):
+            if deep_compare(data, member_data):
                 return True
         else:
             try:
-                self._client.create_async(self.member_path, encoded_data, makepath=True,
+                self._client.create_async(self.member_path, encoded_data, makepath=True, ephemeral=True).get(timeout=1)
-                                          ephemeral=not permanent).get(timeout=1)
+                self.__last_member_data = data
                 return True
             except Exception as e:
                 if not isinstance(e, NodeExistsError):
@@ -300,6 +417,7 @@ class ZooKeeper(AbstractDCS):
                     return False
         try:
             self._client.set_async(self.member_path, encoded_data).get(timeout=1)
+            self.__last_member_data = data
             return True
         except Exception:
             logger.exception('touch_member')
@@ -309,13 +427,41 @@ class ZooKeeper(AbstractDCS):
     def take_leader(self):
         return self.attempt_to_acquire_leader()
 
-    def _write_leader_optime(self, last_operation):
+    def _write_leader_optime(self, last_lsn):
-        return self._set_or_create(self.leader_optime_path, last_operation)
+        return self._set_or_create(self.leader_optime_path, last_lsn)
+
+    def _write_status(self, value):
+        return self._set_or_create(self.status_path, value)
+
+    def _write_failsafe(self, value):
+        return self._set_or_create(self.failsafe_path, value)
 
     def _update_leader(self):
+        cluster = self.cluster
+        session = cluster and isinstance(cluster.leader, Leader) and cluster.leader.session
+        if self._client.client_id and self._client.client_id[0] != session:
+            logger.warning('Recreating the leader ZNode due to ownership mismatch')
+            try:
+                self._client.retry(self._client.delete, self.leader_path)
+            except NoNodeError:
+                pass
+            except (ConnectionClosedError, RetryFailedError) as e:
+                raise ZooKeeperError(e)
+            except Exception as e:
+                logger.error('Failed to remove %s: %r', self.leader_path, e)
+                return False
+
+            try:
+                self._client.retry(self._client.create, self.leader_path,
+                                   self._name.encode('utf-8'), makepath=True, ephemeral=True)
+            except (ConnectionClosedError, RetryFailedError) as e:
+                raise ZooKeeperError(e)
+            except Exception as e:
+                logger.error('Failed to create %s: %r', self.leader_path, e)
+                return False
         return True
 
-    def delete_leader(self):
+    def _delete_leader(self):
         self._client.restart()
         return True
 
@@ -346,6 +492,7 @@ class ZooKeeper(AbstractDCS):
         return self.set_sync_state_value("{}", index)
 
     def watch(self, leader_index, timeout):
-        if super(ZooKeeper, self).watch(leader_index, timeout):
+        ret = super(ZooKeeper, self).watch(leader_index, timeout + 0.5)
+        if ret and not self._fetch_status:
             self._fetch_cluster = True
-        return self._fetch_cluster
+        return ret or self._fetch_cluster

patroni/exceptions.py

@@ -13,6 +13,10 @@ class PatroniException(Exception):
         return repr(self.value)
 
 
+class PatroniFatalException(PatroniException):
+    pass
+
+
 class PostgresException(PatroniException):
     pass
 
@@ -27,3 +31,7 @@ class PostgresConnectionException(PostgresException):
 
 class WatchdogError(PatroniException):
     pass
+
+
+class ConfigParseError(PatroniException):
+    pass

patroni/log.py

@@ -11,6 +11,20 @@ from threading import Lock, Thread
 _LOGGER = logging.getLogger(__name__)
 
 
+def debug_exception(logger_obj, msg, *args, **kwargs):
+    kwargs.pop("exc_info", False)
+    if logger_obj.isEnabledFor(logging.DEBUG):
+        logger_obj.debug(msg, *args, exc_info=True, **kwargs)
+    else:
+        msg = "{0}, DETAIL: '{1}'".format(msg, sys.exc_info()[1])
+        logger_obj.error(msg, *args, exc_info=False, **kwargs)
+
+
+def error_exception(logger_obj, msg, *args, **kwargs):
+    exc_info = kwargs.pop("exc_info", True)
+    logger_obj.error(msg, *args, exc_info=exc_info, **kwargs)
+
+
 class QueueHandler(logging.Handler):
 
     def __init__(self):
@@ -48,9 +62,20 @@ class QueueHandler(logging.Handler):
         return self._records_lost
 
 
+class ProxyHandler(logging.Handler):
+
+    def __init__(self, patroni_logger):
+        logging.Handler.__init__(self)
+        self.patroni_logger = patroni_logger
+
+    def emit(self, record):
+        self.patroni_logger.log_handler.handle(record)
+
+
 class PatroniLogger(Thread):
 
     DEFAULT_LEVEL = 'INFO'
+    DEFAULT_TRACEBACK_LEVEL = 'ERROR'
     DEFAULT_FORMAT = '%(asctime)s %(levelname)s: %(message)s'
 
     NORMAL_LOG_QUEUE_SIZE = 2  # When everything goes normal Patroni writes only 2 messages per HA loop
@@ -59,16 +84,18 @@ class PatroniLogger(Thread):
 
     def __init__(self):
         super(PatroniLogger, self).__init__()
-        self.daemon = True
         self._queue_handler = QueueHandler()
         self._root_logger = logging.getLogger()
-        self._root_logger.addHandler(self._queue_handler)
         self._config = None
-        self._log_handler = None
+        self.log_handler = None
+        self.log_handler_lock = Lock()
         self._old_handlers = []
-        self._log_handler_lock = Lock()
         self.reload_config({'level': 'DEBUG'})
-        self.start()
+        # We will switch to the QueueHandler only when thread was started.
+        # This is necessary to protect from the cases when Patroni constructor
+        # failed and PatroniLogger thread remain running and prevent shutdown.
+        self._proxy_handler = ProxyHandler(self)
+        self._root_logger.addHandler(self._proxy_handler)
 
     def update_loggers(self):
         loggers = deepcopy(self._config.get('loggers') or {})
@@ -87,18 +114,22 @@ class PatroniLogger(Thread):
                 self._queue_handler.queue.maxsize = config.get('max_queue_size', self.DEFAULT_MAX_QUEUE_SIZE)
 
             self._root_logger.setLevel(config.get('level', PatroniLogger.DEFAULT_LEVEL))
+            if config.get('traceback_level', PatroniLogger.DEFAULT_TRACEBACK_LEVEL).lower() == 'debug':
+                logging.Logger.exception = debug_exception
+            else:
+                logging.Logger.exception = error_exception
 
             new_handler = None
             if 'dir' in config:
-                if not isinstance(self._log_handler, RotatingFileHandler):
+                if not isinstance(self.log_handler, RotatingFileHandler):
                     new_handler = RotatingFileHandler(os.path.join(config['dir'], __name__))
-                handler = new_handler or self._log_handler
+                handler = new_handler or self.log_handler
                 handler.maxBytes = int(config.get('file_size', 25000000))
                 handler.backupCount = int(config.get('file_num', 4))
             else:
-                if self._log_handler is None or isinstance(self._log_handler, RotatingFileHandler):
+                if self.log_handler is None or isinstance(self.log_handler, RotatingFileHandler):
                     new_handler = logging.StreamHandler()
-                handler = new_handler or self._log_handler
+                handler = new_handler or self.log_handler
 
             oldlogformat = (self._config or {}).get('format', PatroniLogger.DEFAULT_FORMAT)
             logformat = config.get('format', PatroniLogger.DEFAULT_FORMAT)
@@ -110,17 +141,17 @@ class PatroniLogger(Thread):
                 handler.setFormatter(logging.Formatter(logformat, dateformat))
 
             if new_handler:
-                with self._log_handler_lock:
+                with self.log_handler_lock:
-                    if self._log_handler:
+                    if self.log_handler:
-                        self._old_handlers.append(self._log_handler)
+                        self._old_handlers.append(self.log_handler)
-                    self._log_handler = new_handler
+                    self.log_handler = new_handler
 
             self._config = config.copy()
             self.update_loggers()
 
     def _close_old_handlers(self):
         while True:
-            with self._log_handler_lock:
+            with self.log_handler_lock:
                 if not self._old_handlers:
                     break
                 handler = self._old_handlers.pop()
@@ -130,6 +161,13 @@ class PatroniLogger(Thread):
                 _LOGGER.exception('Failed to close the old log handler %s', handler)
 
     def run(self):
+        # switch to QueueHandler only when the thread was started
+        with self.log_handler_lock:
+            self._root_logger.addHandler(self._queue_handler)
+            self._root_logger.removeHandler(self._proxy_handler)
+
+        prev_record = None
+
         while True:
             self._close_old_handlers()
 
@@ -137,7 +175,18 @@ class PatroniLogger(Thread):
             if record is None:
                 break
 
-            self._log_handler.handle(record)
+            if self._root_logger.level == logging.INFO:
+                if record.msg.startswith('Lock owner: '):
+                    prev_record, record = record, None
+                else:
+                    if prev_record and prev_record.thread == record.thread:
+                        if not (record.msg.startswith('no action. ') or record.msg.startswith('PAUSE: no action')):
+                            self.log_handler.handle(prev_record)
+                        prev_record = None
+
+            if record:
+                self.log_handler.handle(record)
+
             self._queue_handler.queue.task_done()
 
     def shutdown(self):

patroni/postgresql/bootstrap.py

@@ -4,10 +4,11 @@ import shlex
 import tempfile
 import time
 
-from patroni.dcs import RemoteMember
-from patroni.utils import deep_compare, uri
 from six import string_types
-from six.moves.urllib.parse import quote_plus
+
+from ..dcs import RemoteMember
+from ..psycopg import quote_ident, quote_literal
+from ..utils import deep_compare
 
 logger = logging.getLogger(__name__)
 
@@ -54,7 +55,7 @@ class Bootstrap(object):
                     error_handler('Error when parsing {0} option {1}: value should be string value'
                                   ' or a single key-value pair'.format(tool, opt))
         else:
-            error_handler('{0} options must be list ot dict'.format(tool))
+            error_handler('{0} options must be list or dict'.format(tool))
         return user_options
 
     def _initdb(self, config):
@@ -91,15 +92,16 @@ class Bootstrap(object):
         self._postgresql.configure_server_parameters()
 
         # make sure there is no trigger file or postgres will be automatically promoted
-        trigger_file = 'promote_trigger_file' if self._postgresql.major_version >= 120000 else 'trigger_file'
+        trigger_file = self._postgresql.config.triggerfile_good_name
-        trigger_file = self._postgresql.config.get('recovery_conf', {}).get(trigger_file) or 'promote'
+        trigger_file = (self._postgresql.config.get('recovery_conf') or {}).get(trigger_file) or 'promote'
         trigger_file = os.path.abspath(os.path.join(self._postgresql.data_dir, trigger_file))
         if os.path.exists(trigger_file):
             os.unlink(trigger_file)
 
     def _custom_bootstrap(self, config):
         self._postgresql.set_state('running custom bootstrap script')
-        params = ['--scope=' + self._postgresql.scope, '--datadir=' + self._postgresql.data_dir]
+        params = [] if config.get('no_params') else ['--scope=' + self._postgresql.scope,
+                                                     '--datadir=' + self._postgresql.data_dir]
         try:
             logger.info('Running custom bootstrap script: %s', config['command'])
             if self._postgresql.cancellable.call(shlex.split(config['command']) + params) != 0:
@@ -123,28 +125,15 @@ class Bootstrap(object):
         cmd = config.get('post_bootstrap') or config.get('post_init')
         if cmd:
             r = self._postgresql.config.local_connect_kwargs
-
+            connstring = self._postgresql.config.format_dsn(r, True)
-            if 'host' in r:
+            if 'host' not in r:
-                # '/tmp' => '%2Ftmp' for unix socket path
-                host = quote_plus(r['host']) if r['host'].startswith('/') else r['host']
-            else:
-                host = ''
-
                 # https://www.postgresql.org/docs/current/static/libpq-pgpass.html
                 # A host name of localhost matches both TCP (host name localhost) and Unix domain socket
                 # (pghost empty or the default socket directory) connections coming from the local machine.
                 r['host'] = 'localhost'  # set it to localhost to write into pgpass
 
-            if 'user' in r:
+            env = self._postgresql.config.write_pgpass(r)
-                user = r['user']
+            env['PGOPTIONS'] = '-c synchronous_commit=local'
-            else:
-                user = ''
-                if 'password' in r:
-                    import getpass
-                    r.setdefault('user', os.environ.get('PGUSER', getpass.getuser()))
-
-            connstring = uri('postgres', (host, r['port']), r['database'], user)
-            env = self._postgresql.write_pgpass(r) if 'password' in r else None
 
             try:
                 ret = self._postgresql.cancellable.call(shlex.split(cmd) + [connstring], env=env)
@@ -166,19 +155,19 @@ class Bootstrap(object):
         self._postgresql.set_state('creating replica')
         self._postgresql.schedule_sanity_checks_after_pause()
 
-        is_remote_master = isinstance(clone_member, RemoteMember)
+        is_remote_member = isinstance(clone_member, RemoteMember)
 
         # get list of replica methods either from clone member or from
         # the config. If there is no configuration key, or no value is
         # specified, use basebackup
-        replica_methods = (clone_member.create_replica_methods if is_remote_master
+        replica_methods = (clone_member.create_replica_methods if is_remote_member
                            else self._postgresql.create_replica_methods) or ['basebackup']
 
         if clone_member and clone_member.conn_url:
             r = clone_member.conn_kwargs(self._postgresql.config.replication)
-            connstring = uri('postgres', (r['host'], r['port']), r['database'], r['user'])
             # add the credentials to connect to the replica origin to pgpass.
-            env = self._postgresql.write_pgpass(r)
+            env = self._postgresql.config.write_pgpass(r)
+            connstring = self._postgresql.config.format_dsn(r, True)
         else:
             connstring = ''
             env = os.environ.copy()
@@ -223,7 +212,7 @@ class Bootstrap(object):
                                           "datadir": self._postgresql.data_dir,
                                           "connstring": connstring})
                 else:
-                    for param in ('no_params', 'no_master', 'keep_data'):
+                    for param in ('no_params', 'no_master', 'no_leader', 'keep_data'):
                         method_config.pop(param, None)
                 params = ["--{0}={1}".format(arg, val) for arg, val in method_config.items()]
                 try:
@@ -280,7 +269,7 @@ class Bootstrap(object):
 
     def clone(self, clone_member):
         """
-             - initialize the replica from an existing member (master or replica)
+             - initialize the replica from an existing member (primary or replica)
              - initialize the replica using the replica creation method that
                works without the replication connection (i.e. restore from on-disk
                base backup)
@@ -310,22 +299,30 @@ class Bootstrap(object):
         if 'NOLOGIN' not in options and 'LOGIN' not in options:
             options.append('LOGIN')
 
-        params = [name]
         if password:
-            options.extend(['PASSWORD', '%s'])
+            options.extend(['PASSWORD', quote_literal(password)])
-            params.extend([password, password])
 
         sql = """DO $$
 BEGIN
     SET local synchronous_commit = 'local';
-    PERFORM * FROM pg_authid WHERE rolname = %s;
+    PERFORM * FROM pg_catalog.pg_authid WHERE rolname = {0};
     IF FOUND THEN
-        ALTER ROLE "{0}" WITH {1};
+        ALTER ROLE {1} WITH {2};
     ELSE
-        CREATE ROLE "{0}" WITH {1};
+        CREATE ROLE {1} WITH {2};
     END IF;
-END;$$""".format(name, ' '.join(options))
+END;$$""".format(quote_literal(name), quote_ident(name, self._postgresql.connection()), ' '.join(options))
-        self._postgresql.query(sql, *params)
+        self._postgresql.query('SET log_statement TO none')
+        self._postgresql.query('SET log_min_duration_statement TO -1')
+        self._postgresql.query("SET log_min_error_statement TO 'log'")
+        self._postgresql.query("SET pg_stat_statements.track_utility to 'off'")
+        try:
+            self._postgresql.query(sql)
+        finally:
+            self._postgresql.query('RESET log_min_error_statement')
+            self._postgresql.query('RESET log_min_duration_statement')
+            self._postgresql.query('RESET log_statement')
+            self._postgresql.query('RESET pg_stat_statements.track_utility')
 
     def post_bootstrap(self, config, task):
         try:
@@ -347,8 +344,8 @@ END;$$""".format(name, ' '.join(options))
                         sql = """DO $$
 BEGIN
     SET local synchronous_commit = 'local';
-    GRANT EXECUTE ON function pg_catalog.{0} TO "{1}";
+    GRANT EXECUTE ON function pg_catalog.{0} TO {1};
-END;$$""".format(f, rewind['username'])
+END;$$""".format(f, quote_ident(rewind['username'], postgresql.connection()))
                         postgresql.query(sql)
 
                 for name, value in (config.get('users') or {}).items():
@@ -361,7 +358,8 @@ END;$$""".format(f, rewind['username'])
                     self._running_custom_bootstrap = False
                     # If we don't have custom configuration for pg_hba.conf we need to restore original file
                     if not postgresql.config.get('pg_hba'):
-                        os.unlink(postgresql.config.pg_hba_conf)
+                        if os.path.exists(postgresql.config.pg_hba_conf):
+                            os.unlink(postgresql.config.pg_hba_conf)
                         postgresql.config.restore_configuration_files()
                     postgresql.config.write_postgresql_conf()
                     postgresql.config.replace_pg_ident()
@@ -369,7 +367,7 @@ END;$$""".format(f, rewind['username'])
                     # at this point there should be no recovery.conf
                     postgresql.config.remove_recovery_conf()
 
-                    if postgresql.config.hba_file and postgresql.config.hba_file != postgresql.config.pg_hba_conf:
+                    if postgresql.config.hba_file:
                         postgresql.restart()
                     else:
                         postgresql.config.replace_pg_hba()
@@ -379,6 +377,9 @@ END;$$""".format(f, rewind['username'])
                             postgresql.reload()
                             time.sleep(1)  # give a time to postgres to "reload" configuration files
                             postgresql.connection().close()  # close connection to reconnect with a new password
+                else:  # initdb
+                    # We may want create database and extension for citus
+                    self._postgresql.citus_handler.bootstrap()
         except Exception:
             logger.exception('post_bootstrap')
             task.complete(False)