Improve protection of DCS being accidentally wiped (#680)

We already have a lot of logic in place to prevent failover in such case and restore all keys, but an accidental removal of `/config` key was effectively switching off pause mode for 1 cycle of HA loop.
This commit is contained in:
Alexander Kukushkin
2018-05-18 11:18:58 +02:00
committed by GitHub
parent 5296336f4a
commit 5ce18a8045
4 changed files with 24 additions and 11 deletions
+4 -1
View File
@@ -9,7 +9,7 @@ from collections import defaultdict
from copy import deepcopy
from patroni.dcs import ClusterConfig
from patroni.postgresql import Postgresql
from patroni.utils import deep_compare, parse_int, patch_config
from patroni.utils import deep_compare, parse_bool, parse_int, patch_config
from requests.structures import CaseInsensitiveDict
logger = logging.getLogger(__name__)
@@ -88,6 +88,9 @@ class Config(object):
def dynamic_configuration(self):
return deepcopy(self._dynamic_configuration)
def check_mode(self, mode):
return bool(parse_bool(self._dynamic_configuration.get(mode)))
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:
+6 -5
View File
@@ -11,6 +11,7 @@ import sys
from collections import namedtuple
from patroni.exceptions import PatroniException
from patroni.utils import parse_bool
from random import randint
from six.moves.urllib_parse import urlparse, urlunparse, parse_qsl
from threading import Event, Lock
@@ -349,14 +350,14 @@ class Cluster(namedtuple('Cluster', 'initialize,config,leader,last_leader_operat
candidates = [m for m in self.members if m.clonefrom and m.is_running and m.name not in exclude]
return candidates[randint(0, len(candidates) - 1)] if candidates else self.leader
def check_mode(self, mode):
return bool(self.config and parse_bool(self.config.data.get(mode)))
def is_paused(self):
return self.config and self.config.data.get('pause', False) or False
return self.check_mode('pause')
def is_synchronous_mode(self):
return bool(self.config and self.config.data.get('synchronous_mode'))
def is_synchronous_mode_strict(self):
return bool(self.config and self.config.data.get('synchronous_mode_strict'))
return self.check_mode('synchronous_mode')
@six.add_metaclass(abc.ABCMeta)
+10 -3
View File
@@ -76,8 +76,15 @@ class Ha(object):
# already running as replica was aborted due to cluster not beeing initialized in DCS.
self._join_aborted = False
def check_mode(self, mode):
# Try to protect from the case when DCS was wiped out during pause
if self.cluster and self.cluster.config and self.cluster.config.modify_index:
return self.cluster.check_mode(mode)
else:
return self.patroni.config.check_mode(mode)
def is_paused(self):
return self.cluster and self.cluster.is_paused()
return self.check_mode('pause')
def load_cluster_from_dcs(self):
cluster = self.dcs.get_cluster()
@@ -288,10 +295,10 @@ class Ha(object):
return follow_reason
def is_synchronous_mode(self):
return bool(self.cluster and self.cluster.is_synchronous_mode())
return self.check_mode('synchronous_mode')
def is_synchronous_mode_strict(self):
return bool(self.cluster and self.cluster.is_synchronous_mode_strict())
return self.check_mode('synchronous_mode_strict')
def process_sync_replication(self):
"""Process synchronous standby beahvior.
+4 -2
View File
@@ -90,7 +90,7 @@ def etcd_read(self, key, **kwargs):
raise etcd.EtcdKeyNotFound
response = {"action": "get", "node": {"key": "/service/batman5", "dir": True, "nodes": [
{"key": "/service/batman5/config", "value": '{"foo": "bar"}',
{"key": "/service/batman5/config", "value": '{"synchronous_mode": 0}',
"modifiedIndex": 1582, "createdIndex": 1582},
{"key": "/service/batman5/failover", "value": "",
"modifiedIndex": 1582, "createdIndex": 1582},
@@ -276,7 +276,9 @@ class TestEtcd(unittest.TestCase):
{'hosts': 'foo:4001,bar', 'retry_timeout': 10})
def test_get_cluster(self):
self.assertIsInstance(self.etcd.get_cluster(), Cluster)
cluster = self.etcd.get_cluster()
self.assertIsInstance(cluster, Cluster)
self.assertFalse(cluster.is_synchronous_mode())
self.etcd._base_path = '/service/nocluster'
cluster = self.etcd.get_cluster()
self.assertIsInstance(cluster, Cluster)