gitops/patroni
1
0
Fork 0
mirror of https://github.com/outbackdingo/patroni.git synced 2026-08-27 08:00:28 +00:00
Code Issues Packages Projects Releases Wiki Activity
1,190 Commits 15 Branches 66 Tags
3181c4e59f2581ef13630eb17f34e95e6ff2fc03
Commit Graph
1190 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
Oleksii Kliukin 3181c4e59f Code review, asynchronous restarts. 
- Make the restart initiated by the schedule asynchronous
- Fix the placeholders in logs.
- Fix the regexp to detect the PostgreSQL version.
 2016-07-12 20:25:01 +02:00
Oleksii Kliukin ec160f0d59 Do not send 2 quotes for the empty request, instead, send None. 2016-07-11 18:20:42 +02:00
Oleksii Kliukin bf95b75489 Use the parameter that really sets the pending_restart flag. 2016-07-11 18:20:15 +02:00
Oleksii Kliukin b17483b7dd Fix the PG version regex. 2016-07-11 15:21:31 +02:00
Oleksii Kliukin c91eda8d78 Merge branch 'master' into feature/scheduled_restarts 2016-07-11 12:56:24 +02:00
Oleksii Kliukin 6da2eecb90 Increase the test coverage. 2016-07-11 11:51:07 +03:00
Oleksii Kliukin 8548956370 Bumped version to 1.0 v1.0 2016-07-05 17:03:19 +02:00
Oleksii Kliukin 34d18cc182 Set the standard name for the docker-compose file. 2016-07-05 17:02:40 +02:00
Alexander KukushkinandGitHub 58ad8399cf Merge pull request #238 from zalando/bugfix/pg_rewind 
One more try to fix pg_rewind
 2016-07-05 16:25:03 +02:00
Alexander Kukushkin 659f7617f5 New option: remove_data_directory_on_rewind_failure 
One more try to fix pg_rewind
 2016-07-05 12:11:15 +02:00
Oleksii Kliukin 8834f929aa Improve the unit tests/coverage. 2016-07-05 10:07:29 +02:00
Oleksii Kliukin 36a86c67d0 Enable the conditions on normal restart. 2016-07-04 15:46:22 +02:00
Alexander KukushkinandGitHub a19dbfaddf Merge pull request #232 from zalando/bugfix/pg_rewind 
Start readonly when holding leader lock
 2016-07-04 13:11:35 +02:00
Alexander KukushkinandGitHub 40d48eae76 Merge pull request #237 from zalando/bugfix/etcd-retry 
Calculate numer of retries and timeout
 2016-07-04 13:11:25 +02:00
Alexander Kukushkin f7b9709907 Calculate numer of retries and timeout 2016-07-04 12:01:54 +02:00
Alexander Kukushkin 2944a4bcbd Start readonly when holding the leader lock 2016-07-04 11:08:24 +02:00
Alexander Kukushkin b84e22c4ea Implement more checks in the follow method 
Although such situation should not happen in reality (follow method is
not supposed to be called when when the node is holding leader lock and
postgres is running), but to be on the safe side it is better to
implement as much checks as possible, because this method could
potentially remove data directory.
 2016-07-04 10:56:37 +02:00
Alexander KukushkinandGitHub f9298d30ca Merge pull request #231 from zalando/bugfix/etcd-retry 
Fix retry logic in etcd.py
 2016-07-04 10:37:55 +02:00
Alexander KukushkinandGitHub e025f335dc Merge pull request #233 from zalando/bugfix/initialize 
Initialize key can be present but empty
 2016-07-04 09:48:09 +02:00
Alexander KukushkinandGitHub a6d840d6be Merge pull request #234 from zalando/feature/zookeeper-connect-strategy 
Implement different connect strategy for zookeeper
 2016-07-04 09:47:55 +02:00
Alexander KukushkinandGitHub cd752fea14 Merge pull request #235 from zalando/bugfix/docker-volumes 
Volume could be passed to the docker only with absolute path
 2016-07-04 09:47:35 +02:00
Alexander KukushkinandGitHub 3f1d06da32 Merge pull request #236 from zalando/bugfix/strtol 
bugfix: strtol didn't worked correctly with 1 digit numbers
 2016-07-04 09:47:09 +02:00
Alexander Kukushkin bc9aec9076 bugfix: strtol didn't worked correctly with 1 digit numbers 2016-07-01 16:19:57 +02:00
Alexander Kukushkin b7639f5b22 Volume could be passed to the docker only with absolute path 
In addition to that add zookeeper support to the docker.
 2016-07-01 12:43:33 +02:00
Alexander Kukushkin f7c6bd4eab Implement different connect strategy for zookeeper 
Originally it was trying to connect during session_timeout time.
Such strategy doesn't work good during short network hiccups...
 2016-07-01 12:31:29 +02:00
Alexander Kukushkin ee529669d2 Start readonly when holding leader lock 
Not starting of postgres was causeing situation when there were no
master running...
 2016-07-01 12:28:02 +02:00
Alexander Kukushkin 8bd071d9a9 Initialize key can be present but empty 
Nodes were trying to grab initialize key when it didn't contained sysid
 2016-07-01 12:25:00 +02:00
Alexander KukushkinandGitHub dc27a30800 Merge pull request #230 from zalando/bugfix/pg_rewind 
Try to cover as much as possible pg_rewind corner-cases
 2016-06-30 12:09:10 +02:00
Alexander KukushkinandGitHub 489ed20331 Merge pull request #229 from zalando/bugfix/env-zookeeper-hosts 
Document that every single zookeeper host:port MUST be quoted
 2016-06-30 12:09:03 +02:00
Alexander Kukushkin aa10f42913 checkpoint method returns string status message 2016-06-30 10:45:54 +02:00
Feike Steenbergen 72f8fcbb5b Document the per node timeout 2016-06-29 16:40:39 +02:00
Alexander Kukushkin 876cfdfb2d Fix retry logic in etcd.py 
Client class takes care about retrying when connection to the etcd node
fails. It calculates amount of retries and timeout depending on etcd
cluster size.

Etcd class should not retry when EtcdConnectionFailed exception is
raised (this case is already handled in the Client).

Besides that adjust retry timeouts in the Client class.
 2016-06-29 15:30:54 +02:00
Alexander Kukushkin 4b67008488 Try to cover as much as possible pg_rewind corner-cases 
rewind is not possible when:
1) trying to rewind from themself
2) leader is not reachable
3) leader is_in_recovery

All these cases were leading to removing of data directory...
In all cases except 1) it should "retry" when leader will became
available and not is_in_recovery.
 2016-06-29 14:29:31 +02:00
Alexander Kukushkin ae88e7c96e Document that every single zookeeper host:port MUST be quoted 
otherwise yaml library can not parse the list.
And make visible yaml exception when trying to parse this list.
 2016-06-29 14:25:50 +02:00
Oleksii Kliukin 7a1e2e0c72 Fix the assert message. 2016-06-28 17:11:13 +02:00
Oleksii Kliukin d2832ee43b Address the code review. 
Fix return  value in the should_run_scheduled_action and the comments.
Correct the json composition in the scheduled_restart test.
Fix the delete in case there is no scheduled restart.
Fix the usage of format in the logger output.
Fix the indentation in the evaluate_scheduled_restart.
Fix the condition related to the body_is_optional in the do_POST_restart.
Fix a few typos in the error messages.
Fix the _read_json_content
Make the scheduled restart unit-tests a bit less ugly
 2016-06-28 16:54:20 +02:00
Alexander KukushkinandGitHub 0c6a1307da Merge pull request #228 from zalando/bugfix/always-expose-role 
bugfix: api must report role=master during pg_ctl stop
 2016-06-28 15:24:16 +02:00
Alexander KukushkinandGitHub cae3b8ec1e Merge pull request #226 from zalando/bugfix/sigterm 
Setup signal handler before creating dcs
 2016-06-28 15:24:05 +02:00
Alexander KukushkinandGitHub 6fdd01a8eb Merge pull request #225 from zalando/feature/confd 
sample haproxy templates for confd
 2016-06-28 15:23:54 +02:00
Alexander KukushkinandGitHub fe0bc2f3eb Merge pull request #223 from zalando/feature/patronictl-at 
Acceptance tests for patronictl
 2016-06-28 15:23:44 +02:00
Alexander Kukushkin 0318749b56 bugfix: api must report role=master during pg_ctl stop 
In addition for that make pg_ctl --timeout option configurable.
If the stop or start didn't succeeded during given timeout when demoting
master, role will be forcibly changed to 'unknown' and all needed
callbacks executed.
 2016-06-28 14:14:42 +02:00
Oleksii Kliukin 854ff27e56 Allow normal conditional restarts. 
In addition, use the RLock instead of the Lock in async executor
to make sure the lock can be acquired more than once from a single
thread.
 2016-06-27 09:50:09 +02:00
Oleksii Kliukin 568eb730bc Clear the scheduled restart after the normal one. 
Make sure the scheduled restart flag is cleared when the
postmaster_start_time changes since the time restart was scheduled.

Additionally, separate the logic of checking the restart conditions
into the function in order to support conditions for the normal
restart as well.
 2016-06-24 17:39:04 +02:00
Oleksii Kliukin 29845dd383 Restart the node according to the schedule. 
The scheduled restart data structures are now independent of those
used by the normal restarts. This would be fixed in subsequent
commits.
Add the behave tests, that cover the POST /restart (but not DELETE).
 2016-06-23 10:43:54 +02:00
Alexander Kukushkin 44433c2d42 Setup signal handler before creating dcs 
Otherwise it was swallowing SysExit exception in an infinite loop.
 2016-06-22 09:13:47 +02:00
Alexander Kukushkin d65d1028a7 Add patroni-compose-etcd-3.yml 
For starting up cluster easy with docker-compose.
And unify Dockerfile and scripts to be able to work with docker-compose
and the old one dev_patroni_cluster.sh script
 2016-06-21 17:07:24 +02:00
Oleksii Kliukin e5cf06101a Fix line is too long warnings. 2016-06-21 11:19:33 +02:00
Oleksii Kliukin 6a8bfdeb76 Decouple the schedule check from the failover. 2016-06-21 11:02:10 +02:00
Oleksii Kliukin 70195bec2d Handle empty body correctly when reading requests. 2016-06-21 10:56:18 +02:00
Oleksii Kliukin c2490d4831 Merge branch 'master' into feature/scheduled_restarts 2016-06-20 15:38:20 +02:00