Re: PATCH: add "--config-file=" option to pg_rewind - Mailing list pgsql-hackers

Hi,

Am Donnerstag, dem 10.03.2022 um 15:28 +0100 schrieb Gunnar "Nick"
Bluth:
> Am 10.03.22 um 14:43 schrieb Michael Banck:
> > some minor comments, I didn't look at the added test and I did not
> > test the patch at all, but (as part of the Debian/Ubuntu packaging
> > team) I think this patch is really important:
> 
> Much appreciated!
> 
> [...]
> 
> Fixed.
> 
> [...]

Thanks for the updated patch.

The patch applies, make is ok, make check is ok, pg_rewind TAP tests
are ok.

I did some tests now with Patroni 2.1.3 and the attached patch applied.

The following test was made:

1. Deploy 3-node (pg1, pg2, pg3) patroni cluster on Debian unstable
   running postgresql-15 (approx. master)
2. Switch on archive_mode, and set archive_command and restore_command
3. Switchover so that pg1 is the primary (if not already)
4. Kill pg1 hard
5. Wait till a new leader has taken over and the (now 2-node) cluster 
   is healthy again
6. Restart pg1

without this patch:

|Apr 03 19:09:01 pg1 patroni@15-main[99]: 2022-04-03 19:09:01,084 INFO: running pg_rewind from pg3
|Apr 03 19:09:01 pg1 patroni@15-main[99]: 2022-04-03 19:09:01,102 INFO: running pg_rewind from dbname=postgres
user=postgres_rewindhost=10.0.3.227 port=5432 target_session_attrs=read-write
 
|Apr 03 19:09:01 pg1 patroni@15-main[99]: 2022-04-03 19:09:01,135 INFO: pg_rewind exit code=1
|Apr 03 19:09:01 pg1 patroni@15-main[99]: 2022-04-03 19:09:01,136 INFO:  stdout=
|Apr 03 19:09:01 pg1 patroni@15-main[99]: 2022-04-03 19:09:01,136 INFO:  stderr=postgres: could not access the server
configurationfile "/var/lib/postgresql/15/main/postgresql.conf": No such file or directory
 
|Apr 03 19:09:01 pg1 patroni@15-main[99]: no data was returned by command "/usr/lib/postgresql/15/bin/postgres -D
/var/lib/postgresql/15/main-C restore_command"
 
|Apr 03 19:09:01 pg1 patroni@15-main[99]: 2022-04-03 19:09:01,149 ERROR: Failed to rewind from healty master: pg3
|Apr 03 19:09:01 pg1 patroni@15-main[99]: 2022-04-03 19:09:01,149 WARNING: remove_data_directory_on_diverged_timelines
isset. removing...
 
|Apr 03 19:09:01 pg1 patroni@15-main[99]: 2022-04-03 19:09:01,149 INFO: Removing data directory:
/var/lib/postgresql/15/main
|Apr 03 19:09:01 pg1 patroni@15-main[99]: 2022-04-03 19:09:01,245 INFO: Lock owner: pg3; I am pg1
|Apr 03 19:09:01 pg1 patroni@15-main[99]: 2022-04-03 19:09:01,248 INFO: trying to bootstrap from leader 'pg3'

So pg_rewind fails and Patroni does a re-bootstrap.

with this patch:

|Apr 03 19:12:35 pg1 patroni@15-main[99]: 2022-04-03 19:12:35,576 INFO: running pg_rewind from pg2
|Apr 03 19:12:35 pg1 patroni@15-main[99]: 2022-04-03 19:12:35,590 INFO: running pg_rewind from dbname=postgres
user=postgres_rewindhost=10.0.3.145 port=5432 target_session_attrs=read-write
 
|Apr 03 19:12:37 pg1 patroni@15-main[99]: 2022-04-03 19:12:37,147 INFO: pg_rewind exit code=0
|Apr 03 19:12:37 pg1 patroni@15-main[99]: 2022-04-03 19:12:37,148 INFO:  stdout=
|Apr 03 19:12:37 pg1 patroni@15-main[99]: 2022-04-03 19:12:37,148 INFO:  stderr=pg_rewind: servers diverged at WAL
location0/1D000180 on timeline 38
 
|Apr 03 19:12:37 pg1 patroni@15-main[99]: pg_rewind: rewinding from last common checkpoint at 0/1D000108 on timeline
38
|Apr 03 19:12:37 pg1 patroni@15-main[99]: pg_rewind: Done!
|Apr 03 19:12:37 pg1 patroni@15-main[99]: 2022-04-03 19:12:37,151 WARNING: Postgresql is not running.
|Apr 03 19:12:37 pg1 patroni@15-main[99]: 2022-04-03 19:12:37,151 INFO: Lock owner: pg2; I am pg1

Here, pg_rewind is called and works fine.

So I think this works as intended, and I'm marking it Ready for
Committer.


Michael

-- 
Michael Banck
Teamleiter PostgreSQL-Team
Projektleiter
Tel.: +49 2166 9901-171
E-Mail: michael.banck@credativ.de

credativ GmbH, HRB Mönchengladbach 12080
USt-ID-Nummer: DE204566209
Trompeterallee 108, 41189 Mönchengladbach
Geschäftsführung: Dr. Michael Meskes, Geoff Richardson, Peter Lilley

Unser Umgang mit personenbezogenen Daten unterliegt
folgenden Bestimmungen: https://www.credativ.de/datenschutz