Thread: Detect streaming replication failure

Detect streaming replication failure

From
Lists
Date:
For reference:
https://wiki.postgresql.org/wiki/Streaming_Replication

Assume a master -> slave streaming replication configuration, Postgresql
9.2.
Assume that the master has been chugging away, but the slave PG service
has been offline
for a while and the wal archive has updated enough that the slave cannot
catch up.

When I start the slave PG instance, pg launches and "runs" but doesn't
update. It also doesn't seem to throw any errors. The only outward sign
that I can see that anything is wrong is that
pg_last_xlog_replay_location() doesn't update. I can look in
/var/lib/pgsql/9.2/data/pg_log/postgresql-Thu.csv and see errors there EG:

2014-07-17 22:38:23.851 UTC,,,21310,,53c8505f.533e,2,,2014-07-17
22:38:23 UTC,,0,FATAL,XX000,"could not receive data from WAL stream:
FATAL:  requested WAL segment 000000070000000500000071 has already been
removed

Is that the only way to detect this condition? I guess I'm looking for
something like

select * from pg_is_replicating_ok();
1

on the slave. At the moment, it appears that I can either parse the log
file, or look for pg_last_xact_replay_timestamp() > acceptable threshold
minutes in the past.

http://www.postgresql.org/docs/9.2/static/functions-admin.html

Thanks,

Ben


Re: Detect streaming replication failure

From
wd
Date:
you can run select * from pg_stat_replication on master to check all the salve stats.


On Fri, Jul 18, 2014 at 6:50 AM, Lists <lists@benjamindsmith.com> wrote:
For reference:
https://wiki.postgresql.org/wiki/Streaming_Replication

Assume a master -> slave streaming replication configuration, Postgresql 9.2.
Assume that the master has been chugging away, but the slave PG service has been offline
for a while and the wal archive has updated enough that the slave cannot catch up.

When I start the slave PG instance, pg launches and "runs" but doesn't update. It also doesn't seem to throw any errors. The only outward sign that I can see that anything is wrong is that pg_last_xlog_replay_location() doesn't update. I can look in /var/lib/pgsql/9.2/data/pg_log/postgresql-Thu.csv and see errors there EG:

2014-07-17 22:38:23.851 UTC,,,21310,,53c8505f.533e,2,,2014-07-17 22:38:23 UTC,,0,FATAL,XX000,"could not receive data from WAL stream: FATAL:  requested WAL segment 000000070000000500000071 has already been removed

Is that the only way to detect this condition? I guess I'm looking for something like

select * from pg_is_replicating_ok();
1

on the slave. At the moment, it appears that I can either parse the log file, or look for pg_last_xact_replay_timestamp() > acceptable threshold minutes in the past.

http://www.postgresql.org/docs/9.2/static/functions-admin.html

Thanks,

Ben


--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general