>> You can check if PITR is disabled by executing:
>> SHOW archive_command;
>> At a psql prompt. If that's empty, then you turned it off, and it isn't
>> involved in the shutdown issue you have anymore.
> Yes, I could see empty value here.
>> Take a look at what processes are still running against the database and
>> see if there are clients attached after the fast shutdown attempt.
>> If so, those are your problem, not something to PITR.
> Yes, I can see the following processes still running. I used the command
'ps
> -ef |grep postgres' to list.
> postgres 3376 1 0 07:20 ? 00:00:00
> /mnt/database/pgsql/bin/postgres -D /usr/local/pgsql/data
> postgres 3381 3376 0 07:20 ? 00:00:00 postgres: logger process
> postgres 3383 3376 0 07:20 ? 00:00:00 postgres: writer process
> But why logger and writer process are still running/showing up here for a
> very long time after issuing 'service postgresql stop'. What could be the
> problem? Also, am sure here that no other clients (database connections)
> are connected at this time.
I can CONSISTENTLY reproduce this problem (PITR walarchive drive is not
attached/mounted) and I still don't know the reason behind this issue:
1) If I disable PITR and start postgresql service, consistently server does
not shut down on 'service postgresql stop' command. I have tried and
checked this upto 4 times.
NOTE: After disabling PITR, on checking "SHOW archive_command" from psql,
it is showing empty value. But on 'ps -ef |grep postgres', it is always
showing up
logger and writer process running.
2) But if I enable back PITR and start postgresql service, consistently
server shuts down properly on 'service postgresql stop' command. Again I
have tried this cycle also upto 4 times.
When I mean enable/disable PITR, it is about enabling/disabling 2
configuration parameters: "archive_command" and "archive_timeout".
Since this is our Production server, experts assistance to solve this issue
are highly appreciated.
