Thanks a lot Richard. The culprit in my case is the restore command , I modified it as follows per your instructions , it is fine now
restore_command = 'pg_standby -d -s 5 -t /tmp/pg_standby.trigger.5432 /opt/postgres/archive %f %p %r 2>>/tmp/standby.log'
One small doubt I have is , do we have to keep the backup_label on the standby, I deleted it before I start the restore process , it still worked .. Just curious
appreicate your help
Regards
On Mon, Nov 23, 2009 at 1:04 PM, Richard Huxton
<dev@archonet.com> wrote:
akp geek wrote:
> Hi experts -
>
> I am running into issue with pg_standby. May be my
> understanding is not correct. Please help. here is what I did .
>
>
> 1. I made changes in the postgresql.conf ( archive_mode = on
> ,archive_command = 'cp -i %p /opt/postgres/archive/%f' , archive_timeout =
> 60s )
> 2. pg_ctl start -D $PGDATA -l /opt/postgres/logfile
> 3. postgres=# select pg_start_backup('BKP_LBL');
> 4. I have done the base backup and I have used the following command
> for restore in the recovery.conf
> 5. restore_command = 'pg_standby -d -s 3 -t
> /tmp/pg_standby.trigger.5432 /opt/postgres/archive/%f %p %r'
From memory, do you not want a space between ".../archive/" and "%f" - I
think they're separate parameters.
Also, you'll want to redirect STDERR to a file - add the following to
the end of the command: 2>>/tmp/standby.log
That way we'll be able to see what's happening.
> 6. postgres=# select pg_stop_backup();
> 7. Now I have made some changes to the master database and I waited for
> 30 minutes
Or just generate lots of changes so WAL files get filled.
> 8. I have stopped the master database
> 9. I have started the slave. But I did not find the changes I have done
> after I issued the command elect pg_stop_backup();
> 10. The log files have shipped to archive folder
Good. That's a useful fact. Let's see if anything shows up in our
standby logfile once we have it.
--
Richard Huxton
Archonet Ltd