Re: BUG #7500: hot-standby replica crash after an initial rsync

On Tuesday, August 21, 2012 10:08:11 AM maxim.boguk@gmail.com wrote:
> The following bug has been logged on the website:
>
> Bug reference:      7500
> Logged by:          Maxim Boguk
> Email address:      maxim.boguk@gmail.com
> PostgreSQL version: 9.0.8
> Operating system:   FreeBSD
> Description:
>
> Hi,
>
> Reinitialization of the replica after failover is procedure performed by me
> a lot times in that installation. And I never had any issues. But today
> after following the same pg_start_backup/rsync procedure replica doesn't
> start with following error:
>
> 2012-08-21 11:17:07.777 MSK 34329 @ from  [vxid: txid:0] []LOG:  entering
> standby mode
> 2012-08-21 11:17:07.780 MSK 34329 @ from  [vxid:1/0 txid:0] []LOG:  redo
> starts at 925/2C0004C8
> 2012-08-21 11:17:07.783 MSK 34329 @ from  [vxid:1/0 txid:0] []FATAL:  could
> not access status of transaction 361788644
> 2012-08-21 11:17:07.783 MSK 34329 @ from  [vxid:1/0 txid:0] []DETAIL:
> Could not read from file "pg_multixact/offsets/1590" at offset 114688:
> Unknown error: 0.
> 2012-08-21 11:17:07.783 MSK 34329 @ from  [vxid:1/0 txid:0] []CONTEXT:
> xlog redo create multixact 361788644 offset 822580965: 1768508061
> 1768508078 2012-08-21 11:17:07.791 MSK 34327 @ from  [vxid: txid:0] []LOG:
>  startup process (PID 34329) exited with exit code 1
> 2012-08-21 11:17:07.791 MSK 34327 @ from  [vxid: txid:0] []LOG:
> terminating any other active server processes
>
>
> I have kept all that database files for the future investigation.
>
> What I should look into first?
Could you reproduce the error with log_error_verbosity=verbose? Or even better
provide a backtrace with gdb?

Greetings,

Andres
--
 Andres Freund                       http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services