Re: pg_basebackup WAL -- invalid pages

On Mon, Dec 3, 2012 at 12:44 PM, Darron <darron(at)realtyserver(dot)com> wrote:
> Running PostgreSQL 9.2.1 on x86_64-unknown-linux-gnu, compiled by gcc
> (Debian 4.4.5-8) 4.4.5, 64-bit
>
>
>
> Having some issues with streaming replication. After creating a basebackup
> with pg_basebackup, I get the following in the logs when starting the backup
> as a steaming hot standby.
>
>
>
> 2012-12-03 20:27:08 GMT 21494 dLOG: database system was interrupted while
> in recovery at log time 2012-11-29 21:32:38 GMT
>
> 2012-12-03 20:27:08 GMT 21494 dHINT: If this has occurred more than once
> some data might be corrupted and you might need to choose an earlier
> recovery target.
>
> 2012-12-03 20:27:08 GMT 21494 dLOG: entering standby mode
>
> 2012-12-03 20:27:08 GMT 21494 dDEBUG: checkpoint record is at 1C/E28B7978
>
> 2012-12-03 20:27:08 GMT 21494 dDEBUG: redo record is at 1C/E2000020;
> shutdown FALSE
>
> 2012-12-03 20:27:08 GMT 21494 dDEBUG: next transaction ID: 0/1903870;
> next OID: 249837974
>
> 2012-12-03 20:27:08 GMT 21494 dDEBUG: next MultiXactId: 1; next
> MultiXactOffset: 0
>
> 2012-12-03 20:27:08 GMT 21494 dDEBUG: oldest unfrozen transaction ID:
> 673, in database 1
>
> 2012-12-03 20:27:08 GMT 21494 dDEBUG: transaction ID wrap limit is
> 2147484320, limited by database with OID 1
>
> 2012-12-03 20:27:08 GMT 21494 dDEBUG: resetting unlogged relations:
> cleanup 1 init 0
>
> 2012-12-03 20:27:08 GMT 21494 dDEBUG: initializing for hot standby
>
> 2012-12-03 20:27:08 GMT 21494 dLOG: consistent recovery state reached at
> 1C/E28B79D8
>
> 2012-12-03 20:27:08 GMT 21494 dLOG: redo starts at 1C/E2000020
>
> 2012-12-03 20:27:08 GMT 21494 dDEBUG: recovery snapshots are now enabled
>
> 2012-12-03 20:27:08 GMT 21494 dCONTEXT: xlog redo running xacts: nextXid
> 1903870 latestCompletedXid 1903869 oldestRunningXid 1903870
>
> 2012-12-03 20:27:08 GMT 21492 dLOG: database system is ready to accept
> read only connections
>
> 2012-12-03 20:27:09 GMT 21494 dWARNING: page 27 of relation
> base/103111746/231701662 does not exist
>
> 2012-12-03 20:27:09 GMT 21494 dCONTEXT: xlog redo visible: rel
> 1663/103111746/231701662; blk 27
>
> 2012-12-03 20:27:09 GMT 21494 dPANIC: WAL contains references to invalid
> pages
>
> 2012-12-03 20:27:09 GMT 21494 dCONTEXT: xlog redo visible: rel
> 1663/103111746/231701662; blk 27
>
> 2012-12-03 20:27:09 GMT 21492 dLOG: startup process (PID 21494) was
> terminated by signal 6: Aborted
>
> 2012-12-03 20:27:09 GMT 21492 dLOG: terminating any other active server
> processes
>
> 2012-12-03 20:27:09 GMT 21493 dDEBUG: logger shutting down
>
>
>
> The Autovacumm launcher is running on the master postgres server, would that
> cause issues with pg_basebackup. Any ideas what the problem could be?

autovacuum is definitely not the cause of this. That's a default
setting, and if it was problematic with pg_basebackup it would be well
documented.

Are you certain that the pg_basebackup ran successfully to completion?
Can you provide the exact command options you used when running it?
Are all the postgres servers running the same version of postgresql
with the same OS version?
Can you document the exact steps you took when extracting the
basebackup on the server?