Keith Pierno <kpierno@lulu.com> writes:
> <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
> <html>
> <head>
BTW, please do not post all-HTML mail to this list.
> The timeline for the events all dates MM/DD/YYYY
>
> 06/09/2009 1310 EDT - Hardware fault on primary database server db01pri
> 06/09/2009 1325 EDT - Failover to warm standby db01sec
Hmm, how exactly was the failover managed?
> [postgres@db01pri ~]$ ls -l
> total 98468
> -rw------- 1 postgres postgres 74 Jul 10 2008 00000002.history
> -rw------- 1 postgres postgres 74 Jun 9 13:29 00000003.history
> -rw------- 1 postgres postgres 16777216 Jun 16 08:45 0000000400000749000000C9
> -rw------- 1 postgres postgres 16777216 Jun 16 08:46 0000000400000749000000CA
> -rw------- 1 postgres postgres 16777216 Jun 16 08:47 0000000400000749000000CB
> -rw------- 1 postgres postgres 74 Jun 9 13:33 00000004.history
> drwxr-xr-x 2 postgres postgres 32768 Jun 16 08:46 archive_status
> -rw------- 1 postgres postgres 16777216 Jun 9 13:45 xlogtemp.17243
> -rw------- 1 postgres postgres 16777216 Jun 9 13:45 xlogtemp.17244
> -rw------- 1 postgres postgres 16777216 Jun 9 13:52 xlogtemp.17397
The xlogtemp files are a bit suspicious; they shouldn't be there. The
timestamps suggest that something odd was going on right after the
failover. Do you have the slave's postmaster log from that time
interval, and if so, does it show anything out of the ordinary?
regards, tom lane