On Mon, 2007-06-18 at 13:26 -0400, Woody Woodring wrote:
> I am wondering if there is a proper procedure for shutting down the
> warm_standby server (8.2.4)? I am using pg_standby as my restore script in
> my testing:
>
> [postgres@jbeam]$ cat recovery.conf
> restore_command = 'pg_standby -m -d -s 5 -w 0 -t /tmp/pgsql.trigger.5432
> /usr/local2/pg_archive %f %p 2>> standby.log'
>
> My issue is it looks like the when the warm_standby comes back up, it is
> looking for a file that has already been loaded (and deleted). It was
> looking for log '*30' when it was shut down, but upon startup again it is
> looking for '*2F'.
The -m command is not supported in the most recent version of pg_standby
because it results in the error your point out. The latest version has a
-k option that works around this error and a server patch is in the
queue for 8.3 that will allow a more flexible approach to this.
I'll add a --version option to pg_standby to allow us to discuss which
version is in use, to avoid such issues in future. Thanks,
--
Simon Riggs
EnterpriseDB http://www.enterprisedb.com