Re: Server Panic when trying to stop point in time recovery

On Tue, Apr 27, 2010 at 4:02 PM, Joshua D. Drake <jd(at)commandprompt(dot)com> wrote:
>>
> Well the problem is that the files are missing. How are you copying
> them?
>
> Joshua D. Drake
>

I am using rsync to copy the files from the master into a directory
/opt/wal/archwalremote on the standby.

My restore script copies the requested file from
/opt/wal/archwalremote to wherever it is asked to
(pg_xlog/RECOVERYXLOG).

Here is the end of the log from my restore script. First,
00000001000000A000000001 is requested and copied. Then
00000001000000A000000002 is requested, but it has not been synced yet
so the script is sleeping and polling for that file to exist. Before
the file exists, the trigger is detected and the restore script starts
returning 1 to all requests.

Tue Apr 27 15:06:00 CDT 2010 > cp
/opt/wal/archwalremote/00000001000000A000000001 pg_xlog/RECOVERYXLOG
Tue Apr 27 15:06:00 CDT 2010 > cp returned 0, script returning 0
Tue Apr 27 15:06:00 CDT 2010 Starting up... From
/opt/wal/archwalremote/00000001000000A000000002 TO
pg_xlog/RECOVERYXLOG
Tue Apr 27 15:06:00 CDT 2010
/opt/wal/archwalremote/00000001000000A000000002 not found
Tue Apr 27 15:06:00 CDT 2010 Sleeping
Tue Apr 27 15:09:11 CDT 2010
/opt/wal/archwalremote/00000001000000A000000002 not found
Tue Apr 27 15:09:11 CDT 2010 Sleeping
Tue Apr 27 15:09:21 CDT 2010 Triggered
Tue Apr 27 15:09:21 CDT 2010 Starting up... From
/opt/wal/archwalremote/00000001000000A000000001 TO
pg_xlog/RECOVERYXLOG
Tue Apr 27 15:09:21 CDT 2010 Triggered

00000001000000A000000001 is found and restored (see original log
file). I guess the question is: why, after the server should be out
of recovery mode, is it looking for 00000001000000A000000001 in the
pg_xlog directory?