From: | David Steele <david(at)pgmasters(dot)net> |
---|---|
To: | Venkata Balaji N <nag1010(at)gmail(dot)com>, "pgsql-hackers(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: recovery_target_time ignored ? |
Date: | 2015-03-23 22:54:23 |
Message-ID: | 5510999F.3070106@pgmasters.net |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On 3/23/15 12:42 AM, Venkata Balaji N wrote:
> Hi,
>
> Assuming that this might require a patch, i am posting this in
> pgsql-hackers. Apologies, if this is not the appropriate mailing list to
> start this discussion.
>
> I performed a PITR and saw the below message in the log file is a bit
> confusing.
>
> 2015-03-23 13:49:09.816 GMT-10 DB= PID=4707 LOG: *database system was
> interrupted; last known up at 2015-03-23 10:30:26 GMT-10*
> 2015-03-23 13:49:09.817 GMT-10 DB= PID=4707 LOG: *starting point-in-time
> recovery to 2015-03-23 10:00:26+10*
> 2015-03-23 13:49:09.827 GMT-10 DB= PID=4707 LOG: restored log file
> "000000010000000B00000020" from archive
> 2015-03-23 13:49:09.888 GMT-10 DB= PID=4707 LOG: redo starts at B/20000090
> 2015-03-23 13:49:09.937 GMT-10 DB= PID=4707 LOG: consistent recovery
> state reached at B/200000B8
> 2015-03-23 13:49:09.947 GMT-10 DB= PID=4707 LOG: restored log file
> "000000010000000B00000021" from archive
> 2015-03-23 13:49:09.950 GMT-10 DB= PID=4707 LOG: *recovery stopping
> before commit of transaction 16267, time 2015-03-23 13:22:37.53007+10*
>
>
> By mistake i gave "recovery_target_time" as 10:00 GMT which is 25/30
> minutes behind the backup start/end time registered in the backup_label.
>
> The parameter recovery_target_time is ignored and recovery proceeds
> further applying all the available WAL Archive files finally ends up
> bringing up the database.
>
> I think it would make sense if the recovery does not proceed any further
> and error out with a message like "recovery_target_time is behind the
> backup time.. please consider using the backup taken prior to the
> recovery_target_time"
I just tried it with 9.3.5 and I do get an error:
LOG: starting point-in-time recovery to 2015-03-23 17:26:02.721307-04
LOG: restored log file "000000010000000000000003" from archive
LOG: redo starts at 0/30000C8
LOG: recovery stopping before commit of transaction 1001, time
2015-03-23 18:26:01.012593-04
LOG: redo done at 0/3000228
FATAL: requested recovery stop point is before consistent recovery point
Here's my recovery.conf file:
restore_command = '/usr/bin/pg_backrest.pl --stanza=db archive-get %f "%p"'
recovery_target_time = '2015-03-23 17:26:02.721307 EDT'
> recovery.conf file is as follows :
>
> restore_command='cp /data/pgdata9400backup/pgwalarch9400backup/%f %p '
> recovery_target_time='2015-03-23 10:00:26 GMT-10'
> recovery_target_inclusive='true'
You have '2015-03-23 10:00:26 GMT-10' in recovery.conf but the log says
'starting point-in-time recovery to 2015-03-23 10:00:26+10'. Note the -
vs +.
Could you check your log and recovery.conf and make sure the timezone
offsets are actually different?
--
- David Steele
david(at)pgmasters(dot)net
From | Date | Subject | |
---|---|---|---|
Next Message | Andrew Gierth | 2015-03-23 23:08:48 | Re: Abbreviated keys for Numeric |
Previous Message | Tomas Vondra | 2015-03-23 22:14:58 | Re: PATCH: numeric timestamp in log_line_prefix |