Re: after using pg_resetxlog, db lost - Mailing list pgsql-performance

From Shea,Dan [CIS]
Subject Re: after using pg_resetxlog, db lost
Date
Msg-id 644D07D3D59D8F408CD01AC2F833D8C62B9218@cisxa.cmc.int.ec.gc.ca
Whole thread Raw
In response to after using pg_resetxlog, db lost  ("Shea,Dan [CIS]" <Dan.Shea@ec.gc.ca>)
List pgsql-performance
Tom, thank you for your help.
I increased 000E to 81920 and the databse is working now.
We are using RHAS 3.0 and it does have /dev/zero.

Dan.

-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Thursday, June 24, 2004 12:34 PM
To: Shea,Dan [CIS]
Cc: pgsql-performance@postgresql.org
Subject: Re: [PERFORM] after using pg_resetxlog, db lost


"Shea,Dan [CIS]" <Dan.Shea@ec.gc.ca> writes:
> I determined the largest was 12,293,162 and set it to
> pg_resetxlog -x 15000000 /var/lib/pgsql/data

Okay, but it looks like you will also need to adjust pg_clog to cover
that transaction ID range.  (I had thought pg_resetxlog would handle
this for you, but it looks like not.)

> Jun 24 15:02:05 murphy postgres[28061]: [11-1] PANIC:  could not access status of transaction 15000030
> Jun 24 15:02:05 murphy postgres[28061]: [11-2] DETAIL:  could not read from file "/var/lib/pgsql/data/pg_clog/000E"
atoffset 73728: Success 

You need to append zeroes (8K at a time) to
/var/lib/pgsql/data/pg_clog/000E until it's longer than 73728 bytes.
I'd use something like
    dd bs=8k count=1 </dev/zero >>/var/lib/pgsql/data/pg_clog/000E
assuming that your system has /dev/zero.

            regards, tom lane

pgsql-performance by date:

Previous
From: Tom Lane
Date:
Subject: Re: after using pg_resetxlog, db lost
Next
From: Chris Cheston
Date:
Subject: postgres 7.4 at 100%