On 02/24/2013 10:05 PM, Jeff Frost wrote:
>
> On Feb 24, 2013, at 7:16 AM, Rafael Martinez Guerrero
> <r.m.guerrero@usit.uio.no <mailto:r.m.guerrero@usit.uio.no>> wrote:
>
>> We reported this back in 2011, but we did not get to any conclusion:
>> http://www.postgresql.org/message-id/4DE89072.7070305@usit.uio.no
>>
>> In our case, we had this problem when creating a GIN index.
>>
>> I think the problem has something to do with checkpoints, I think the
>> number of WAL files will grow beyond the limit defined because the
>> system can not be finish with checkpoints. A good start to try to
>> explain what it is happening would be to try to identify or describe the
>> situations where checkpoints can take very long to complete or fail
>> altogether.
>>
>> We are interested in this subject because we have had this problem a few
>> times. But we have not been able to identify the reason that triggers
>> this.
>
> Rafael, did you do a followup post on pgsql-performance, or did the
> thread die at the end of the one you post above?
>
Jeff, we didn't have the time to continue investigating this problem
that time. Our workaround was to increase the size of our pg_xlog
partition so we could finish the export/import job we had to do.
regards,
--
Rafael Martinez Guerrero
Center for Information Technology
University of Oslo, Norway
PGP Public Key: http://folk.uio.no/rafael/