Re: What's canceling autovacuum tasks? - Mailing list pgsql-general

From Alban Hertroys
Subject Re: What's canceling autovacuum tasks?
Date
Msg-id 13EA19F5-17D4-4444-818B-82487FF08FD0@solfertje.student.utwente.nl
Whole thread Raw
In response to Re: What's canceling autovacuum tasks?  (Herouth Maoz <herouth@unicell.co.il>)
List pgsql-general
On 6 Feb 2011, at 18:52, Herouth Maoz wrote:

> on 06/02/11 18:16, quoting Tom Lane:
>>
>> Most likely, some other session requested an exclusive lock on the
>> table.  Autovacuum will quit to avoid blocking the other query.
>>
> That's strange. During the day, only selects are running on that database, or at worst, temporary tables are being
createdand updated. And that particular table gets updated only on weekends (it's one of my archive tables). Besides, I
assumethat a simple update/insert/delete is not supposed to request an exclusive lock, or autovacuum would not work at
allin an average database. Even backups don't run during the day, and I think backups also don't create an exclusive
lockor I'd never see a vacuum process run more than a day. 
>
> This is really inexplicable.


You could try turning on statement-level logging. On a busy database the logs would probably grow huge, but you should
beable to see what statements coincide with autovacuum aborting. 

Alban Hertroys

--
If you can't see the forest for the trees,
cut the trees and you'll see there is no forest.


!DSPAM:737,4d4f250611735773614733!



pgsql-general by date:

Previous
From: "Francisco Figueiredo Jr."
Date:
Subject: Re: "could not accept SSPI security context"
Next
From: Jean-Armel Luce
Date:
Subject: Question about switchover with PG9 replication