Re: ERROR: out of free buffers: time to abort! - Mailing list pgsql-general

From Joseph Shraibman
Subject Re: ERROR: out of free buffers: time to abort!
Date
Msg-id 3E7A41F9.9030903@selectacast.net
Whole thread Raw
In response to Re: ERROR: out of free buffers: time to abort!  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: ERROR: out of free buffers: time to abort!  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-general
Tom Lane wrote:
> Joseph Shraibman <jks@selectacast.net> writes:
>
>>Unlikely.  I run this cron job every day, and every day I get the same
>>error.  The whole thing should be pretty quick.
>
>
> Well, I can't reproduce the problem here, and in general this isn't an
> error message we hear of very often.  So there's got to be something
> unusual about what you're doing.  Any chance that you're invoking
> triggers recursively, or something like that?  Could you possibly get
> a stack trace from the point of the elog call?
>
>             regards, tom lane

My update looks like:
UPDATE tablename SET intfield = 2 WHERE keyfield IN( ... )

If I lowered the number of items in the IN() then I didn't get the error, but what that
number is is a moving target.  205 used to work a few minutes ago, but now 200 doesn't
work.  A vaccuum seems to help matters.  In previous versions of postgres I was able to do
up to 10000.

I tried to make a simple test with a table with 10000 entries, but that had no problems.
Maybe I would need a bigger table.


pgsql-general by date:

Previous
From: "Ed L."
Date:
Subject: 32/64-bit transaction IDs?
Next
From: Flower Sun
Date:
Subject: system table backup strategy