When grilled further on (Mon, 18 Jul 2005 00:10:53 -0400),
Tom Lane <tgl@sss.pgh.pa.us> confessed:
> The context swap problem was no worse in 8.0 than in prior versions,
> so that hardly seems like a good explanation. Have you tried reverting
> to the cron-based vacuuming method you used in 7.4?
>
Ran 7 hours on 741 with VACUUM ANALYZE every 5 minutes. The largest CS I saw
was 40k, with an average of 500 (via script which monitors vmstat output).
I've done a VACUUM FULL ANALYZE on 803 and have switched the cron based VACUUM
ANALYZE to 803 also. The tests are now running again.
> Hmm, did you read this thread?
> http://archives.postgresql.org/pgsql-performance/2005-07/msg00088.php
I just glanced at it. Once I've reproduced (or not) the problem on 803 with the
VACUUM FULL, I'll turn off the vacuum delay.
Cheers,
Rob
--
07:10:06 up 4 days, 10:45, 6 users, load average: 0.28, 0.40, 0.29
Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004