On 22 March 2010 14:29, Tom Lane
<tgl@sss.pgh.pa.us> wrote:
>> Force a checkpoint, if one hasn't happened already.
> Yeah, I've run a CHECKPOINT too, but no joy. Still the same number of
> files.
Hm, it works for me. What PG version is this exactly? Are you sure
the active value of checkpoint_segments really changed? (Use SHOW)
regards, tom lane
This is version 8.4.3
postgres=# show checkpoint_segments;
checkpoint_segments
---------------------
3
(1 row)
Not sure if it's of any relevance, but the file name sequence seems to go a bit weird.
Nov 25 19:49 0000000100000000000000FA
Nov 26 17:25 0000000100000000000000FB
Dec 16 14:50 0000000100000000000000FC
Dec 17 04:00 0000000100000000000000FD
Dec 17 15:26 0000000100000000000000FE
Dec 17 15:27 000000010000000100000000
Dec 17 15:29 000000010000000100000001
Mar 22 14:28 0000000100000000000000BF
And yes, this particular cluster has very low usage at the moment.
Thom
