Philip Warner kirjutas P, 03.11.2002 kell 15:41:
> At 03:25 PM 3/11/2002 +0500, Hannu Krosing wrote:
> >a separate backend in a loop that
> >kept doing VACUUM TABLE with only 5 seconds sleep between
> Good grief! I thought 5 minutes was bad enough. Can't wait for b/g vacuum. 
> Thanks for the input; I'll wait for a day or so to get some figures as you 
> suggest.

The 5 sec number was for case when tens of worker threads were updating
as fast as they could a table with just a few of hundreds of rows. I
guess your case is not _that_ intensive, so you can probably use much
bigger intervals.


---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster

Reply via email to