В сообщении от Sunday 18 November 2007 05:00:35 Scott Marlowe написал(а):
> On Nov 16, 2007 11:59 AM, <xeb@xxxxxxx> wrote:
> > Hello!
> > Process postmaster completly eat my proccessor for a long time and i see
> > that message in logs.
> > Does anybody know what does the subj means and why it occures ?
>
> You're giving us WAY too little information to troubleshoot this problem.
>
> What message in the logs? What does top / vmstat / ps / iostat really
> show? What kind of query is running? Have you got query logging
> turned on for long running queries? Is this actually affecting the
> performance of your machine? Is the process chewing through all your
> memory? If you know what query is causing this, what does explain
> analyze of the query look like?
>
> The more info you can provide about the issue, the more we can help.
> But first we have to be sure there's a problem.
>
> If you're asking postgresql to sort 1billion rows it is going to eat
> up a lot of CPU and there's little you can do about it.
OK, when it occures again i'll collect all information.
> Have you got query logging
> turned on for long running queries?
How can i do it ?
> Indeed, but it seems to have something to do with a broken selectivity
> estimator function (see restriction_selectivity()). What PG version
> is this? Do you have any add-on datatypes or operators installed?
> Exactly what is the query that triggers the problem?
PG 8.2.4
Database containes phpBB forum, jabberd-2 and gentoo-wiki mirror and who
causes trouble and what query i don't know.
I have droped gentoo-wiki yesterday and trouble did not occured yet, it seems
gentoo-wiki was causing, but i want to restore this database.
As you may know this database is large enough.
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
