It's a 4 processor Intel xeon machine with more than enough ram. The
entire database can fit in memory, and while the CPU is pegged,
nothing is chewing up I/O bandwidth, and nothing is getting swapped
out of RAM.
I'm running Debian stable with only a few tweaks to the kernel's
memory settings. As far as I'm aware, I have not changed anything
that would impact scheduling.
Other queries do respond, but it's more like every couple of seconds
one query which normally takes 300ms might take 8000ms. Nothing
terrible, but enough that our users will notice.
Bryam
On 8/2/07, Alan Hodgson <ahodgson@simkin.ca> wrote:
> On Thursday 02 August 2007 09:02, "Bryan Murphy" <bryan.murphy@gmail.com>
> wrote:
> > My question: Is there a way I can decrease the priority of a specific
> > query, or determine the PID of the process it is running in? I'd like
> > to throw together a quick shell script if at all possible, as right
> > now I have to monitor the process manually and we'll have fixed the
> > problem long before we have the chance to implement proper database
> > clustering.
>
> select procpid from pg_stat_activity where current_query
> like '%stored_proc%' and current_query not like '%pg_stat_activity%';
>
> requires stats_command_string to be enabled
>
> I'm surprised your operating system doesn't automatically lower the priority
> of the process, though ..
>
> --
> "Remember when computers were frustrating because they did exactly what
> you told them to? That actually seems sort of quaint now." --J.D. Baldwin
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 6: explain analyze is your friend
>