Re: Execution variability

On Thursday 28 June 2007 12:00:40 Richard Huxton wrote:
> Vincenzo Romano wrote:
> > On Wednesday 27 June 2007 23:46:25 Vincenzo Romano wrote:
> >> Hi all.
> >> I understand this can be a ridiculous question for most you.
> >>
> >> The very same query on the very same db shows very variable
> >> timings. I'm the only one client on an unpupolated server so I'd
> >> expect a rather constant timing.
> >
> > What's really weird is that after some time the timings get back
> > to normal. With no explicit action. Then, later, timings get
> > worse again.
> >
> > I fear it can be a "DBA problem" but but still have no clue.
>
> So what's the computer doing? If it is the "very same" query on the
> "very same" data then something else must be happening in the
> background. Check the output of top/vmstat (or Windows equivalents)
> when this happens and see if there's a clue.

From the "top" command (I'm running Linux) the only process that jumps
high with the load is just the postrgres instance managing the SQL
connection.
I agree about "something else must be happening in the background".
All rthe available RAM gets used as well as some swap.
During "fast" operations the used RAM remains low and no swap
happens.
I would exclude any other "system" process.

How can I log what the PGSQL is actually doing?

--
Vincenzo Romano
--
Maybe Computer will never become as intelligent as Humans.
For sure they won't ever become so stupid. [VR-1988]