On Mon, 2008-01-28 at 14:36 +0100, Cyril SCETBON wrote:
> 
> Simon Riggs wrote:
> > On Thu, 2008-01-17 at 15:52 +0100, Cyril SCETBON wrote:
> >
> >   
> >> Here are the plots of several metrics taken with the original slon 
> >> binary (25 september 2007) and with the new slon binary (including 
> >> modifications of static variables, 16 october 2007)
> >>     
> >
> > Cool graphs.
> >
> > First graph the effect you highlighted is shown very clearly.
> >
> > Second one may need repeating because if the performance really does get
> > a lot worse *sometimes* then we may want to go back.
> >
> > Does this include Chris's latest changes to avoid VACUUM altogether?
> >   
> no.
> 
> NB: sorry I think my email was in a moderated state.

Cyril,

Sorry, I don't understand that.

Your earlier graphs highlight the problem very well. Chris mentions that
the proposed solution has already been committed, based upon my
analysis. 

Re-running the tests with the latest version of the software would help
confirm that the problem has gone away. Would you be able to repeat the
tests to confirm that the response time no longer peaks periodically as
it did before?

-- 
  Simon Riggs
  2ndQuadrant  http://www.2ndQuadrant.com