[Info-Ingres] Lock escalation caused by bad page estimate

[John Dennis]

Marty,

As far as I can see for "normal" tables (not partitioned) although the
relpages attribute on iirelation is updated, it is updated to the
value it previously held. I just set relpages on a table to a dummy
value (1111), ran optimizedb, and the statdump showed the page count
to be 1111.

Are you able to run this again, printing off relpages from iirelation
before and after the optimizedb?

John

On Sep 20, 5:29 pm, "Martin Bowes" <martin.bo... at ctsu.ox.ac.uk> wrote:
> Hi John,
>
> II 9.0.4 (a64.lnx/105)NPTL + patch 12343
>
> Marty
>
> -----Original Message-----
> From: info-ingres-boun... at kettleriverconsulting.com
>
> [mailto:info-ingres-boun... at kettleriverconsulting.com] On Behalf Of John
> Dennis
> Sent: 20 September 2007 05:11
> To: info-ing... at kettleriverconsulting.com
> Subject: Re: [Info-Ingres] Lock escalation caused by bad page estimate
>
> Standard support question - what version of Ingres?
>
> John
>
> On Sep 20, 12:57 am, "Martin Bowes" <martin.bo... at ctsu.ox.ac.uk>
> wrote:
> > Hi Everyone,
>
> > I have a uniquely keyed btree table with 158pages. Ithas all the
> > defaullt fillfactors etc. I have freshly modified it, I have rebuilt
> the
> > stats on it.
>
> > Why does the optimiser think there are only 40 pages in it?
>
> > Sadly, because it doesn't think there are more than maxlocks (50)
> pages,
> > it goes into page reading and then has to escalate. Which is filling
> the
> > errlog with escalation messages.
>
> > Marty
>
> _______________________________________________
> Info-Ingres mailing list
> Info-Ing... at kettleriverconsulting.comhttp://www.kettleriverconsulting.com/mailman/listinfo/info-ingres