AW: drop distributions
Posted in 2008
Hi Jack
controlling the btree cleaner with onstat -C you could see one cleaner
working on one table all the time with type leaf. Informix support suggested
to start more cleaners with type range.
onmode -C rangesize 500
onmode -C threshold 50000
onmode -C stop
onmode -C stop
onmode -C start
onmode -C start
My impression is that some long checkpoint which where waiting on threads in
critical section disappeared.
Why do you assume that it has influence on the optimizer?
Regards,
Reinhard.
> -----Ursprüngliche Nachricht-----
> Von: informix-list-bounces@iiug.org
> [mailto:informix-list-bounces@iiug.org]Im Auftrag von Jack Parker
> Gesendet am: Dienstag, 18. März 2008 13:37
> An: Habichtsberg, Reinhard; Informix-List (E-Mail)
> Betreff: RE: drop distributions
>
> Problem may be btree cleaner threads, not the drop distributions.
>
> j.
>
> Sane ego te vocavi. Forsitan capedictum tuum desit.
>
> -----Original Message-----
> From: informix-list-bounces@iiug.org
> [mailto:informix-list-bounces@iiug.org]On Behalf Of Habichtsberg,
> Reinhard
> Sent: Tuesday, March 18, 2008 2:07 AM
> To: Informix-List (E-Mail)
> Subject: drop distributions
>
>
> Hi.
>
> After upgrading from IDS 9.40.FC4W2 to 9.40.FC9W2 some
> queries run extremly
> slow and I'm under the impression that the optimizer chooses
> the wrong way
> especially when OPTCOMPIND is 1 or 2.
>
> My idea is to drop distributions. The problem is "update
> statistics low on
> table xxx drop distributions" runs very, very long on large
> tables. So, is
> it safe, to delete the rows for the table in the sysdistrib
> table? Update
> statistics low could be run anyway. The advantage is that it
> would cost no
> time to test the optimizer without distributions.
>
> BTW: We have an OLTP and Decison Support mixed environment
> with very large
> tables. What do you think is the best value for OPTCOMPIND in
> onconfig. In
> 9.40 is not possible to "set OPTCOMPIND" with the session and we have
> applications that cannot set an environment variable.
>
> And is there any advantage of distributions if I set OPTCOMPIND to 0?
>
> Thanks in advance for your help.
>
> Reinhard.
> _______________________________________________
> Informix-list mailing list
> Informix-list@iiug.org
> http://www.iiug.org/mailman/listinfo/informix-list
> _______________________________________________
> Informix-list mailing list
> Informix-list@iiug.org
> http://www.iiug.org/mailman/listinfo/informix-list
>