Re: [Info-Ingres] Optimization Problem on Ingres2006 database

We also have the same problem with 9.1.0 (R2) on Linux Itanium. A select
based on any date field will
cause this unless the stats are either dropped or ignored.
 
Andre

________________________________

From: info-ingres-bounces@[EMAIL PROTECTED]
 on behalf of Paul
White
Sent: Fri 25/04/2008 6:51 AM
To: Ingres and related product discussion forum
Subject: Re: [Info-Ingres] Optimization Problem on Ingres2006 database



I sounds like we had the same bug - 119583
Talk to ingres sup****t.  The turnaround was very fast.


Problem.
Simple select on table recently moved from 2.6 to 2006 using unloaddb

Server errlog.log
E_OP04AA_HISTOMOD Histogram modification failed. Original histogram will
be
used.

E_OP03A2_HISTINCREASING consistency check - histogram on
ta_employee_clockings attribute actual_clock_datetime is not monotonically
increasing

client
select clocking_source, actual_clock_datetime
from ta_employee_clockings where employee_no= 123
and employee_attend_id= 456
and (actual_clock_datetime= '07/07/2007 10:00:00' or
actual_clock_datetime= '07/07/2007 11:00:00')

E_OP03A2 consistency check
    - histogram on ta_employee_clockings attribute actual_clock_datetime
is
    not monotonically increasing

optimizedb did not fix.
verifydb did not detect problems.
restored database to another machine worked. (sorry not sure version)

Tem****ary fix was to remove stats or optimize with -zna flag. eg
statdump -zdl  mwp -rta_employee_clockings

version.rel
II 9.1.0 (int.w32/123)

Paul White

-----Original Message-----
From: info-ingres-bounces@[EMAIL PROTECTED]
 Behalf Of
weasel
Sent: Friday, 25 April 2008 8:20 PM
To: info-ingres@[EMAIL PROTECTED]
 Re: [Info-Ingres] Optimization Problem on Ingres2006 database


On Apr 25, 6:06 am, marcopolo...@[EMAIL PROTECTED]
 wrote:
> We have some optimization statistics problem.
>
> First we run optimizedb -zk sales
>
> But then we get this messages on every table we try to use on our
> application.
>
> E_OP03A2_HISTINCREASING      consistency check
>     - histogram on taliprovta attribute idmes is not monotonically
> increasing
>
> We fix this running an optimization at 50% with this command:
> optimizedb -zk -zs50.0 ventas -rtaliprovta.
>
> So at noon we run an optimization like this optimizedb -zk -zs50.0
> sales, but now the problem persist not in all the table but more than
> a 70% of all the database, we are at stage of migration from IngresII
> 2.6 version, and think maybe in this new version of ingres is diferent
> form of optimization somebody can give us a clue?
>
> Thanks in advance.

Can you post the statdump for one of these tables and columns?
_______________________________________________
Info-Ingres mailing list
Info-Ingres@[EMAIL PROTECTED]
 mailing list
Info-Ingres@[EMAIL PROTECTED]