Up Prev Next
From: Roleigh Martin <rol at uhc dot com>
Subject: Re: after 5000 dbe deletes, internal directory control layer
error - why?
Date: Mar 30 2001 3:48PM
Thanks! I originally posted the message to both lists at the same time but
got a bounce-back on the first and reposted to the correct email address
the second time--wanted to know if anybody had suffered the problem.
My top-notch DBI perl programmer read your reply and he surmises that he
needs to do a bind each fetchrow rather than either (a) a do() or a (b)
(prepare/execute/finish) each fetchrow loop. He guesses that the finish
command must not be cleaning up all stuff. He doesn't see where he is
accidentally opening another connection each cycle however.
I hope I interpreted his reply correctly.
We're going to try the bind option and we'll let yo know what we find.
Also, we've submitted a request to the systems group to update our Sybase DBD
and DBI modules to the latest versions.
Last, we've had problems using the CT-Lib Sybase 12.x libraries -- and so we're
stuck using the CT-Lib Sybase 11.x libraries with our current versions of
stuff. The databases we're going against are Sybase 12.x version. Is this
the norm for the versions we're using?
Thanks again, Michael!
Specs about the program versions/machine, etc:
rol/ftp/perl>perl -e "print join('; ',
Adaptive Server Enterprise/22.214.171.124/P/SWR 9268 ESD 3/RS6000/AIX 4.3.2/1629/
32bit/FBO/Fri Sep 8 01:43:53 2000
IBM RS/6000 AIX 4.3
# $Id: Sybase.pm,v 1.22 2000/03/24 04:44:49 mpeppler Exp $
# $Id: CTlib.pm.in,v 1.30 2000/05/13 22:50:56 mpeppler Exp $
# $Id: DBD.pm,v 10.7 1999/05/06 17:29:14 timbo Exp $
# $Id: DBI.pm,v 10.25 1999/07/12 02:02:33 timbo Exp $
At 12:38 PM 3/29/2001, you wrote:
>Roleigh Martin writes:
> > We have a DBI perl program that after about 5,000 deletes we get this
> > error message -- something similar happened with another program that did
> > about 5,000 updates. The update program worked around the problem by
> > removing use of the second database handle. The program logic basically
> > has a parent SQL select command that uses one database handle to get a
> > list of unique index keys of records to update, then inside a while
> > next record loop, it does the update or delete for only one database
> > at a time to minimize database locking.
>I answered this on the other list where you posted - did you not see
>Michael Peppler - Data Migrations Inc. - firstname.lastname@example.org
>http://www.mbay.net/~mpeppler - email@example.com
>International Sybase User Group - http://www.isug.com
>Sybase on Linux mailing list: firstname.lastname@example.org