Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.misc -> OCI: olog/orlon core dump problem on AIX
Dear all,
I read the following message posted by Charles Hooks to comp.databases.oracle.misc a bit more than a month ago. I emailed him last week to find out whether he had received any suggestions or solutions to his problem, since I am also having trouble with a similar issue. I haven't had a reply, so I take it he is on leave (being in the "business" of education)
Does anybody have a clue how to tackle this problem ?
Thanks a lot.
Willem MOORS,
Belgium
>From shire_at_ecst.csuchico.edu (Charles W. Hooks)
Organization California State University, Chico
Date 28 May 1997 05:09:23 GMT Newsgroups comp.databases.oracle.misc Message-ID <5mgem3$agt_at_charnel.ecst.csuchico.edu> I've had several problem with the shared libraries on variousplatforms, previously a patch for Oracle has fixed the problem, after one phone call. However Oracle is taking a long time to get back to me about the AIX platform... So I thought I'd see if anyone else on the net has had my problem, and could help me lead Oracle support in the correct direction of a patch, or at least help speed a patch into existence... :)
The commmon problem for many platforms has been for a core dump
to occur durign an 'orlon' call when linking with shared libraries. If
you switch over to linking statically the problem goes away. Thus I can
get around the problem until Oracle comes up with a patch. Of course
Oracle and thier documentation says to use 'olog', however we still have
an support call in on olog problem, which I'll tackle another day. :)
On the AIX machines however, the static linking has a bad side-effect
that prevents my product from working. :( I use cda.rpc to report how
many rows are fetched. Works great on other platforms. On the AIX
boxes,
however I am gettign bogus data... A consistently large negative
number
in some cases, which I treat as an error condition. This bad value
occurs
just after I finish getting all the data(rows) from the select. From
my testing, I've noticed that I can jsut treat that large ugly number
as an '0' and things do work. However since the application work on
several RDMS and platforms, this 'magical' hack is not appropriate.
Besides, I hate stuff that work magically! Leave to much area for
something else to go wrong and/or compound the problem...
Has anyone come across this problem before? If so, did Oracle build a patch for it?
-- Charles W. Hooks aka Tribbles _ /| aka Moonshae \'o.O' aka (shire_at_ecst.csuchico.edu) =(___)= California State University, Chico UReceived on Mon Jul 07 1997 - 00:00:00 CDT