Hi!
I am having really weird problem installing latest DBD::Oracle (tried
with 1.08 and 1.09) on our IRIX 6.5 server running Oracle 8.1.7.0.0. It
all worked just fine with 1.06 and .07 under Oracle 8.1.5 and 8.1.6
before, but three days ago we had to upgrade Oracle to 8.1.7 due to a
nasty bug in 8.1.6 SQLLoader and since then it all went south. I won't
go into details of build problems here -- every time I am installing
another version of DBD::Oracle under IRIX I have to edit out numerous
errors in makefile by hand, and each time it is a different set of
errors :(. This time the amount of fresh ones was really tremendous, but
after a day of hard work I managed to fix them all and the build finally
ran through successfully (or at least it looked like), with only 3
warnings from compiler. It used to generate much more warnings before
when installing DBD::Oracle 1.06 under Oracle 8.1.6, and everything
worked smooth afterwards, so I was actually very optimistic, but...
first time I started 'make test' it crashed the machine at t/plsql test
55! In my 6 years of experience with SGI/IRIX this was the first time
I've seen a system crashing so hard. At first I thought it was just a
coincidence, we rebooted the server, restarted Oracle, and repeated the
test run. Another hard crash exactly at the same point! Gees, this is
our production server and our system administrator was not happy at all.
He also has never seen this before with *any* software, only once it
happened due to a broken SCSI adapter. The machine is 4-way Octane
server with 1GB memory installed, BTW. So I downloaded DBD-Oracle-1.09
and went through the same routine of fixing makefile errors by hand
after MakeMaker. This time they were slightly different once again, and
I took it as a good sign. Unfortunately, I was too optimistic again.
Make test crashed machine hard at the same point. Twice in a row.
OK, I am really desperate now. Our project is coming into a final stage,
e.g. "go production or die" and we simply can't make it without
DBD::Oracle working and we can't backout to Oracle 8.1.6 because SGI
support people are insisting the only fix for SQLLoader is to stick with
8.1.7.
Any help is APPRECIATED. Even moral one...
FYI: I am trying this build/test under IRIX 6.5, Perl 5.6.1 n32 ABI
version, using -n32 Oracle libs (Oracle server itself is n64 binary of
course). All DBD tests prior to t/plsql 55 are passing Ok. As far as I
can see, #55 is about testing nested cursors. A nice way to crash SGI
servers! We filed a complaint with SGI support of course, but these guys
are neither fast nor impressive, IMO.
Ivan Adzhubei
---
Lowery's Law:
If it jams -- force it. If it breaks, it needed replacing
anyway.
