QUOTE - "Another major one is some of the RMAN jobs scheduled
through GC just disappeared last week. I've been ranting about OEM's
reliability since 8i. I'm still far from satisfied, unfortunately."
We ran into this bug and I worked with Oracle for months to try to reach a resolution. We have since upgraded the OMS to 10.2.0.4 and that has seemed to fix the problem.
Before upgrading, I ran this query against the repository database daily. It should show any jobs that are supposed to repeat, but are not.
SELECT j.job_name
FROM mgmt_job_schedule s, mgmt_job j
WHERE s.frequency_code > 1
AND s.schedule_id = j.schedule_id
AND j.job_status = 0
AND j.nested = 0
AND j.is_library = 0
AND (s.end_time > SYSDATE OR s.end_time IS NULL)
AND NOT EXISTS
(SELECT 1
FROM mgmt_job_exec_summary e
WHERE e.job_id = j.job_id
AND e.status NOT IN (3,4,5,8))
ORDER BY j.job_owner, j.job_name
Rob Pegram
Duke University
- Original Message ----
From: Rich Jesse <rjoralist_at_society.servebeer.com>
To: Chris.Taylor_at_ingrambarge.com
Cc: oracle-l_at_freelists.org
Sent: Thursday, November 29, 2007 3:02:47 PM
Subject: Re: Anyone else experience this?
More often? Not sure, but I do know that since my intro to Oracle back
on
7.3.4, I've been an unknowing accomplice of bugdom in our database by
utilizing more functionality of Oracle. A few that come to mind:
- CURSOR_SHARING=FORCE(SIMILAR)
This has caused me the most grief of anything I've used in Oracle,
including the solitary "wrong results" (9.2.0, I think). At the time,
though, it was the best solution, because we weren't taught the
importance of bind variables in our SQL classes. I know better now!
- OEM/Database Control/Grid Control
4/6 of my current SRs are about Grid Control 10.2.0.3. And I have
issues that I haven't started an SR because there is no reference to
any
part of the error messages on Metalink or even Google. And with the
pain and time I seem to have with relatively "known" bugs, I
unfortunately can't afford to blaze the trail for new ones. One major
bug I did file included a metric that failed to collect, and failed to
notify the collection failure, which led to a full tablespace, and a
discrepancy in our financials from the failed transaction. This was
never resolved and Support relayed that they are not able to allocate
time for root cause analysis (apparently I'm supposed to debug their
software now?). Another major one is some of the RMAN jobs scheduled
through GC just disappeared last week. I've been ranting about OEM's
reliability since 8i. I'm still far from satisfied, unfortunately.
- Oracle Client on Windows
In a past life, we used homegrown client/server apps (e.g.
VB/.NET).
8.0.5 was very stable. Very few issues that we could attribute to the
Oracle Client. The other two versions we replaced it with, 9.2.0 and
then 10.2.0, were buggy to the point of not being usable. I had an SR
for the 9.2.0 client open for 10 months (escalated to Development for
most of that time) that was never resolved.
- Oracle UltraSearch
It's probably that I'm just not a Java-type (or class!). It would
work
great for weeks (10gR1, I think?!?), then puke unexplicably, throwing
30+ errors/second until the drives filled. Only a bounce could fix it.
The next *patch-level* release of the database included a fix, but also
*introduced* (in a patch-level release!) a different third-party engine
for the Text portion, if I remember correctly. Because of the testing
requirements of that huge change, the patch could not be justifed to
the
business.
- Oracle Internet Directory
Version 9iR2 (or was it even 10gR1?) was not usable with replication
on
Linux. It flat out didn't work, and Support couldn't tell me what was
wrong or how to fix it. To be fair, 10gR2 seemed nicely stable with
the
newer (and simpler!) LDAP-based replication agreements instead of the
OID
Advanced Replication. And at least one person on the OID Support team
was downright fantastic in helping me troubleshoot the miniscule error
in
my install (and I left feedback about it, too).
All that being said, I still think the core database is the best out
there,
bar none. It's just that I don't feel that 11 years experience with a
free
safety ringing my bell after catching a high pass in the flat should be
needed to know what options and features I should not use (Sorry for
the
football reference, but the Superbowl of the NFC is on tonight. GO
PACK!).
My $.02000000001, (w/rounding error)
Rich
p.s. Best of luck to you in kicking the habit!
> Do you find yourself saying more often, "It's an Oracle bug. Fixed
in
> 'xx' patch." Or "It's an Oracle bug and its not fixed yet." Or
"It's
> an Oracle bug and it looks like its fixed in 11g."???
>
>
>
> We're running 10.2.0.3 (can't remember the patchset off the top of my
> head) and we've got a few bugs that aren't fixed and we seem to hit
> another every other week.
>
>
>
> The latest is ORA-01008: Not all variables bound when setting
> cursor_sharing = SIMILAR for PeopleSoft performance.
>
>
>
> I just looked at the 10.2.0.4 release notes (which has not been
released
> for any platform, nor are any release dates provided) and I think
it's
> the biggest list of Oracle bugs I've ever seen in a patch.
>
>
>
> I know with added complexity comes added potential for bugs, but this
is
> becoming more and more bothersome.
>
>
>
> Thoughts?
>
>
>
> Disclaimer:
>
> I've recently quit smoking and taking Chantix so maybe I'm just in a
> very bad mood.
>
>
>
> Chris Taylor
--
http://www.freelists.org/webpage/oracle-l
____________________________________________________________________________________
Get easy, one-click access to your favorites.
Make Yahoo! your homepage.
http://www.yahoo.com/r/hs
--
http://www.freelists.org/webpage/oracle-l
Received on Thu Nov 29 2007 - 14:49:33 CST
