RE: Constant corruption in DB2-UDB on the i5.

["Jim Damato" <jdamato@xxxxxxxxxxxxxxxxx>]

I've always been frustrated by the concept of damaged objects.  Here we
have a file, and we can read/write/change the data.  We just can't back
it up.  I wonder whether there's a good reason why a damaged object
can't be backed up with a warning, and why even a restore w/damages
might or might not be permitted with the appropriate errors and
warnings.  And if we have to have a message every stinkin' time a job is
held or released, why can't we have a message every time a damaged
object is encountered?  It would be a lot smarter than having to dig
into the job log of the failing backup.

I was overly preachy about the standards for monitoring backups, sorry.

A message that would flag when damage occurs?  Hey!  That would be
something.  I don't think the process or the OS knows about the damage
when it happens -- it's kind of like finding the mangled tricycle in
your grill when you come out to your driveway in the morning.  I think
we found our damage by noting the type of damage based on a message in
the job log -- indicating a process that was aborted.  Then we scanned
QHST's CPF1164 messages looking for end code of 50.  We had successfully
backed up the files in question on Saturday night, then hit the error on
Sunday night's backup.  So we had a good idea of the time-frame.  It's
always enlightening to scan CPF1164's for unanticipated end codes.
Otherwise you'll never know who's doing something really, really scary.

-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx
[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of rob@xxxxxxxxx
Sent: Tuesday, April 11, 2006 1:53 PM
To: Midrange Systems Technical Discussion
Subject: RE: Constant corruption in DB2-UDB on the i5.

You are right on regarding the CPF3777.  There is a project request to 
modify the backup to perform better handling of the CPF3777.  We are 
rightfully open to your criticism.

Yes, operations should be tailored to gracefully exit.  I don't know yet

if I want to take responsibility for damaged objects.  I would hope that

some nice error recovery in database would handle all of these, except 
perhaps catastrophic disk failure.

I don't believe that we are accessing that ldap directly with any of our

software.  For our use we normally use our Domino driven ldap.  Shutdown

of i5/os ldap currently involves ENDTCP.  Any other suggestions welcome.

Would there be a QHST, or a QSYSMSG, message that would flag when damage

occurs?  That would be nice.  What would that be?

Rob Berendt