During the test restore we found that we may face data corruption on
media since our environment matches to conditions described below:
http://seer.support.veritas.com/docs/280140.htm 

at the end we found that issues with restore were related to bad SCSI
cable (backups are 100% good, restores are 100% bad - "bad image header"
reported).
But during the troubleshooting we migrated to 5.1 MP5.

Next day we noticed an increased number of invalid images reported by
'bpdbm -consistancy 2'. Started to have troubles with bpexpdate and
bpimagelist for certain images... Investigation showed that bpdbm dumps
core during the attempt to process certain entries in catalog.
Solutions: wait until Symantec will fix the daemon with positibilty to
have increased number of broken images, downgrade to MP3/MP4 and face
potential data corruption on tapes or move on to 6.0 (were ready,
internally trained and prepared).

To clarify - I'm not 100% sure if MP5 introduced the problem with
catalog or we had one before with MP3.. Sorry, had no time to
investigate, was too scary to lose data in catalog ...

Dmitri

-----Original Message-----
From: Jack Forester, Jr. [mailto:jack.l.forester AT lmco DOT com] 
Sent: Wednesday, May 17, 2006 4:06 AM
To: Dmitri Smirnov
Cc: veritas-bu AT mailman.eng.auburn DOT edu
Subject: Re: [Veritas-bu] Upgrade to 6.0

DB corruption with 5.1 MP5?  I must be living in a shell as I hadn't
heard about that.  We're getting ready to install MP5 in a couple of
weeks.  Can you elaborate?

Dmitri Smirnov wrote:

>Still working on this one ... Overall impression - not worth it - too 
>many changes, too many bugs, documentation and support site are
useless.
>But between 5.1 MP3/4 data corruption and MP5 with DB corruption 
>Symantec leaves not much choice.
>
>Few questions:
>- nbpem is dying every 2-5  mins with segmentation fault (Linux 2.4). I

>may be wrong but it can be related to attempts to forward email using 
>clients... should be fixed in MP3 - is anyone had same problem or have 
>a fix?
>- few libraries transferred w/o problem to EMM, one finished with all 
>tapes in Netbackup pool... Is anyway to assign tapes correctly to right

>pools and recover the rest of mediaDB information?
>- Jobs are going bananas... Bytes are growing on jobs in 'Queued' 
>state, they go to 'Completed' state after backup is finished and 
>restart after that once again...How to fix that one?
>- Firewalling changed big time.. Hope I'll fix that based on Symantec 
>documentation.
>- Is only 'root' allowed to use Java admin interface remotely w/o NBAC?
>
>Dmitri
>
>_______________________________________________
>Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu 
>http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
>  
>


--
Jack L. Forester, Jr.
UNIX Systems Administrator, Stf
Lockheed Martin Information Technology
(304) 625-3946