On Wed, Apr 14, 2004 at 09:23:26AM -0500, Ed Wilts wrote:
> We did the upgrade to 5.0 last night (Solaris master, Solaris and Win2K
> media servers).  In short, it was a major disaster.  We are seeing a
> *LOT* of error 84 failures.  Note that our environment was very stable
> before our upgrade but now it's tough to get our backups completed at
> all.  

To follow up my own post...  The error 84 failures were caused by a
flaky SAN interswitch link.  We repaired our ISLs and these problems
went away.  Veritas tech support was very helpful in helping us
troubleshoot this issue.

> MP1 came out last night and we applied it right away as per Veritas tech
> support's recommendation.  It was supposed to solve at least one bpsched
> core dump that we also saw last night.
> 
> This morning, after the MP1 install, we're no better.  It's a mess that
> we're working on with tech support.

It turns out that we didn't need MP1 broke include/exclude processing on
Solaris clients.  We turned on some debug logging and Veritas has
confirmed that's broken.  If you're in need of include/exclude
processing, do not upgrade your clients to MP1 - it was apparently ok in
5.0GA.

We're also having issues with Windows backups that Veritas has
confirmed (we weren't the first customer to see this one).  We've seen
the scheduler request a backup of the same drive twice in the same
stream.  If you look at the job in the Activity Monitor, you may see W:
appear twice in the same job.  It's also possible that a separate job
may be backing up W: as well.

> The admin console in Windows is much snappier, but throughput seems to
> have really slowed down.

These were our SAN issues.  We fixed the SAN and the throughput is back
to what we were getting before.

-- 
Ed Wilts, Mounds View, MN, USA
mailto:ewilts AT ewilts DOT org