On 12/13/2006 9:09 PM, JAJA (Jamie Jamison) wrote:
> I'm running NetBackup 5.1MP5 on a Sun V880 server with 4CPUs and 8Gb of
> RAM. My tape storage is IBM Ultrium LTO Gen 2 drives on a fiber
> connected L180 library. Lately I'm seeing a lot of jobs requeue when I
> watch my backups kick off. Other jobs just queue up and sit there but
> some jobs queue and then requeue a few times before finally starting.
> What I see when I look at the status for the job is something like this:

This is a design issue in 5.1 and is caused by the master and media 
servers having different views of what drives are really available. 
Fixing this type of problem is one of the major features in 6.0. 
There's nothing you can do until you upgrade.

> This confuses me. I assume that NetBackup is actually trying to mount
> this tape and then finding that all of the drives are in use. Is there a
> communications breakdown here? Why isn't NetBackup just queueing these
> jobs until a drive and tape are available. I haven't seen any failures
> yet, but it concerns me and I don't understand why some jobs start and
> sit in the queue until a drive is ready and then start and others
> requeue.

The jobs don't really fail under 5.1.  After you upgrade, you won't see 
these at all.  The root issue is that under 5.1, the media servers keep 
track of tape drives so the master doesn't really know if there's a 
drive available or not.  With 6.0, the master keeps track of all of the 
drives across all of your media servers.

        .../Ed

-- 
Ed Wilts, Mounds View, MN, USA
mailto:ewilts at ewilts.org