Yura Pismerov wrote:

> "Porotikov, Vitaly [IT]" wrote:
> >
> > Yuriy,
> >
> >    Did you try to kill nsrmm process which is responsible for this tape
> > drive?
> >
> > give a try:
> >
> > kill -9 `lsof |grep /dev/rmt/<idle devcie>|awk '{print $2}'`
>
> Thanks. My point is there should be no manual intervention in the
> process.
> I have analyzed the messages (nsradmin->NSR jukebox->messages) and I
> found that the problem occurs only when nsrjb request involves volume
> labeling (ie. a new/blank tape is requested).
> Last night backup did not label any tapes so it went smoothly.
> My question is, is it expected behavior when request for labeling is
> treated like this ?
> It looks like Networker proceeds with it only in case there is an empty
> drive by that time.
> Otherwise it waits, no matter if the tape that is currently in the drive
> is idle and can be unloaded.
>
> >
> >  It helps for me. (Linux RedHat Legato 6.1.3)
> >
> >   Regards, Vitaly
> >
> > -----Original Message-----
> > From: Yura Pismerov [mailto:ypismerov AT TUCOWS DOT COM]
> > Sent: Monday, April 21, 2003 10:23 AM
> > To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
> > Subject: [Networker] "Waiting for 1 writable volumes"
> >
> >         Sometimes (not always) after a group start Networker keeps asking
> > for a
> > tape that could be mounted on second drive that is idle at the moment
> > but for some reason it does not eject the tape from another pool that is
> > already in the drive.
> > So instead of using 2 drives it ends up with one drive. I watch nsrjb
> > process that is issued for the second drive and keeps looping/trying.
> > Does anybody have an idea what is wrong ? There is not much in the logs
> > files that could shed light on the problem.
> > The version of Networker is 6.1.3 (running on Solaris 9).
> > Interesting thing is, if I kill the queued nsrjb process and issue
> > manual tape umount (nsrjb -u), the requested tape gets mounted right
> > away next time it retries.
> > What do I do to troubleshoot the problem ?
> >
> >         TIA.
> >

Yuriy,

 I guess this is problem of particular library.

In my case (Storage Tek L40) each nsrjb labeling request goes to  nsrjb queue   
 and has time-out value.

If tape drive is busy this current request will  die after time-out.

We do tape labeling via script. This script tries to label blank tape several 
times (3) and if it's not lucky goes further.

I think mentioned above situation arises in case we have

 - more then one nsrjb (any tape library, for instance savegrp ) requests,

  - at least one free tape drive,

  - jukebox parallelism is more then 1 (set up automatically and Legato doesn't 
recommend to change it)

   - jukebox incorrect resolves  some sequence of nsrjb(any tape library) 
commands

On the other hand, Legato has to have a remedy for this non standard (for 
Networker) situation.

--
Yours,  Vitaly Porotikov aka vp12929 on ws22a963

 Trap full -- please empty.

--
Note: To sign off this list, send a "signoff networker" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=