MSL6000 Library - Offline

[Kehad Snydewel]

My customer has a MSL6000 lib on the SAN and 4 servers (Windows). Veritas is the backup software. When ever a scheduled backup starts the tape drives goes in an offline state in veritas and the backups then fail. This happens within less than a minute. Please note that this setup did work but for the pass 3 days the drives go in an offline state in veritas if a backup starts.

I also checked in device manager and the tape drives + media changer also disappears. I then have to power cycle the lib and reinsert the FC cable for device manager to detect the tape drives + media changer. When I start the backup again, still the same problem (drives go offline).

I changed the GBIC on the NSR and Switch port GBIC and also used another switch Port but still the same issue. Removable storage on all windows servers is disabled. I did a firmware update (NSR, Drives and Library) and library is on its latest firmware revision. No errors on the switch port and library event log. I also checked the NRS for errors but could not find any appropriate errors. Zoning is not configured on SAN switches.

please assist asap.

thanks

[Marino Meloni_1]

not easy to understand why it disappear, it looks like the path is not stable, so when the backup application start managing the drive, the lack of communication make the drive to be put offline.

You should analyse the logs from the NSR and the switch in order to understand where the problem come from

Calling HP support may help you to indentify the problem

[Luk Vandenbussche]

It is best / advised to implement zoning on your switches.

I suppose each server has two fiber cards?
So each tape drive is seen twice by each server. Perhaps this causes the problem.

Check also the presentation of the hosts on your NSR

[Kehad Snydewel]

i will go through the logs of the NSR and SAN switch again. the NSR is connected to Fabric A only. i will also replace the FC Cable.

[Vihang K.]

Hi Kehad,

It would be advisable to have zoning in your environment as you have windows servers.
The problem could be happening as the server which scans the drives & robots first must be reserving the drives.
Veritas uses the SCSI reserve/release features on the host.
There is a chapter on SCSI Reserve/Release in Veritas Netbackup Admin guide.
You can try scsi releasing the drive before starting a backup.
One thing you can try (if this is not a productions setup) is to shutdown 3 hosts, reboot the library & then try the backup from the 4th host. If the backup happens, this is probably a scsi reserve issue.