How to avoid master/media servers reboot after SAN LTO2 rebooted?

[CheeHoong_CHONG]

LTO2 drives, master/media servers using SAN fibre channels over SAN switch. Everytime a hardware problem with a drive, ie power off/on resulting in master/media servers cannot 'see' the affected drive. Is there a solution to avoid the need for reboot? This is Win2k platform with NBU5.0 MP3 ES.

Thanks in advanced.

[Stumpr2]

I can't say for sure that this will help but at least it is good reading that may initiate further thoughts.

Troubleshooting Device Manager initialization problems in a Storage Area Network after device communication failures
http://seer.support.veritas.com/docs/272091.htm

How to delete "ghost" or "phantom" devices from the Windows 2000 operating system
http://seer.support.veritas.com/docs/259595.htm

Another thing to consider is the vmglodb
There are several technotes on it and its use.
How to rebuild a corrupted VERITAS NetBackup (tm) Shared Storage Option configuration from 'scratch'
http://seer.support.veritas.com/docs/263343.htm

DOCUMENTATION: How to rebuild the globDB (global device database file) on the master server
http://seer.support.veritas.com/docs/243792.htm

The tape drive serial number or the firmware version is not correct in the global database.
http://seer.support.veritas.com/docs/259899.htm

[Scott_Murray]

Hi there,
We have had similar problems when a Tape Library is switched off/on in a large SAN. We have SDLT tape libraries, and Windows 2003 with NBU 5.1 MP3.
After much investigation, the only resolution was:
1. Ensure HBA firmware and driver is up-to-date.
2. Use the Veritas Plug-n-Play tape device driver (I think there's one for LTO - not sure ??) Refer Veritas TechNote #270183.
3. Implement the Microsoft TUR disable - Refer Veritas TechNote #268245, and MS TechNote #842411.
4. Implement Microsoft HotFix as per TechNote #873337.
5. Shutdown NBU services on all Media Servers/SAN Media Servers.
6. Must do a manual Device Manager hardware scan under Windows OS after device comes back on-line.
7. Bring Master Server NBU up and detect devices.
8. Bring other NBU Servers up.

This relates more to Win2003, but there may be some useful tips here. We manage a very large environment with many SAN Servers and spent much anguish trying to get the best result.
This is as good as we could get.

Other issues which caused concern are:
we had HBA's which had tape and disk shared (not optimum, but couldn't change this). Also we were unable to implement Persistent Binding.

Hope this might be useful, although not quite the same as your environment.
Additional comments or findings would also be appreciated.

[CheeHoong_CHONG]

Scott,
You mentioned persistent binding. I read something about it but don't quite understand how it works. We are using QLogic HBAs at Win2k servers.

[patrickkuah]

When you startup a machine, the OS will assign a SCSI id to the WWNN of the FC device it first discovered. If you used persisten binding, the OS will assign the same SCSI id to the WWNN the next time you reboot the server so your Netbackup SSO device database will be consistent ;)

The Qlogic HBA should have a utility to configure persistent binding.

Same with me here, got no change to configure persistent binding so everytime i have to reboot the servers once the tape library is off and on.

patrickkuah

[CheeHoong_CHONG]

Patrick, sounds great, why do you say you have no chance to implement persistent binding as the benefits were spelt out clearly?

Anyone implemented persistent binding on Win2k servers and care to share your experience?

Thanks

[patrickkuah]

well.... my job is to strictly implement Netbackup, that all. can't really touch anything related to storage as it is setup by someone else....

You should have SANsurfer HBA Manager installed if you are using qlogic card. Use that to do persistent binding :)