tape devices unavailable

[Erkan Saygili]

Hi,

I have a trucluster(three member cluster) 5.1b pk4, where the connection from the server to the drives are via a san switch to a NSRouter, connected to two MSL5000 series autochangers (1master & 1 slave). Using in total of 8 SDLT320 drives.
The Backup software is NetWorker v7.1.2.
We have been a lot of issues lately where the driver owner number is set to 4 on one member. Some of the things i see are:

# hwmgr -view hier -id 445
HWID: hardware hierarchy (!)warning (X)critical (-)inactive (see -status)
-------------------------------------------------------------------------------
445: X tape bus-3-targ-2-lun-1 tape0


# hwmgr status comp -id 445

STATUS ACCESS INDICT
HWID: HOSTNAME SUMMARY STATE STATE LEVEL NAME
------------------------------------------------------------------------------
445: critical online unavailable tape0


However when i run the same command on another member everything comes back ok.

I have run:
hwmgr refresh comp
hwmgr refresh scsi
hwmgr scan scsi

across all three members.

Here we go, is there anyway without a reboot :) where the tape devices can be re-initialised through a hwmgr command? I've been looking through manuals and nothing seems to work for these tape devices. I may have missed something?
I have also had the autochangers power-cycled

Cheers,

Erkan.

[Johan Brusche]

Is the KGPSA (dedicated to tape traffic ?) on this server still OK, ie: see binary.errlog ?

Is switchport OK, is zoning on switch still OK, see SupportShow on switch ?

Is the NSR's mapping still OK, firmware up-to-date ?

__ Johan./

[Ivan Ferreira]

It seems that you lost connection with the tape device, I get that output when I remove the device from the system.

What is the output of hwmgr show scsi?

126: 16 tape none 0 1 tape0 [0/3/0]
127: 3 tape none 0 1 tape6

Tape0 has a valid path and tape6 does not. This is because I have lost the connection with tape6.

If you delete the device:

hwmgr delete -id 445

And then run hwmgr scan scsi, the device appears again?

[Erkan Saygili]

Thanks Ivan, although this is not the case.

Johan, the NSR mapping is ok, and the firmware is all up to date.

I will check on your other suggestions though.

Thanks,

Erkan.