Trying to use QLogic card with C7200 DLT library in SAN environment

[Fedon Kadifeli]

We recently converted our SCSI DLT-8000 library to an FC library, connected to SAN and trying to get HP DataProtector backups of some Windows servers. Hardware configuration is as follows:

FYSRV: Win NT 4.0 SP6 servers running DP media and disk agents with 2 x QLogic QLA23xx PCI FC Adapters (one is used solely for disk I/O, the other is used solely for tape I/O).

SAN Switches: 2 x 2109-F16 IBM SAN Switch Model F16. These are actually Brocade switches.

DLT library: HP SureStore C7200-A5587A 60 slot, 4-drive library with 2 Fibre Channel cards.

Connection is as follows:

FYSRV + IBM (Qlogic) FC-133 HBAs
->
FC connections
->
2109-F19 IBM (Brocade) SAN Switches
->
FC connections
->
HP SureStore C7200 DLT Library.

DataProtector version is 5.00 and Cell Manager runs on HP-UX 11.00 box.

For backups that span multiple tape cartridges we CONSISTENTLY (I tried with different NT servers, different QLogic cards, different FC paths, different tape drives and cartridges) get the following errors:

....
<<< backup starts >>>
....
<<< first cartridge is filled up >>>
....

[Normal] From: BMA@fysrv "FYSRV-DLT3" Time: 24/07/2004 04:59:57 PM
Unloading medium to slot 18 from device scsi3:0:1:0C

[Normal] From: BMA@fysrv "FYSRV-DLT3" Time: 24/07/2004 05:00:40 PM
Loading medium from slot 17 to device scsi3:0:1:0C

[Major] From: BMA@fysrv "FYSRV-DLT3" Time: 24/07/2004 05:03:23 PM
[90:180] Cannot rewind medium. (Details unknown.)

[Critical] From: VBDA@fysrv "N:" Time: 24/07/2004 05:03:23 PM
Received ABORT request from SM => aborting.

[Critical] From: VBDA@fysrv "N:" Time: 24/07/2004 05:03:23 PM
Connection to Media Agent broken => aborting.

The problem (I think) is not in DataProtector, but in the FC connectivity. Presumably some misconfiguration in QLogic HBA's. Has anybody used such a configuration without problem?

[Fedon Kadifeli]

UPDATE:

We replaced the QLogic HBA with HP A5158A HBA (Windows NT 4.0 detects this card as "Agilent PCI Fibre Channel Controller"). Exactly the same problem occured:

[Normal] From: BMA@fysrv "FYSRV-DLT1" Time: 26/07/2004 05:43:19 PM
scsi4:0:0:1C
Medium header verification completed, 0 errors found

[Normal] From: BMA@fysrv "FYSRV-DLT1" Time: 26/07/2004 05:43:19 PM
Unloading medium to slot 24 from device scsi4:0:0:1C

[Normal] From: BMA@fysrv "FYSRV-DLT1" Time: 26/07/2004 05:44:04 PM
Loading medium from slot 26 to device scsi4:0:0:1C

[Major] From: BMA@fysrv "FYSRV-DLT1" Time: 26/07/2004 05:46:08 PM
[90:180] Cannot rewind medium. (Details unknown.)

[Critical] From: VBDA@fysrv "N:" Time: 26/07/2004 05:46:08 PM
Received ABORT request from SM => aborting.

[Critical] From: VBDA@fysrv "N:" Time: 26/07/2004 05:46:08 PM
Connection to Media Agent broken => aborting.

[Fedon Kadifeli]

Next UPDATE:

I tested the same configuration bypassing the SAN switch and connecting the HBA directly to the library. The error did NOT occur. It looks like we need to do something for SAN environment.

Does anybody know whether I have to do any configuration changes in the SAN switch ports and/or HBAs?

[Marino Meloni_1]

hi,

Generaly speaking, when you have a library in a SAN, all the servers connected to it could potentialy take the ownership of the tapes drives, and probably what it happens is that when the server who is writing on the tape release it because it have to load another one, another server take the ownership of this unit, and send a Reserve command in order to be sure he can manage exclusively the drive. To recover of this situation, or the server who sent the "reserve command" issue a "Release command", or you should reboot the library.

This kind of issue in a eterogeneous environment happens frequently, and for that it is recommanded to stop all the software that can cause interferrences with tape drives, like (for windows environment) insignt manager FC agent(or Toptools specifics agents, depending of the platform), Removable Storage manager, and other backup software that may be active on all servers on the san, also for UX, there is some advise to stop some sw.Some time, a bad communication or problem in the software between the media agent and the cell server may cause also thit behaviours.

What you can do to troubleshoot or to reduce the problem is to stop all this kind of sw that may disturb the Drives, and create a "Backup zone" with only the servers that have manage the drives, and the robotic.