SAN failover problem after applying NW65SP8

[toomas_aas]

Hello!

We have a simple SAN consisting of MSA1500cs to which 3 servers are attached in dual path configuration:

1. Netware 6.5
2. Linux (SLES10 SP2)
3. Windows 2003

So far, SAN failover has always worked flawlessly. I have tested this multiple times by disconnecting the FC cable from the active MSA controller. This weekend I applied Support Pack 8 to the Netware server (which previously had SP7). Last night I needed to perform a SAN failover for maintenance purposes. Linux and Windows servers failed over successfully, but the Netware server lost connection to it's LUNs, giving this error message:

NWPA-004: The CDM driver deactivated device [V599-A2-D0:4] COMPAQ MSA1000 VOLUME f/w:5.20 due to a device failure.

Interesting thing is that this happened almost 10 minutes after failover. Since it was late night it is not entirely impossible that no IO was attempted for 10 minutes, but still hard to believe.

Anyway, I noticed that SP8 update had updated the QL2X00.HAM driver from version 6.90.15 to 7.00.01. I'm inclined to just put the old driver back, unless someone has other ideas or can convince me that this was just a random fluke :)

[toomas_aas]

After changing back to old driver (QL2X00.HAM ver 6.90.15) the problem is gone. I just performed another failover last night and it worked flawlessly.

[Rob Leadbeater]

Hi,

It might be that the updated QLogic driver expects the MSA to be in an Active/Active configuration, rather than Active/Standby as you are at the moment (5.2 firmware)...

Hope this helps,

Regards,

Rob

[Rob Leadbeater]

If you check on Spock, you'll see that your MSA should be on at least firmware 7.0 to be supported...

http://www.hp.com/storage/spock

Cheers,
Rob

[toomas_aas]

7.00 is the active/active firmware, right? As far as I know, running Netware in A/A configuration is not supported in heterogenous environment (where other OSes connect to the same MSA1500), has this changed?

According to PDF freshly obtained from SPOCK, 5.20 is still supported A/P firmware. The driver version, however, is listed as 6.90m, so maybe driver 7.00 really is not supported in this configuration.