Rebooting FI causes errors on SAN

[r_armstrong]

I have 2 fabrics consisting of a 6120XP in FC switching mode connected to a MDS 9124 with a 4 port port channel.  There are also HP Proliant servers connected directly to the each fabric on the 9124s. Storage is a EMC Clariion NS480. All servers are running ESX 4.1u1.  I want to share luns between the HP servers and the UCS blade servers in order to vmotion VMs over to the UCS servers.

When I upgraded the firmware on one of the 6120s and rebooted, the HP ESX servers temporarily lost access to the shared luns. This also happened when I added the LUNS to the storage group for one of the UCS ESX servers.

Is there a configuration issue that can cause this or am I wrong to expect that I can share LUNS between the HP and UCS servers

[richard.pugh]

What was the purpose of changing to FC switching mode if you are using 9124's as uplinks?

[r_armstrong]

It is my understanding that to enable NPIV on the 9124 requires a reload. Is that not true or necessary?

We are small and not concerned about running out of domain IDs.

So switching mode will cause problems?

[Robert Burns · Accepted solution]

Enabling NPIV does NOT require a reload.  Enabling NPV does.

Regards,

Robert

[r_armstrong]

Since we are not yet in production on the UCS, I guese I can change modes.  I am still wondering why the current configuration causes problems.

[Robert Burns]

Have you checked the MDS logs?

Assuming your zoning is setup correctly and each host is segemented from others, the rebooting of any host or NPV device should not affect anything else in your SAN fabric.

Q.  How are you zoning your UCS and non-UCS initiators? (wwpn, port etc)

Run a show tech on your MDS(s) and check the logs around the time of the interuption/upgrade and see if there are any clues.

Regards,

Robert

[HAROLD MEIER]

Since these were VMware hosts sharing the luns they would have been zoned together. I've seen this behavior when a HBA sends a global TPRLO request. I've never seen that in a UCS CNA though.