NPIV FAILS??? with HP Virtual Channel Fiber Connect Module

Unanswered Question
Nov 16th, 2007
User Badges:

The setup is HP C-Class 8x BL480 Blades w/QMH2462 dual port hba going out via 2 VC_FC Interconnects. Egress Ports connect one each to 2 gps uplink to two MDS9509 Switches vsan 200 on switch A and vsan 400 on switch B. NPIV is enable on each switch. Both VSANs have single initiator zones configured for each Blade host with EVA5000 Vdisk storage luns selectively presented to Blades. Pair Windows 2003 EE running MSCS Active/Passive Cluster - 3 Luns are Cluster Disk Resources presented to cluster nodes. Luns presented to hosts appear and are available, however when Active or Passive is rebooted when host comes back up shared LUN is no longer showing up in Disk Manager or presented to host. If Cisco MDS9509 F_Port which is connected to HP Blade VC_FC Module is "shut/No Shut" or disabled then re-enable LUNS appear or are again presented to rebooted host.

Has anyone seen this problem?? Cisco Switches are running O.S. 3.2.1a per HP Blade Compatibility list.


Please advise

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
stephen2615 Sun, 11/18/2007 - 13:43
User Badges:
  • Bronze, 100 points or more

If the switch needs a port to be re-enabled to get the storage "visible" again, I get the feeling that the EVA is the issue here. Just for interests sake, does this setup work when you don't use the VC module?


Also, does the same thing happen over and over again? In other words, have you done this more than once and you still can't see the LUN's?


Can you see flogi's when you do what you do? I would debug the interfaces while failing over.


Stephen

CKLSailing217 Wed, 11/21/2007 - 13:33
User Badges:

Stephen,

First, thank you for your reply.

I do not have another storage system to test against other than an EVA5000 VCS3.110 Active/Passive. I have not tested agaisnt pass-throughs or Brocades/Cisco switches.


I have single host connected to this eva via the VC_FC same Q-Logic HBA and when rebooted do not lose access to their luns???


It appears to be a problem with shared luns as stated with MS-Clusters hosts and Poly_Serv connected hosts. I have not been able to setup PAA on the Cisco switch to capture any flogi data.


CKL

briancoffey Mon, 11/26/2007 - 07:48
User Badges:

Having a similar Problem I have a 9140 with 3.12A and it is ISL'd to a 9140 with 2.03 FW on it NPIV is only Enabled on the SW with the 3.12A do you know if you can have two switches in a VSAN one with NPIV enabled and the other without it enabled?

CKLSailing217 Tue, 11/27/2007 - 12:01
User Badges:

I am connecting one uplink per HP VC_FC interconnect to two seperate Core MDS9509 switches both, NPIV enabled, runing Cisco O.S. 3.2.1a.

I have read per Cisco storage documentation you can run switches with mismatch code going 2 versions back.

CKLSailing217 Wed, 12/12/2007 - 09:40
User Badges:


Final post findings: I swapped out the Q-Logic hba's for Emulex hba's and everything is functional.

CKLSailing217 Tue, 01/08/2008 - 20:41
User Badges:

Q-Logic QMH2462 hba's now working fine.

Needed latest and greatest Windows 32bit/64bit Storport driver v9.13.18.

Also, required firmware upgrades to Cisco fw v3.2(1a)with Qlogic HBA bios v1.26 and fw 4.00.90.

Tested all servers no longer losing access to luns on reboot.

Actions

This Discussion

 

 

Trending Topics: Storage Networking