HBA Cards only see some of the LUNS

[Dharmesh Kotak]

Hi all,

I have the following configuration and a curious problem.

Config is as follows:-

1) Two physical servers with Windows 2003 Enterpirse Edition 2 64-Bit. HP DL380's with 32Gb RAM.
2) Both servers each have two HPAE312A (QLE2462 - qlogic) dual port 4GB Fibre Channel HBA's
3) Two HP (Brocade 200E) Fibre Switches
4) HP MSA 2012FC SAN
5) Two LUNS on the SAN. One for a Quorum, and the other for data
6) The ACL's on the SAN are setup correctly AND point to the required WWN of the HBA's
7) The zoning is configured correctly picking up the required WWN's of the SAN and HBA's.
8) MPIO software on both servers and all Windows Patches and hotfixes applied and up to speed.
9) Servers have Windows 2003 Clustering in operation, in preperation for Clustering of SQL 2005.

The problem I have can be seen in the attached Daigram. But, I will also explain.

As can be seen in the attached diagram, all HBA's on both servers can see LUN0 (data volume). However, only one HBA on each server can see both LUN's 0 and 1(Quorum LUN1 and Data), therefore when it comes to failing the servers to test clustering, the test is only ever 50% correct. If I remove the fibre attached to the HBA that can see only LUN0, the cluster does not failover. If I remove the Fibre attached to the HBA that can see both LUN's, the cluster fails over, as the quorum is no longer avilable. This is correct as the other HBA is unable to SEE LUN1. The failover should only occur when both fibres have been detached.

I have checked the configuration on all devices concerned over and over again, but, cannot see why this happening.

Any help would be much appreciated.

[IBaltay]

Hi,
It seems like not all MSA LUNs are presented to both the HBA port wwns. Check the MSA SSP (Selective storage presentation) and if missing, add the remaining assignments.

[Dharmesh Kotak]

I have checked the assignments on the MSA and they are all correct.

[Bill Costigan]

It looks like the LUNs are only available through one of the controllers.

Note: the connections with both LUNs are labled disk and the connections with one LUN is labled SAF-TE. I'm guessing this is an active/passive array and the controller labled SAF-TE is the standby controller.

Does that sound right?

I also found a link that says that the
SAF-TE LUN is for management of the enclosure.

Check out this link:

http://www.safte.org/


In either case, I don't think the two LUN 0s are the same disk. You are not seeing either LUN0 or LUN1 through the second HBA,

[Dharmesh Kotak]

Hi Bill, thanks for your reply.

Your are quite correct in your thinking.

However, I did notice that that the SFP that the MSA (second hba card) attaches to on the Fibre Switch was slightly different to all the others, but, still 4Gb.

I swapped over the SFP's in that switch, so that MSA was using the same one as the others, thinking that this may be the problem. But, I still have the same issue.

Any more suggestions are most welcome.

Thanks in advance.

[Kritao]

Hi,

First thing, please once again check the SSP using ACU and verify both WWN of HBA are selected. Then using SAN Surfer utility check the Target Persistence of the Connection and enable it.

Regards,

Cherish