MSA1000 hangs on server boot

[Anthony.Crawford]

I am trying to set up a Windows 2003 cluster on 2 DL585 servers, each with dual QLogic QLA2340 HBA's, connecting to a new MSA1000 with dual controllers and dual MSA 2/3 hubs.

I can setup the cluster OK, and move disks from one node to the other. The problem occurs when I cold boot or reboot one of the nodes. When I boot either server, the MSA1000 becomes unresponsive, the heartbeat LED on the MSA1000 stops flashing, there is no response to the buttons on the front of the MSA1000, and there is no response to the CLI. The MSA1000 stays "hung" until part way through the Windows startup, when I assume the HBA windows drivers are loaded, and the MSA1000 responds again. OR, if I press control-Q during startup to go into the Qlogic FastUtil utility, the MSA1000 "unfreezes". While the MSA1000 is hung, the MSA Hub ports for the booting server are flashing rapidly between the 1G and 2G LEDs.

If one cluster node is up and the other is rebooted, then the running node loses sight of the MSA1000 and its disks, and the cluster fails. Also at this time, SANSurfer reports both HBA's as "loop down", until the other node is partway through booting, when the HBA's status changes to "loop up, LIP detected".

Both MSA1000 controllers are running firmware V4.48.B342. PORT_1_TOPOLOGY on both controllers is set to "Arbitrated Loop". All HBA's have driver version "SCSI miniport 9.0.0.13", BIOS V1.45, firmware V3.02.28, and the Connection Options is set to "2 - Loop preferred, Otherwise point to point". I am using MPIO for multi-pathing.

Has anyone seen anything similar? Any advice or suggestions would be most appreciated.

Thanks.

[DanDelcroix]

Do you encounter the same problem in a single path config?
Also, be sure to use the basic MPIO:
http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareIndex.jsp?lang=en&cc=us&prodNameId=421495&prodTypeId=18964&prodSeriesId=421492&swLang=8&taskId=135&swEnvOID=1005
The MSA1000 does not support full featured MPIO.

[Anthony.Crawford]

I did try using a single path, by removing 1 hub. Then I powered on 1 server, and the MSA1000 controller in front of the remaining hub hung (heartbeat LED stopped, no response to front panel buttons), until part way through the Windows startup. The other MSA1000 controller with no hub remained responsive.

I am using V1.1 of MPIO referenced by the link in your posting.

[DanDelcroix]

The HP FCA2210 HBA is the OEM version of the Qlogic QLA2340 HBA you're using. They are not supported under Windows for the MSA1000.
Have a look at the compatibility guide:
http://h20000.www2.hp.com/bc/docs/support/SupportManual/c00222327/c00222327.pdf
or at the Qlogic site:
http://www.qlogic.com/interopguide/vendor.asp?vendor_id=252

Do you have other HBA's compatible with the MSA1000 to test with ?

[BR894765]

First of all, use HP-supplied firmware for HBA (HP name FCA2214DC)
http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=us&prodTypeId=12169&prodSeriesId=315890&prodNameId=315892&swEnvOID=1005&swLang=8&mode=2&taskId=135&swItem=co-26946-3
Try to use latest HP driver (9.1.0.13)
http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=us&prodTypeId=12169&prodSeriesId=315890&prodNameId=315892&swEnvOID=1005&swLang=8&mode=2&taskId=135&swItem=co-37098-1

[Uwe Zessin]

The FCA2210 is the same hardware as the FCA2214 - it was just a package with documentation for NetWare instead of Linux/Windows.

[Anthony.Crawford]

The HBA's being used were actually HP part number 281541-B21 which I believe are FCA2214 HBA's (apologies for describing them as QLogic QLA2340, which is what SANSurfer called them).

These HBA's have been replaced with FCA2404 HBA's and appropriate drivers, and the issue is now resolved.

Thanks.

[Michael Ratta]

I encountered a very similar issue with the MSA1000 2/3 fibre hubs and FCA2214 (Qlogic). I escalated it up to HP Stoage Support and they finally admitted (after months of troubleshooting) that there is an incompatibility with the MSA 2/3 hubs and the FCA2214. The old FCA2101 (Emulex) work fine with hubs and the FCA2214 will work with switches but not hubs. They are 'working' on a fix but I have yet to see an advisory about the issue.

[Gregory Nazar]

Greetings all..

Regarding the issue concerncing the MSA 2/3 hub in a dual node clustered environment, HP has released a support document which can be
viewed here:

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&objectID=c00625693&jumpid=reg_R1002_USEN

I believe this document is specifically adressing the problem that is being discussed in this thread. Enjoy!