StoreEasy Storage

NAS9000 and CPQKGPSA.sys PCI Problem

 
SOLVED
Go to solution
BR794491
Advisor

NAS9000 and CPQKGPSA.sys PCI Problem

Hi
I am pretty new to the world of NAS9000s.
We have a cluster of 2 NAS9000s connecting to a VA7100. This week one of the nodes failed and is unable to mount the quorum disk and the data disk. The other node continues no problem.
Rebooting, disabling and then re-enabling the switch ports does nothing.
The server has crashed a couple of times with errors related to CPQKGPSA FCA driver.
I have know found a related article at HP

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=PSD_OI040715_CW01


Has any one successfully applied this patch or have more info on it.

Are there more tests that I can do? Secure Path Manager shows the LUNs and controllers as OK. Disk Manager even reports the correct LUN numbers, I just can't see them from one of the cluster nodes

Any help/info gratefully received

Regards
Jonathan
2 REPLIES 2
CA1118839
Valued Contributor
Solution

Re: NAS9000 and CPQKGPSA.sys PCI Problem

Jonathan,

First, have a current backup before we go too far...

This could be as easy as:
1. The disks are active on the working node and therefore not usable on the second node
a. Move the Cluster group over to the non-working node
2. It is not entirely evident that you are seeing the LUNs correctly from both nodes
a. Verify that zoning is not enabled or is setup correctly
b. Verify that secure manager is not enabled or is setup correctly
3. The signature on the disks is slightly corrupt
a. Evict the node
b. Run a cleanup - /forcecleanup is OK other than for cluster-aware NFS. There is a release note on how to modify the registry though, so you should be OK doing this.
c. Re-add the node

Also, it is not clear if you are having a HBA problem. Can you cut and paste the event log message?
BR794491
Advisor

Re: NAS9000 and CPQKGPSA.sys PCI Problem

thanks for your advice
all zoning was ok.
following the message I couldn't connect to either the data or quorum LUNs from either node.
I have applied service release 2 and this seems to have improved things considerably.
many thanks Jonathan