cluster validation problem at "List All Disks"

Thu, 01 May 2008 15:41:40 Z

I'm attempting to build out a two-node Windows Server 2008 Failover Cluster between an HP Superdome partition and an HP Integrity rx8640 (Itanium-based servers), but I can't get passed the "cluster validation checks". I have shared HP XP24000 storage between the nodes, and both nodes see these disks as expected. The issue is that the validation routine .. whether run from the 'Dome or the rx8640 .. gets hung at "List All Disks" for the rx8640 only. It claims that it is having a problem "preparing" for storage validation on that node. It smells like a some sort of node access problem, 'cause I can successfully run the validation between the 'Dome partition and another partition in my environment. (Note that this other partition doesn't have the shared XP storage attached, but the test finishes successfully . . as "List All Disks" polls both nodes for all disks, local or shared . . .

Any thoughts on why the rx8640 cannot be accessed correctly for the disk polling? Again, I've tried all kinds of validation permutations between the two 'Dome partitions and the rx8640, and the rx8640 is the only one with access issues. My gut feel is that it has nothing to do with the XP storage, 'cause the problem occurs with a routine that is supposed to just poll all candidate servers for all disks. Maybe a permission problem somewhere?

Thanks for any help in advance!
Joey

cluster validation problem at "List All Disks"

Thu, 01 May 2008 16:30:13 Z

Resolved

Well, the cluster validation process, as well as the cluster formation process, doesn't like software-based mirroring on your system drive. I had software-mirrored the system drive on rx8640, not remotely thinking that Failover Clustering would care what I'm doing to local drives. Wrong. For whatever reason, the disk polling routine can't deal with software-mirrored system drives. Interestingly, I have another software mirror for a large pagefile on this system, and it seems to be fine with that. . . .

I hadn't noticed this incompatibility listed in any Windows Server 2008 Failover Cluster documentation, and it really doesn't make sense . . Maybe it's a bug??

cluster validation problem at "List All Disks"

Thu, 01 May 2008 17:12:47 Z

Hi Joey,

Can you please provide more information on your configuration so we can setup a repro to investigate? You say that you were using software mirroring on the system drives... were you using in-box Dynamic Disks? Or were you using some 3rd party product (and if so, which one). Did you purchase the add-on Symantec Storage Foundations for Windows and use their Dynamic Disks?

When you broke the mirror, did Validate pass? (with no other changes?)

Thanks!
Elden

cluster validation problem at "List All Disks"

Thu, 01 May 2008 20:12:39 Z

Hi Elden . . .

I used the process to mirror the system drive as detailed in "KB Article ID : 814070". And, yep, when I broke the system drive mirror, the validation passed with no problems. Again, I didn't have to break my "pagefile" volume mirror that I created among two other local drives . . .

Thanks for your help,
Joey

cluster validation problem at "List All Disks"

Thu, 03 Jul 2008 14:35:46 Z

I will confirm the EXACT same problem. Breaking the mirror on the system disks causes the validation to finish successfully.

Config

Windows Server 2008
2 x Dell PowerEdge 2950 with a pair of 73gb 15K in a RAID1 config using Disk Management
2 x Emulex LPE-1150E HBA / server
EMC CX300 SAN

cluster validation problem at "List All Disks"

Fri, 04 Jul 2008 17:15:09 Z

Hi,

Thank you for the information, we've been able to reproduce internally and have confirmed it's a bug. We plan to address this in the next release. If this is a blocking issue for you and you need a hotfix for Win2008, please open a case with Microsoft Product Support Services.

Thanks!
Elden

cluster validation problem at "List All Disks"

Fri, 06 Mar 2009 17:06:59 Z

Elden,

Unfortunately, product support services refuses to help me with a hotfix without a KB article, which I cannot find. Can you please reference a KB article so that I can get a hotfix for this problem?

Thanks,
David Pfeffer