Here is my situation. I have 4 VMWare hosts set up to view two HP MSA1000's connected in series. The other day one of the fiber switches failed on the array, so I routed all the fiber connections to the secondary switch (no redundant connections). I shutdown the arrays, rebooted them, and the bottom one displayed an NVRAM error. Great... So I rebooted it again, both arrays came up fine this time. I booted up my ESX hosts, and they could see the storage in the top MSA1000, but not the bottom MSA1000. Both arrays were fully allocated and presented to the ESX servers before. The fiber switch is in the bottom MSA. Now if I go to "Add Storage" all the LUNs from the bottom array show up there as available. The problem is they are chock full of data and should be listed with all the other existing resources from the top array, but they are not. Attempts to add one back (which I know will destroy the file system) results in this error:
Error during the configuration of the host: failed to update disk partition information.
Now the old LUNs are listed as "blank" in the add storage wizard and empty but they definitely aren't, nor will it allow me to add them back and format them? Has anyone ever seen something like this before? Can anyone lend any advice? Are all the VMs on the bottom array truly gone?
Thanks in advance.
You may want to check out this post http://www.yellow-bricks.com/2008/12/11/enableresignature-andor-disallowsnapshotlun/
Short of it is you may want to set enableresignature to one on your ESX hosts and resacn.
That didn't do the trick sorry...
Have you open a report with VMW on this?
Can you post the vmkernel log file of the systerm after it has boot but has not mounted the VMFS volumes from the LUNs on the boot MSA?
New wrinkle... "Missing" datastores now show up under the "datastores" view. They report the correct size, and number of VMs, however you cannot browse the contents of the ones that were "missing." Essentially they are showing up as there, but not there. They also do not appear under "Storage" in configuration... In light of this new development, I tried the trick again from earlier in the thread, and that failed to get them back either...
Engage HP. Engage Vmware.
Check config of the MSAs with "show connections" on the terminal session on them, correct it when it shows anything other than Linux as profile. If this fixed search in the community for "saxnot" you will find the needed steps there.