Weird Issue: Datastore not appearing in 1 ESX Host of the Cluster

[chaicka]

Background Info:

Cluster Size: 3 ESX Hosts (each with 2 HBAs)

LUN: 3 hosts map to same LUN, eg. vmhba0:0:5 and vmhba0:1:5

Datastore: XP1-00-0b-OS

On ESX hosts 1 & 3, the datastore is shown under storage configuration. However, on ESX host 2, the same datastore just refuse to show up under storage configuration.

Likely, if the datastore is created on host 2, it will not appear in host 1 & 3.

We have tried remove and re-map the LUN to all the hosts, delete and recreate the datastore, etc. None seems to work. However, the LUN allocation is confirmed configured correctly, just like other LUNs on these 3 ESX Hosts.

Any other actions that we can try?

[Gerrit_Lehr]

Are there other LUNs vissible to all Servers (to make sure the Zoning is correct)? Lun Masking checkes, twice? Some servers just seem to need a while to recognize new LUNs, did you scan for new LUNs and VMFS and Refresh the Storage Tab?

Kind Regards,

Gerrit Lehr

If you found this or other information useful, please consider awarding points for "Correct" or "Helpful".

[mike_laspina]

Hello,

Any messages in /var/log/vmkwarning & vmkernel?

What does esxcfg-vmhbadevs show?

[chaicka]

Hi Sinac,

There are 5 other LUNs for the 3 ESX Hosts, all working fine except this particular 6th LUN.

The mentioned steps - LUN configuration, storage adapters scanning, storage refresh, etc have all been done prior to posting in here.

Anyway, we finally resolved it after trying all possible actions or combination of actions. It's weird but it's resolved.

Resolution (very specific steps):

1. Delete Datastore.

2. Refresh on all 3 ESX Hosts | Storage tab.

3. Unmap/Unallocate the particular LUN from all 3 ESX Hosts.

4. Scan on all 3 ESX Hosts | Storage Adapter tab.

5. Remap/Re-allocate the particular LUN to all 3 ESX Hosts.

6. Scan on all 3 ESX Hosts | Storage Adapter tab.

7. Recreate Datastore.

8. Refresh on all 3 ESX Hosts | Storage tab.

A similar behavior has occurred 1 other occasion, for new clusters. For the other occasion, a simple remap/reallocate on that particular affected ESX host would resolve the issue. But this time round, that didn't work.

[chaicka]

Hi,

Another occurrence of datastores 'disappear' after reboot of ESX hosts, and there is LVM-related error message on the console screen of the ESX hosts.

The error "ALERT: LVM: 4475: vml.020005000060060e80042c530000002c5300000b294f50454e2d56:1 may be snapshot: disablingaccess. See resignaturing section in SAN config guide"

[mike_laspina]

ESX issues this message when it finds a UUID on a vmfs volume that was presented to the host in the past on a different LUN than is now being presented.