einstein-a-go-g
Hot Shot
Hot Shot

Snapshotted LUNs

When I present a Snapshotted LUN (copy of original LUN) back to the ESX 3.0.1 servers, something very strange happens, when I re-scan for VMFS, the ESX server seems to get confused and swaps the "original" LUN for the copy LUN, and it gets in a real mess.

e.g. have a LUN, vmfs datastore named vi3test with a single Guest VM which happens to be Win2k3 Server. Snapshot the LUN using SAN Tools, this creates a Read Only Copy of the original LUN. Present this back to the original esx server which already has a mapped datastore name of vi3test, after re-scanning for VMFS, the new LUN is mapped as vi3test, and the original LUN becomes vi3test(1)!

Anyone seen this issue?

Do I need to present this LUN to a "special ESX DR server" to recover Guest VMs from this LUN.

0 Kudos
5 Replies
nadger
Enthusiast
Enthusiast

It should work depends on the SAN hardware. This works fine for us on the NetApp products.

Have you tried enabling volume resignaturing under LVM in advanced settings for the host?

0 Kudos
einstein-a-go-g
Hot Shot
Hot Shot

It's a NetApp we are using, yes resignaturing under LVM is enabled. It's just very odd behaviour, and since I've seen the latest NetApps docs ref ESX 3.0, I suspect it's an issue we our experiencing. I don't wont to have to raise another SR, we currently have four oustanding!

0 Kudos
wowen
Contributor
Contributor

I experienced a similar situation when migrating between EMC SANs. We mirrored our LUNs to the new SAN and connected to the ESX server. The ESX server saw them as new disks and we were forced to go in and add all 100+ servers back to the ESX servers.

Can anyone tell me what ESX sees to think the drives are new drives? Is resignaturing supposed to fix this? The documentation I have found on resignaturing is to present a SNAP or a Clone back to the same ESX server is it not?

0 Kudos
Hernandes
Contributor
Contributor

Hello,

Don't if this can helps you today. Just seen the post.

We've experienced quite the same thing testing our BCP.. LUN were presented to the other ESX but without being able to see the volumes..

We tried several things and finally found this solution :

Advanced settings : LVM.DisallowSnapshotLUN=0 and LVM.EnableResignaturing=0 (not to have a renamed volume, with the inconvenience we have ..).

It appears now that if any VMs was running on the volume that "crashed", you have to force the poweroff of this VMs from VC. (or any other way).

At that point we were able to see the volumes.

Hope that will help anybody, tell me if you need more infos.

Kind Regards

0 Kudos
Hernandes
Contributor
Contributor

Hello,

Don't if this can helps you today. Just seen the post.

We've experienced quite the same thing testing our BCP.. LUN were presented to the other ESX but without being able to see the volumes..

We tried several things and finally found this solution :

Advanced settings : LVM.DisallowSnapshotLUN=0 and LVM.EnableResignaturing=0 (not to have a renamed volume, with the inconvenience we have ..).

It appears now that if any VMs was running on the volume that "crashed", you have to force the poweroff of this VMs from VC. (or any other way).

At that point we were able to see the volumes.

Hope that will help anybody, tell me if you need more infos.

Kind Regards

0 Kudos