How to force Windows Automount to resignature a VMFS volume on SAN?
Hi,
we are in the process of deploying a new backup solution (Veeam Backup & recovery) in our vSphere 5 environment and want to use Direct SAN backup within our FC SAN. I've read a lot of posts that say when Windows Automount is enabled it could
cause Windows to resignature a presented VMFS volume and thus destroy your VMFS partition (though there are ways to recover from this). Before going live I'd like to test such a scenario - enable automount and let Windows issue a signature to a VMFS volume
presented to the VBR server. Then see how to recover from it.
However, I've not been able to produce such a situation. Our VBR server is 2008 R2 SP1 and with automount enabled I've tested the following scenarios:
1.
- created a new LUN on SAN, presented it to an ESXi host and created a VMFS datastore
- presented the LUN (while sill being present to ESXi) to the VBR server
- nothing happended! Windows disk manager just shows the partition as online, healthy and labels it as Primary Partiton.
2.
- created a new LUN on SAN, presented it to an ESXi host and created a VMFS datastore
- unmounted the datastore on ESXi and unpresented the LUN from ESXi
- presented the LUN (while now not being present to ESXi) to the VBR server
- nothing happened - just as above!
So, in what scenario does Windows resignature a disk that is presented to it? Anyone knows? Or is this just a myth?
Thanks
Michael
December 25th, 2011 10:53am
Mybe you would have a better chance on the Storage forums:
http://social.technet.microsoft.com/Forums/en-us/winserverfiles/threadsMCTS - Please remember to click Mark as Answer on the post that helps you, and to click Unmark as Answer if a marked post does not actually answer your question. This can be beneficial to other community members reading the thread.
http://mariusene.wordpress.com/
Free Windows Admin Tool Kit Click here and download it now
February 4th, 2012 12:36pm
When a LUN is presented to ESX host, it get’s a signature. This signature is in the form of UUID. When you take a snapshot of LUN, the exact UUID is copied in the snapshot. So when you wish to restore
the snapshot in case of DR, in ESX4.0 you are given a option to re-signature the disk i.e. to assign a new signature to the disk or assign the existing signature
Remember you cannot mount datastore whose signature collides with existing signature. You either have to give new signature to datastore or unmount the exisiting datastore. There are pros and cons
So what happens if you decide to re-signature the copy of datastore
New signature is assigned to datastore
It looks distinct than the existing datastore as it assigns new name it as snap-<snapID>-<oldlabel>
But most important concern, once you re-signature, you will have to register vmx file, vmsd and vmsn file
In general in production scenario, snapshot is taken for every lun and placed in DR site. So when production fails, this lun is mounted in the production site without resignaturing and all your vm’s are up in a 5-10 minutes time
For Storage queries, post here.
http://social.technet.microsoft.com/Forums/en-us/winserverfiles/threads/
Thanks
February 4th, 2012 1:51pm