Backup job failing on 1 VM

[Daveyd]

I have a VM running 2008 x64 with 2 vmdks. 1 vmdk for the O/S that is 40GB and the other that is 1.4TB. I have not been able to sucessfully backup the VM. The initial backup and the 3 retries error out with the following message: Creating snapshot CreateSnapshot failed, vmRef "vm-701", timeout "1800000", snName "VEEAM BACKUP TEMPORARY SNAPSHOT", snDescription "Please do not delete this snapshot. It is being used by Veeam Backup.", memory "False", quiesce "True" Operation timed out.

I have tried to de-select the 1.4TB drive, thinking that may be the issue, but I still receive the above error message.

Any idea?

[Gostev]

Please try to create snapshot of this VM with vSphere Client.

[Daveyd]

When I do I get: File <unspecified filename> is larger than the maximum size
supported by datastore '<unspecified datastore>

The 1.4TB vmdk is on a VMFS datastore with a 8MB block size

[Vitaliy S.]

Dave,

Could you tell me what datastore the VMX file is located on for this VM? What is the block size for this particular datastore? I guess you have 1.4 TB VMDK file located on the separate LUN, right?

If this is the case, then you should have larger VMFS block size for the datastore, where VMX file is located, as all snapshots are automatically placed on this datastore. This is the reason why you see the error message while creating snapshot manually.

[Daveyd]

Dave,

Could you tell me what datastore the VMX file is located on for this VM? What is the block size for this particular datastore? I guess you have 1.4 TB VMDK file located on the separate LUN, right?

If this is the case, then you should have larger VMFS block size for the datastore, where VMX file is located, as all snapshots are automatically placed on this datastore. This is the reason why you see the error message while creating snapshot manually.

The VMX file is located on the datastore with the 40GB O/S vmdk. That datastore is formatted with 1MB block size.

Yes, the 1.4TB vmdk file is on a separate datastore/LUN that is formatted with 8MB blocks

Of note, looking at the datastore with the O/S vmdk, there are 32 snapshot files named <ServerName>.Snapshotxxx.vmsn. I assume that is from Veeam each time it failed?

So the resolution is to reformat the datastore that the vmx file is on to 8MB blocks? If so, can you please explain why that is? Thanks!

[Gostev]

This is because snapshots can potentially grow to the size of original VMDK, so VMFS has to be able to accommodate them.

[Daveyd]

So, I need to move all the VMs off the datastore that this particular vmx is on and reformat it with 8MB block size, correct?

[Vitaliy S.]

Dave, that's correct.

[Daveyd]

So, doing a Veeam backup of just the O/S vmdk will not work either?

[Vitaliy S.]

Actually, VMware snapshot operation cannot be performed on per-disk level, it is only possible to create snapshot of the entire VM with all disks included. However, If you want to exclude VM disk from snapshots, mark the 1.4 TB disk as Independent in the VM hardware settings using vSphere Client and re-run the job, that would allow you to backup just O/S vmdk file. Hope it helps!

[Daveyd]

One more quick question...Are there cons to formatting all datastores with 8MB blocks? Performance issues or anything like that?

Thank you guys for all of your help!

[Vitaliy S.]

As far as I know there is no noticeable I/O disk performance difference by using a larger block size. Although, If you re-format your datastore with 8 MB block size, please be aware that, an 12KB log file will actually take up 8MB of disk space (1 block) and a 9 MB file will take up 16MB of disk space (2 blocks).

[Daveyd]

So my solution is going to be to mark the 1.4TB drive an independent, backup that drive with Backup Exec to our dedupe appliance (Exagrid) and backup the O/S vmdk with Veeam