VM Snapshots not being removed/consolidated by the volume collection jobs.

[supraedd101]

We have a problem with a few servers in a certain volume, this doesn't affect all servers in that volume. The volume collection runs each night to create a snapshot and they are successful. But often we find there will weeks of snapshots left on the VM's, where we have the collection set to only keep 7.

Any thoughts/suggestions would be appreciated.

Edd...

[Nick_Dyer]

Hi Edd,

I've seen this in the past where the VM cleanup operation times out after a vCenter synchronised snapshot; the most common resolution to this is to reduce the amount of VMs you have in the volumes that are being snapshotted.

How many VMs are part of this synchronised snapshot? And is it only occuring within a single volume inside of the snapshot collection?

[supraedd101]

Hi Nick,

We had 16 VMs in the volume when getting the problem. I have started to reduce this now. I'll be keeping a close eye on them over the next few days to see if the cleanup starts working.

Is there a recommended maximum do you know?

Thanks,

Edd...

[rbrowne40 · Accepted solution]

Hi Edd,

There is no maximum per se it really comes down to the amount of IO the VMs themselves are generating. It gets harder to quiesce datastores as IO increases, obviously the more VMs in a single datastore the more IO we are going to see and the greater number the of VMs that need to be sync'd at the same time - hence the recommendation to reduce the number of VMs per datastore if you are having synchronisation issues.

Nick posted a link in another thread to a good blog post discussing this very question: VAAI and the Unlimited VMs per Datastore Urban Myth »  boche.net – VMware vEvangelist.

The blog recommendation was: "10 HIGH IO VMs, 15 AVERAGE IO VMs or 20 LOW IO VMs" This a is a good starting point & can be adjusted with a bit of trial and error.

Hope this helps.

Rols