Commit 6 month old Snapshot

[aderrNIC]

Background Info:

In an effort to free up some space on our datastores I am going to clean up some of our VMs and more specificially the snapshots that some of them are running on.

There are two servers that have been running on snapshots that were created in July of 2011, one is a production Windows 2003 SQL server, the other an IIS 2003 production server (both on different datastores).

The physical blades are running ESXi 4, using vSphere 4 Update 3 to mange them.

The datastore with the SQL server has ~70GB of free space available.

From what I can see the snapshot (000001.vmdk) has grown to about 34.5GB and the original disk is 55GB Thick provisiong.

There is only one snapshot of the server.

Concern:

I have read that vSphere 4U3, when using Delete All Snapshots, will just merge the two files together and then delete the snapshot file. However, how am I assured that there will be enough space on the datastore for it to combine the two files. In my calculations, worst case senario would be that the file merged file would be 78.5GB roughly. That is to say, that every thing that is in the snapshot will be merged into the original file.

Question:

Based on the above information, would this cause the deleting of snapshot file to cause the datastore hit capacity and as such, the merging of the two files to fail?

Also, based on previous expirences, can anyone estimate how long it may take to merge and delete the snapshot, assuming that everythig does workout?

Thank you in advance very much!!

- Alex

[weinstein5]

Welcome to the Community - Concerns are well founded and yes if there is not enough space the consolidation will fail. From what yo describe sounds like you might have enough space but it will be close -  A new file is not created but the snapshot is merged into the -flat.vmdk so at worse you will end up with 2x the swap file - that is my understanding if I am wrong I am sure someone will chime in. Also to free up space if you have other VMs on this datastore that do not need to be running - shut them down and this will free the space used for the per VM vmkernel swap space -

[aderrNIC]

Thanks for the good info. I didn't think to shut down the other VMs. There are a couple running on that datastore.

Would you think that shuting down the Guest OS of the VM that I am going to merge the snapshots of would help as well.

     In addition to freeing up space, I would assume that it may make the process run smoother, or is that just a nice assumption I am making?

Thanks again!

[a_p_ · Accepted solution]

Let me put this together:

• the ESXi host is running version 4.0 Update 3
• the VM has a thick provisioned virtual disk (55GB)
• the current size of the snapshot file is 34.5GB
• free disk space on the datastore is 75GB

If all of the above is true, then there should be no issues with running a "Delete All" from the snapshot manager. Since the virtual disk is thick provisioned the data blocks from the snapshot will only replace the according blocks in the base disk and the base disk will not grow in size. The only additional disk space you temporarily need is for the "Consolidate Helper" snapshot which is created if you run the "Delete All" with the VM powered on. However, with 75GB of free disk space this is no problem.

Due to the snapshot size the consolidation/merging may run a couple of hours - depending on your storage system - (please don't become inpatient) and depending on the host's performance the VM's respond slower. What you absolutely need to make sure is that no other snapshot is created while the deletion process is running (e.g. through a backup application).

André

[aderrNIC]

Great! I had a suspicion that the thick formating of the drive would be useful here. Unforunetly, that is not the standard practice for the data center, the majority of our VMs are set up as thin provisioning.

But for now, that puts my mind at ease some.

Thanks for all the help!