VM Snapshots issue

anything I don't mind · 20 May 2013 at 15:38

We have a esxi4 host that has ran out of space on the datastore due to unconsolidated snapshots.

At this point we are having to update the host which only has internal storage to esxi 4 update 3 so we can migrate the host to our new shared storage infrastructure. But hopefully we don't run in to any problems with that.

But the question is, is there any magic commands we can do to clear off the snapshots if we don't have any space on the datastore?

Little_Crow · 20 May 2013 at 16:26

Is there any reason you can't use VMWare converter to move your existing VM's to an ESX host on the shared storage now?

Or is this the only ESX server you have?

anything I don't mind · 20 May 2013 at 16:35

That was one of the first things i tried this morning. It got to 10% and then failed with a network error.

We are going to migrate it to shared storage this evening once we have upgraded the source host to a compatible version. At the moment when we migrate it says that the current version does not support migrating vms with snapshots or delta disks. please update to esx 4 update 3.

But it looks like there are no other options but this. This site was poorly set up and they even put esxi on one hp server and then put two exchange servers on the esxi with local storage. completely crazy solution, but what happened was three years ago previous admin made a snapshot and it created duplicate set of disks. I thought the space would last until we upgrade to 2010 in the coming months, but looks like it was thin provisioned. But as no free space can't consolidate snapshots.

anything I don't mind · 20 May 2013 at 17:09

ok we have a new plan now, we going to add a hba in to the problem host and then connect it to the shared storage, move one disk to the shared storage and try to start up the guest. Then we going to try consolidate all snapshots.

Little_Crow · 21 May 2013 at 12:36

If you wanted to pursue the VM conversion you'd have to dig into the log files for the exact problem.

However, one thing to check is that network link speed and duplex is negotiated correctly. This will always kill a conversion due to the amount of IP errors generated.

anything I don't mind · 21 May 2013 at 17:06

In the end had to recover exchange from backup due to time constraints. But we ran a clone to a datastore on the shared storage and a tape backup at the same time. The clone failed on a specific file and the tape backup worked and we got exchange back up. We first tired hba card but for some reason didnt work in a hp dl380 g6, but it came out of a new gen8 server. I think if given enough time i probably would have copied the vmdk to another data store with more space and then tried to consolidate snapshots. Funny thing is that datastore never had enough space left to deal with the snapshots. Someone made a snapshot in 2010 called test and never deleted it. Lesson of the day is to delete snapshots older than 3 hours, especially when using thin provisioning.