Hi there,
if somebody can help me with our problem, I would appreciate.
We have ESXi 3.5 server now with just one VM. We did only snapshot last week (27/01/09). Today morning the VM stopped responding because of low space in datastore. I have deleted the second VM to make some space, but now I am not able to start it again - with error VMWare ESX Server cannot find the virtual disk "SERVERNAME-000001.vmdk". In the datastore, there is only SERVERNAME-000001-delta.vmdk file.
There are these files in the datastore -
SERVENAME.nvram
SERVENAME.vmdk
SERVENAME.vmsd
SERVENAME.vmx
SERVENAME.vmxf
SERVENAME-000001.vmdk
SERVENAME-flat.vmdk
SERVENAME-Snapshot1.vmsn
vmware.log
vmware-10.log
vmware-5.log
vmware-6.log
vmware-7.log
vmware-8.log
vmware-9.log
Can somebodz help me?
One of the worst things IMHO that can happen to a VM is to fill up the LUN where it resides. Except for performance reasons that is another very good reason not to keep a production server in snapshot mode - since the delta file will grow outside of the allocated VM size.
Try to commit the snapshot through VMware infrastructure client.
You can also try to clone the VM to another LUN using vmkfstools -i
Thanks for reply.
Well, i tried to do a snapshot, but with error - an invalid snapshot configuration was detected. =o(
Or is there a way how to restore data from the snapshot? I think it could be enough for us to start the server at the 27/1 and then recover from the backup that is on the drive.
When you create a snaphot a delta file is created and all changes that occur while you are in snapshot mode will be written to the delta file. The delta file will grow outside of the flat file. That is also the reason you ran out of space on the LUN. When you commit the snapshot the delta file is "merged" with VMNAME-flat.vmdk. This obviously also allows you to jump to any snapshot and "restore" the system state to exactly the way it was when you created the snapshot. Your problem seems to be that the delta file was corrupted once your LUN filled up
Did you try to clone the entire VM to a different LUN (using the command line)? I don't think it will work, since it seems that your VM is corrupted, but it's worth a try. Do you have a image backup and a file level backup of the VM?
Finally I have found the reason of the corrupted snapshot - the snapshot header file (meta file?) was deleted. So I have recreated that file and than the VM was able to start again. We were very lucky.
I did not know how to use the cloning using the command line, so I haven't tried that. But in fact it was not necessary in the end.
Anyway many thanks for your responses.