Hi,
I've a ESXi 4.1 host with a guest OS (CestOS). The total space of the VMFS(data store) partition is 550 GB. Out of 550 only 25 GB is available now. There are 4 snap shot of that guest OS (the last one is 250 GB and the first one 45 GB). I want to delete the 1st snapshot.
1. Which is the best way to delete this snapshot?
2. If I want to delete the snapshot using vSphere Client, what are the negetive impact.
The Guest is a hosting server with 200 client so I can't bare the down time. Please give me a best way to do this. If possible please provide links to perform this steps.
Thanks in advance.
Basically I agree with continuum and cloning would be safest way to clean this up. However if I understand the situation correctly you should be able to delete the snapshots with the VM without downtime. What I understood is:
can you confirm this?
In this case you can delete the first snapshot (the one closest to the base disk) without the need of additional disk space, except for the changes that go to the newest snapshot. With 25GB free disk space this should be ok!? After the first snapshot has been merged to the base disk, you should have additional 80GB free disk space (according to the time stamps, snapshot file 000002.vmdk is the oldest snapshot, the file names do not necessarily be in an ascending order), in which case you can now decide to continue cleaning up the snapshots one-by-one or using "Delete All" to get rid of all snapshots.
André
Hi,
Take a look at (v.good tips for snapshot removal):
http://communities.vmware.com/thread/283021
and more of an overview of snapshots
http://kb.vmware.com/kb/1015180
Many tx
post a filelist of the vmdks of this VM
Is the basedisk of that VM thick or thin provisioned ? - that is the most important question at the moment
Hi Continuum,
Thanks for your reply. I am attaching the screen shoot of the following command from the ESXi 4 host.
# df -h
# ls -l /path-to/datastore-dir/
# ls -l /path-to/datastore-dir/my-vm-dir/
Here I also attaching the screenshot of disk properties and snapshot manager
Thank you once again for your reply.
Message was edited by: rhmahfuz
Hi
You should think about the way you use snapshots.
If you were using version 4.0 you would need to add more storage - to get rid of them.
As you have 4.1 it should be safe to delete the first snapshot - the one that directly points to the basedisk.
Deleting any other snapshot would result in a desaster.
If possible do this in a downtime - if that is not possible use the nightime when the server is less busy.
Hi Continuum,
Thanks for the Reply. Please make me clear delete "The first Snapshot"
If this is the scenario. The first snapshot mean the SnapShot 01 that is red marked in the image?
Yes - snapshot 01 - the one that directly links to the basedisk (the vmdk which uses a flat.vmdk) is the only one that can be deleted without filling up the volume.
And that is only because the basedisk is thick provisioned.
If you had a thin provisioned basedisk delting snapshots may not be possible at all without adding extra space.
Delete means using the Snapshotmanager button - do not use rm or Datastorebrowser to delete the name-000001.vmdk manually !!!
Hi,
You are really helpful
Can you please tell me that what could be the negetive impact/risk while I delete snapshot using VMware vSphere Client (snapshot size is 43 GB).
Thanks.
If you fill up the datastore while deleting a snapshot you will get into trouble.
If I were in your shoes I would add another datastore with at least 270 GB and then I would use vmkfstools to create a full clone - I would do this in a scheduled downtime.
Basically I agree with continuum and cloning would be safest way to clean this up. However if I understand the situation correctly you should be able to delete the snapshots with the VM without downtime. What I understood is:
can you confirm this?
In this case you can delete the first snapshot (the one closest to the base disk) without the need of additional disk space, except for the changes that go to the newest snapshot. With 25GB free disk space this should be ok!? After the first snapshot has been merged to the base disk, you should have additional 80GB free disk space (according to the time stamps, snapshot file 000002.vmdk is the oldest snapshot, the file names do not necessarily be in an ascending order), in which case you can now decide to continue cleaning up the snapshots one-by-one or using "Delete All" to get rid of all snapshots.
André
Hi Continuum,
Thanks for your Suggestion and guidance.
Hi André,
Thanks for your answer. The current status of the host is:
Should I take a backup of *000001.vmdk & *000002.vmdk before perform the delete process? Or what should I backup before start delete process?
Regards,
rh.mahfuz
Basically you should - independent of this issue - have an up-to-date backup of the VM/important data. Anyway, I do not expect issues with deleting the snapshots one-by-one with the VM powered off.
André
Hi André,
Thanks. Can you please tell me the approx time to delete the 1st snapshot (42 GB) when vm is powered off.
rhmahfuz
This depends on the speed of your storage/RAID system. Assuming a (conservative) speed of ~30MB/s, 1GB would require ~30 sec. In order to delete the first snapshot (which IMO is the one with 80 GB) this would be result in something between 30-45 min.
André
Thanks to Continuum & André. Specially greatful to Continuum to find out the correct Answer.