Snapshots taking long time to commit

[jurajfox]

Anyone have issues with snapshots taking a long time, up to half an hour, to commit. During this time the server is unavailable.

We have a couple servers that don't really have a time when they are idle so snapshots are taken during quite a bit of activity on the disk.

Any suggestions on making snapshots more usable in this scenario?

[cheeko]

What storage are you running on? Do you have storage extends active?

[jurajfox]

CX500... no storage extends.

[cheeko]

Maybe migrated from VMFS2?

[jurajfox]

No, this is a clean VMFS3 install.

[esiebert7625]

It seems the bigger they are the longer they take to commit. Here's another thread talking about it.

http://www.vmware.com/community/thread.jspa?messageID=584324&#584324

[cheeko]

How big is that snapshot actually?

[admin]

Not only the bigger they are the longer they take to commit, but the busier the VM the longer they take to commit.

[jurajfox]

I think my snapshot was quite big...80GB.

[jurajfox]

So I gather there is no way around this issue. If you have a snapshot that is rather larger or if the VM is quite busy during the snapshot you will suffer an outage on the VM when it comes time to commit.

[esiebert7625]

Check this out...

http://www.vmwarez.com/2006/11/beware-long-snapshot.html

[tcassie]

In response to Cheekos comment re VMFS2, will this make a difference?

I have two VM's that were copied across from VMFS2 to VMFS3 (Netware boxes so converter was no good). Converted the VMDK's to VMFS3 and updated the hardware. Everything is fine except comitting snapshots (deleting a snapshot) makes the VM unavailable until the commit has completed.

[osde_info]

here are some vmwvi vmwesx snapshot apply commit delta "delete all" (sic) remove (sic) redo timings from other users / threads !

18G = 2 hours

80G = 5 hours

450G = 8 hours

regards

clive

[NNy2k]

Has anyone seen this take over 24hrs? We have about 150gb of delta's due to bad backups and they caught up to us and we didn't see them so now we're running 'delete all' so we can clean up but it's almost been 24hrs.

This is our production Exchange 2007 server and has lots of IO so I'm sure it's taking long but would 'love' to see something telling me it's still working so I don't feel in Limbo. Couple that with waiting for VM support to call back and I'm itching to do something.

Thanks in advance everyone!