asked on

How to grow virtual disks in ESX 3.5.0 with VMs that have multiple snapshots?

Hi,

I have a number of VMs running under ESX 3.5.0 that have multiple snapshots each, and their virtual disks are constantly running out of space.

Is there any way to expand the virtual disks even though I have multiple snapshots? It's not an option to just pick one of the states and grow the virtual disk for that. I *must* be able to keep the snapshots!

Also, I have noticed an option in VMware Infrastructure Client that doesn't seem to work that *may* solve my problem - anyone know why it doesn't work? This is what I'm doing:
- Select a VM
- Click on Edit Settings under Commands
- On the Hardware tab, select Hard Disk [x]. Observe the text box next to New Size: and that it is possible to enlarge this size. Add a few GB and select OK.
- I get an error like the following:

----------------------------
Error
---------------------------
A general system error occurred: Internal error
---------------------------
OK
---------------------------

At other times it will look like the disk has been resized, but going back into the Edit Settings dialogue simply shows the original size again.

Any ideas?

Irwin W.

Fitst things first, snapshots are NOT to be kept in perpetuity. They are meant as a troubleshooting tool during patching and upgrading of your OS until it is stable, which is usually after a week or so. At that point you would either commit or revert to snapshot. Also snapshot will greatly affect your performance i/o as well.

Also if you are getting this error when the guest OS is running, try shutting down the guest and then increasing the size of your disk from the VI client>edit settings of the guest>hard disk

Picture-168.png

markzz

What he said.
nappy_d:
Fitst things first, snapshots are NOT to be kept in perpetuity.

You are tempting fate. Don't keep snapshots for long periods. Commit or roll back.
The snapshot function is not intended as a method of perpetually retaining the state of a guest.
If you don't understand why they are running out of space then take the advice given. Commit or remove.
Snapshot is not a replacement for backups.
If things really go wayward with your snapshots you may corrupt the guest disks. (don't ask how I know this)

markzz

OH once you have cleared all you snapshots I expect you'll have no trouble expanding your disks. If you can't commit your snapshots as the inevidable has satarted to occur shut down you guest and use the importer to clone and resize the disk..

DChaffey

Hi,
Firstly - never, ever, ever attempt to expand a disk that has a snapshot. If you manage to partially succeed you'll either lose the snapshot, the disk or the VM where the only recovery is to hope you haven't written new data to the original disk and reconstruct the CID chain.

If you absolutely must keep the snapshots as well as expanding the underlying disk size then your best option is to roll the snapshots back one at a time taking a clone of the disks at each stage so you can retain the versions of the data that you need. Of course this takes up a whole lotta disk space, and assumes you can still commit the snapshots.

I hope you haven't done something fun like a major ESX version upgrade or a storage migration on the snapshot, because that can also stop you from making further changes.

In all respects I very strongly suggest you take an independent backup of the data before you go any further with this just in case you corrupt something you can't fix.

Good luck!

thehoff

ASKER

Thanks for the replies, guys. Unfortunately, these are test servers for our test team and we absolutely *must* use snapshots to go back to/upgrade from various versions. Typically the snapshot would be used for a week of testing, and at the end of that, another snapshot would be taken to keep that state, and the system(s) would be reverted back to one of the earlier snapshots to go back to a release version to upgrade from.

So some of the older servers are starting to fill up with test data and add-on software that's been installed. It is a *lot* of work to build a new test server (as typically there are at least 2, sometimes up to 6 servers that are all linked in certain states), hence my question here.

Is there no workaround or "cheat"? Disk space on the ESX servers isn't so much of a concern (for now) so I'm open to all suggestions!

markzz

I would suggest setting up for VCB and taking a VCB at the end of the week or possibly automating via the scheduler a weekly clod or hot clone depending on your licensing but pleasse listen to the advice that you may and likely will lose the lot if you don't change this practise.

thehoff

ASKER

Yes, we do backup every week.

Please, don't worry about us losing data - we have that area covered. We have played around with resizing before and are aware of the problems it will cause if snapshots are involved, hence this question.

I am looking for a cheat/workaround in this case - something out of the norm - if it exists. At the moment, it is looking very unlikely... :(

ASKER CERTIFIED SOLUTION

DChaffey

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

thehoff

ASKER

We apparently only have 2 free support tickets for every renewal of the software so I don't want to use one on this, especially when it appears that there is no "ideal" solution to the problem. I guess we'll just have to pick one state and stick with that...