Possibly a Zombie VMDK file ! Please Check

[RajuVCP]

Hi All,

I recently installed RVtools,(RVtools is a free tool and lite weight tool to view and get information on all the esxi/esx host)

In RVtools on one of the option vhealth check there i can one of the suggestion as

"Possibly a Zombie VMDK file ! Please Check"

what does it mean , and what can be done to remove that error.

Does any one got the similar message.

Regards

Raju Gunnal

[arturka]

Hi

the zombie VMDK are files which are "most probably"not connected to any virtual machine.

Cheers
Artur

Please, don't forget the awarding points for "helpful" and/or "correct" answers.

[nielse]

A zombie VMDK is as mentioned usually a VMDK which isn't used anymore by a VM. You can double check this by checking if the disk is still linked to the VM which it should be a part off. If it isn't you can delete it from the datastore via the datastore browser. I would suggest moving it first before you delete is, just in case.

[Texiwill]

Hello,

As nielse stated, a Zombie virtual disk is a virtual disk that MAY or MAY not be connected to a VM. I get these from time to time on my monitoring tools as backups are performed as some backup tools use Hot Add to do the backup. If they use HotAdd you will get a snapshot connected from one VM to another and as it is removed from one VM it shows up in the monitoring tools as a Zombie. Other times I get real Zombies.

to find a Zombie you must be able to read the snapshot file and disk meta files to determine if it really is unrelated. While the tool does this for you, if you have backup tools in play, it may not recognize hot  add and mark it as a Zombie.

Ulli Hankln at www.sanbarrow.com is 'the man' for determining what is connected to what within a virtual disk and most likely could assist further.

Best regards,
Edward L. Haletky
VMware Communities User Moderator, VMware vExpert 2009, 2010, 2011, 2012

Author of the books 'VMWare ESX and ESXi in the Enterprise: Planning Deployment Virtualization Servers', Copyright 2011 Pearson Education. 'VMware vSphere and Virtual Infrastructure Security: Securing the Virtual Environment', Copyright 2009 Pearson Education.
vSphere Upgrade Saga -- Virtualization Security Round Table Podcast

[annakalaj87]

You can double check this by checking if the disk is still linked to the VM which it should be a part off. If it isn't you can delete it from the datastore via the datastore browser. I would suggest moving it first before you delete is, just in case.

How how do you do this?

thanks in advance!!!!

[stevenstapleton]

To move or copy the zombie vmdk or any other file.

Depending on if the server has a hold on the file you may have to power off the guest to complete

1. select the host, can get the host from the guest summary screen

2. Configuration Tab

3. Storage Hardware Menu

4. select datastore

5. Right Click - Browse Datastore

6. From there should see a list of Folders

7. Right click in the right panel on a blank spot - create a new folder

8. Now open the server folder that contains the zombie file

9. There will be several files listed, look for the TYPE Virtual Disk

10. carefully select the suspected disk.

10a. disk names should be servername_1.vdmk (where _1 is disk one _2 is disk 2

10b a lost snapshot may be -1-00003 or -1-00004

11. With the correct vdmk selected right click and select copy, select your new folder (step 7) as place to send it to.

12. open the new folder and right click and Paste

13. after the file is copied, shutdown the server is still on

14. Delete the original file

15 attempt to start the guest, if there are no errors the file was not needed.

16 you can now safetly delete the copy from the new folder.

17. if for some REASON the file will not copy or move. you may have to shutdown or relocate other guest on the host, shutdown the remaining guest

18. place the host in maintenance mode and once complete, restart the esxi host. then attempt the copy again.

[bobyuen]

We use RVtools for reports and checking for Zombie and sometimes known orphaned VDMKs.

This usually means that the VM folder contains a VMDK that has no association with the VM based on the VM configuration settings. Each Hard Disk has it own VMDK.

We often get a lot of Zombie/ orphaned files due to our CommVault backups. I'm not the backup admin but we know we find Zombie files after a backup job is incomplete or if someone removed a second hard drive-VMDK from the Setting but did not remove the VMDK permanently. Sometimes it could be a snapshot process that has been corrupt.

If you run the vHealth report and it indicates that you have a zombie file, do the following safety checks.

1) Check to make sure you have no snapshots

2) Check the setting for your VM configuration your your Hard disk(s) is assigned to a proper VMDK file name. Meaning it should NOT have your VMname ending with 000001.vmdk.

3) Browse the VM folder and take a look. If the reports indicates your VM has a zombie file, you should see one or more  VMname.000001.vmdk. The report will tell you if you have more than one.

The simple fix, simple if the VM's VMDK are small. All you have to do is run a Storage vMotion of the VM over to another LUN. Once the move is complete, whatever is left over in the original source LUN can be deleted if you know you don't need those VMDK files.

When this become not so simple is when you have multiple VMDK and if they are larger than 100GB. Larger VM's with multiple disk may take a whole day or longer complete. It will vary depending on your setup.

If you don't clean up your Zombie/Orphaned disk, you'll will soon have to buy more storage.

P.S If in step #2 you find that your VM has a weird 00001.vmdk and vCenter indicates that you DO NOT have any snapshots, you better open a call with VMware support to get that VM corrected.

Good Luck.

[darshanarjoshi]

Hi ,

Is there any other way other than RVTOOLS report to know this?

If there is some script that can be run on powerCLI that could give answers!

Also if there is some other way to know which are the zombie VMDK files. It will be of great help.

Thanks and Regards

Darshana Joshi

[continuum]

One man's zombie.vmdk is another man's temporarily unused vmdk.

So be very careful when deleting zombies - just because a script lists a vmdk as potential zombie.

There are different rules to identify a zombie:

1. all vmdks that do not belong to a currently active VM are potential zombies

2. all vmdks that do not belong to a registered VM are potential zombies.

3. all vmdks that are neither referenced by a vmx-file nor appear as "parentFileNameHint" in any  descriptor-vmdk are potential zombies

For a good Zombie-detection you need to list all available VMDKs across all datastores in a cluster.

Then you must compare that with the list of all VMDKs referenced in any vmx-file.

Depending on your zombie-definition ( see 3 options above) the results will vary.

So before you automatically start to delete zombies you may consider to move the zombie-candidates into a directory for each datastore called "zombies-will-be-deleted-every-last-day-per-month"

I do not remember which Zombie definition is used by RV-tools but last times I looked the buildin logic did a good job.