lanson
asked on
URGENT - Recovering data from orphaned .VMDK files in ESX Server 3.5
have a really messed up scenario with a Windows 2003 Server Virtual Machine running on VMWare ESX Server 3.5
The Story:
1)Overnight, the Datastore that 2 of my VMs run from filled.
2) I moved all of the files for one of the VMs (VM1 - our main Database server) to another datastore, using VeemFastSCP, removed it from the inventory in Infrastructure Client and registered it from the new location.
3) On startup - Poweron failed giving me a file Not Found Error.
4) The files present were as follows:
VM1.VMDK (165GB)
VM1000001.VMDK (5GB)
VM1000002.VMDK (17MB)
VM1000003.VMDK (50GB)
VM1000004.VMDK (50 GB)
VM1000005-delta.VMDK (31 GB)
VM1000006.VMDK (32 GB)
VM1000007.VMDK (23GB)
VM1000008.VMDK (500MB)
VM1000009.VMDK (50 GB)
5) I took a look at the Hard Disk of the VM and it's Disk File was set to VM1000009.VMDK
6) In a panic - I changed the disk file location to VM1.VMK and powered on. The VM powered up, and the guest OS started, but on investigation, all the files very as of the 16Dec09!
7)since this VM is a SQL server, and I have seperate, hourly file backups of all my databases that run on it, i decided to just run a quick restore on all my databases - with the previous night's data. That worked fine and I was able to get all my users back on-line.
8) However, one of our apps dumps raw data text files onto a fileshare on the VM. I have now lost all the files that were created since the 16th Dec.
9) To compound issues, I had overwritten my filebased backup of this server before I noticed the missing files!
My Question:
Does anyone know if it's possible to recoved the data from these now orphaned .VMDK files?
any help would be greatly appreciated.
The Story:
1)Overnight, the Datastore that 2 of my VMs run from filled.
2) I moved all of the files for one of the VMs (VM1 - our main Database server) to another datastore, using VeemFastSCP, removed it from the inventory in Infrastructure Client and registered it from the new location.
3) On startup - Poweron failed giving me a file Not Found Error.
4) The files present were as follows:
VM1.VMDK (165GB)
VM1000001.VMDK (5GB)
VM1000002.VMDK (17MB)
VM1000003.VMDK (50GB)
VM1000004.VMDK (50 GB)
VM1000005-delta.VMDK (31 GB)
VM1000006.VMDK (32 GB)
VM1000007.VMDK (23GB)
VM1000008.VMDK (500MB)
VM1000009.VMDK (50 GB)
5) I took a look at the Hard Disk of the VM and it's Disk File was set to VM1000009.VMDK
6) In a panic - I changed the disk file location to VM1.VMK and powered on. The VM powered up, and the guest OS started, but on investigation, all the files very as of the 16Dec09!
7)since this VM is a SQL server, and I have seperate, hourly file backups of all my databases that run on it, i decided to just run a quick restore on all my databases - with the previous night's data. That worked fine and I was able to get all my users back on-line.
8) However, one of our apps dumps raw data text files onto a fileshare on the VM. I have now lost all the files that were created since the 16th Dec.
9) To compound issues, I had overwritten my filebased backup of this server before I noticed the missing files!
My Question:
Does anyone know if it's possible to recoved the data from these now orphaned .VMDK files?
any help would be greatly appreciated.
Well dont delete them!
You are not supposed to point it the master VM1.VMDK, the behavior is normal when you do snapshots, original vmdk will not be touched and change to read only the moment you take snapshot, that is why you can normally copy/download the main vmdk file as the file is no longer locked by vmfs, any changes will be written to delta file instead
so instead of something like this scsi:0.filename = "VM1.VMDK" it is now like this scsi:0.filename = "VM1000009.VMDK"
For your case, try fixing the CID chains
Refer to this http://redshift10.blogspot.com/2008/04/vmdk-snapshots-and-importance-of-cid.html
Read it slowly I'm sure there are still hope, i had fixed the similar issue for few customers, it worked great! Hopefully it can solve your problem
You are not supposed to point it the master VM1.VMDK, the behavior is normal when you do snapshots, original vmdk will not be touched and change to read only the moment you take snapshot, that is why you can normally copy/download the main vmdk file as the file is no longer locked by vmfs, any changes will be written to delta file instead
so instead of something like this scsi:0.filename = "VM1.VMDK" it is now like this scsi:0.filename = "VM1000009.VMDK"
For your case, try fixing the CID chains
Refer to this http://redshift10.blogspot.com/2008/04/vmdk-snapshots-and-importance-of-cid.html
Read it slowly I'm sure there are still hope, i had fixed the similar issue for few customers, it worked great! Hopefully it can solve your problem
ASKER
ryder0707 - the problem is that vm1.vmdk has been running live now as the main vmdk, for 2 days.
I presume fixing the cid chain won't work since the vm1.vmdk had been updated?
I presume fixing the cid chain won't work since the vm1.vmdk had been updated?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
You may want to duplicate your remaining files to another folder or datastore, then create a new vm using the original disk as the boot image. Leave off a network adapter so you have no conflicts with the existing vm. I would then attempt to fix any cid issues and see if you can get that to boot.
ASKER
out of luck i'm afraid. the CID fix didn't work and I raise a call with VMWare.
It's impossible to recover once the parent .vmdk is modified.
It's impossible to recover once the parent .vmdk is modified.
http://www.ufsexplorer.com/solutions.php