Corrupt VM?

[MG23]

I am tinkering with ESX Server 3.5. The power went off over the weekend. When I boot up my VM which is acting as the only DOmain Controller (running server 2003 ent), it halts at the windows 2003 screen. The one with the bar at the bottom. The bar keeps moveing but never stops. I cant boot into safemode. I tried to run a chkdsk using windows repair but it frezzes when examining disk comes. I tried to view the vmdk using vmkfstools -g my.vmdk -v 5 ... but is says that the specified file is not a valid vmdk. I also tried to transfer the file to my desktop using VC but I get invalid I/O. I unfortunatly have no snapshot or backup. I am assuming that the vmdk is corrupt, but why does it attempt to boot windows?

Any suggestions?

[gguntz]

I have run into a similar situation albeit on a physical machine. Since you have nothing to lose, you might try booting with the network connection "unplugged". I have seen when you have only one Domain controller you get stalled out on boot up especially if there is a big time difference. If you can get logged in with the connection still "unplugged" you can connect it and sync up the clock. I would also reinstall vmtools to ensure you do not have a driver issue. It is likely that the MBR is still okay just something else is out of whack.

HTH

G

[MG23]

Ya but why cant I copy the file over to my desktop?

[COS]

Remove all snapshots first then....

Shutdown the VM, log into the CLI of the ESX server, go to the directory of the VM in question, delete any files .lck or .mem or any long randon GUID named type files that are the same file size as the assigned RAM of the VM, kill all log files than try restarting your guest VM.

[MG23]

Could this problem come from hitting the reset button and not shutting down the guest? I dont see those files through VC browsing the storage. I will go into the CLI in morning and see if any of the mentioned files are there. Thanks

[COS]

yes and also if you are using iSCSI or Fibre and the connection gets lost.

[MG23]

No Luck.. there are no .lck or .mem files in there.

[ianmellor]

Hi,

You can try to clone the vmdk with vmkfstools and then remove the disk from your vm and re add it back and power on. We had a corrupt vm and did this and it worked.

Hope this helps.

[MG23]

didnt work error said cannot clone .. Input/output error

[COS]

It looks like the dirtual disk is corrupt. Try using Acronis Dis Director 10.0 to fix it. If that fails, try using Ghost ro Acronis True Image to cut an image of the VM then wipe out the VM (Delete from the ESX server) recreate the VM and drop the image back on the new vm.

I ran into this before, if in fact the virtual disk is bad use any third party util to boot to and fix it.

[MG23]

Any good documentation on the procedure? I have never used acronis Dis Director. DO I boot the ESX server from this?

[COS]

No, you boot the VM/Guest that's having problems with it. Typically if it sees an error on the drive it will prompt to fix it.

[MG23]

Using Disk Director 10 I can see the NTFS partition. It never asked me to fix anything. However I can see all my files. That seems like a good sign. How would I clone. Create a second Virtual Disk in VC for taht VM then clone to that disk?

[COS]

That's a good sign. DDirector won't clone. Acronis True Image Server or Ghost or whatever 3rd party imaging should allow you to cut an image or clone to disk or pull your files out of the disk.