A bad day .....
this happens on ESX3.5 U2 and U3 (separate server)
When i start a vm or play with the disk file of this machine, Boooom purple screen with a nice code equivalent to "cannot find local disk"
Here is the story of the killing vm
This week-end we had a crash ...... all the ESX cluster was down due to a network outage on the console port ( ok it start bad...)
The vm machine is a small web server with a huge snapshot (bad again ...... i know)
During the crash we lose the vmdk descriptor file, but we rebuild it with the help of some nice post from this community
When we restart he vm ( and this happens normally) the machine was running on the snapshot but we cannot see it in the snap manager
After some research in the forum, we create an new snapshot, run 2 min on it, and ask to delete all existing snapshot. This happens correctly no more snapshot file on the disk
We stop the vm and restart it and ........ purple boooom and 20 vm out
I start it again on another esx .... purple boom and again 20 vm out (really bad days)
I start it again and again and .... on a plenty new ( but empty) ESX 3.5 U3 and the result is always the same, Purple every time
we've tested
A copy on another LUN,
a copy on the local disk
a "mount" of the disk on another vm
For the moment i'm testing a "duplication" of the vmdk file with a DD and i'll test it tomorrow morning
But if somone have an idea during the (european) night , i'll read the forum before restarting it tomorrow.
Any ideas are welcome
Many thanks' in advance
Ph Koenig, aka UnclePhil
I wonder if the snapshot is commited properly to the VMDK. Are you sure you dont see the snapshot files anymore?
I would suggest you to import/clone the VMDK using the vmkfstools and try power on the VM after configuring the VM to the new disk. See if that helps
Do you have a good backup of this VM?
In any case the ESX Server should not cause a PSOD even if it has a corrupted VMDK file. I would suggest you open a Support Request with VMware and let them know. I sounds like a bug to me.
Regards,
Surya
Import was the next solution on the list
But the DD solution have solved the problem, and now i can start the a new clone of the vm without explosion
all snapshot file where removed from the folder and the good backup .... is not a good question
Btw i'll open a case at support, because i can show that a corrupted vmdk can cause a PSOD ..... but i need to finish the crash recover first
Many thanks' for your reaction
Ph Koenig
Good to here that your issue is fixed. And if you could post the DD solution that you have used, It will be helpful for anybody else who may run in to this issue.
Regards,
Surya