Please help! Getting Invalid Configuration For Device 0...i know why but can't fix it.

[sean1017]

I'm in a bad situation here. I have a MSCS file cluster that was previously running with RDMs to a volume on a SAN that has since been deleted. I am used to working with cluster disks through ISCSI and had forgotten that i added those RDMs to the VM (for testing purposes) so after i reconfigured everything within cluster administrator to point to a new volume on a different SAN, i deleted the original volume.

Now when i try to start the VM i get an error message "virtual disk 'hard disk 1' is not accessible on the host. Unable to access the file "

So then I went in to remove those disks (because i don't need them anymore, all my connections are now direct through MS ISCSI) but when i try i get the error message "Invalid configuration for device 0"

So now i'm in a real bad situation...I can't start either node in the cluster.

HELP!

[oreeh]

Post the VMX file and the log

[sean1017]

thank you for replying so quickly.

I zipped the vmx file and all the logs and attached it.

please let me know if you need anything else. I hope it's as easy as manually editing the config.

i really appreciate the help. i need to get this up somehow by tomorrow AM and i only have gold support.

[oreeh]

Remove the VMs from inventory, then open the VMX file in an editor and comment out the following lines:

scsi1:0.present = "true"

scsi1:0.fileName = "/vmfs/volumes/471caa1f-3d6e96f6-df82-001b2103c0dc/FileCluster/quorum_disk0.vmdk"

scsi1:0.deviceType = "scsi-hardDisk"

Then add the VMs back to inventory and check if you can start them.

[sean1017]

Thank you!

I commented out those lines but that didn't work...it kept complaning about the other disks so i commented out that whole section including the SCSI controller those disks were using.

That allowed it to start but for whatever reason the system drive had been removed too (even though i didn't touch that section or that SCSI controller). So i re-added it and it booted! yah!

but then it appeared to reinstall the NICs once it booted up and one of them reverted to DHCP instead of the static address it had. so i added the static back and then realized that i couldn't ping the public either. i then swapped the networks in the config file (one of them is a heartbeat LAN) and that appeared to fix it.

So, after all that, i think it's actually working!

I owe you big time Oreeh!

[oreeh]

Reinstalling NICs after removing disks - weird

[sean1017]

it did the same thing on the other node. do you think it has something to do with removing and re-adding to inventory?

[oreeh]

The only thing that changes when removing and adding a VM to inventory is the UUID (and therefore the MAC address).

This doesn't trigger a new NIC install.

[wensophieyang]

Hi, there,

I got the same problem after the underlining NFS shares were removed, and I tried to remove the Virtual Disk from the VM. The vCenter reported "Invalid Configuration for Device '0'. And the problem happened on Esxi 5.0.1

I tried to follow the procedure in this forum thread. But after that, I got exteremly slow response in "ls" and vCenter "Browsing" datastore of this particular VM direcory.

Here is what happend:

1. I removed the vm from the inventory using vCenter

2. I modified the vmx file by removing all the related entries from scsi0:1 to scsi0:4 of the Virtual Disks on the removed NFS shares, and only leave the scsi0:0 - the VM OS virual disk, which is physically available still.

3. Try to browse the datastore from vCenter of this vm to locate the vmx file, vCenter could not list the contents of this vm directory. Try other VM directory, vCenter responded normmally.

4. Go to esx shell, cd the directory of the vm, do a "ls", it takes a long time (minutes) to display the directoy file list

Due to the slow response, I am not able to add the vm back to the inventory. Any idea about how I can recover the normal response time and add the vm back to the inventory?

Thanks!

-Wen