Your Ad 

Wednesday, January 14, 2009

Could not power on VM No swap file Failed to Power on VM

The following error occurred: Could not power on VM: No swap file. Failed to power on VM.
If you try to rename, delete or move the VM swap file, you get a "file in use" error.

Check the log file, should look something like this
Feb 01 15:03:31.486: vmx| DISKLIB-VMFS : "/vmfs/volumes/464420a7-a1e28a0d-6b66-0019bbd0c70a/MYSERVER/MYSERVER-flat.vmdk" : failed to open (1048585): AIOMgr_Open failed. Type 3

Feb 01 15:03:31.486: vmx| DISKLIB-DSCPTR: Failed to open extents for descriptor file in normal mode
Feb 01 15:03:31.486: vmx| DISKLIB-LINK : "/vmfs/volumes/464420a7-a1e28a0d-6b66-0019bbd0c70a/MYSERVER/MYSERVER.vmdk" : failed to open (Device or resource busy).
Feb 01 15:03:31.486: vmx| DISKLIB-CHAIN : "/vmfs/volumes/464420a7-a1e28a0d-6b66-0019bbd0c70a/MYSERVER/MYSERVER.vmdk" : failed to open (Device or resource busy).
Feb 01 15:03:31.487: vmx| DISKLIB-LIB : Failed to open '/vmfs/volumes/464420a7-a1e28a0d-6b66-0019bbd0c70a/MYSERVER/MYSERVER.vmdk' with flags 0xa (Device or resource busy).
Feb 01 15:03:31.487: vmx| DISK: Cannot open disk "/vmfs/volumes/464420a7-a1e28a0d-6b66-0019bbd0c70a/MYSERVER/MYSERVER.vmdk": Device or resource busy (1048585).
Feb 01 15:03:31.487: vmx| Msg_Post: Error
Feb 01 15:03:31.487: vmx| [msg.disk.noBackEnd] Cannot open the disk '/vmfs/volumes/464420a7-a1e28a0d-6b66-0019bbd0c70a/MYSERVER/MYSERVER.vmdk' or one of the snapshot disks it depends on.
Feb 01 15:03:31.487: vmx| [msg.disk.configureDiskError] Reason: Device or resource busy.----------------------------------------
Feb 01 15:03:31.501: vmx| Module DiskEarly power on failed.
Feb 01 15:03:31.501: vmx| VMX_PowerOn: ModuleTable_PowerOn = 0
Feb 01 15:03:31.502: vmx| WORKER: asyncOps=0 maxActiveOps=0 maxPending=0 maxCompleted=0
Feb 01 15:03:31.746: vmx| vmdbPipe_Streams Couldn't read: OVL_STATUS_EOF
Feb 01 15:03:31.746: vmx| VMX idle exit
Feb 01 15:03:31.757: vmx| Flushing VMX VMDB connections
Feb 01 15:03:31.763: vmx| IPC_exit: disconnecting all threads
Feb 01 15:03:31.763: vmx| VMX exit (0).
Feb 01 15:03:31.763: vmx| AIOMGR-S : stat o=1 r=0 w=0 i=0 br=0 bw=0
Feb 01 15:03:31.763: vmx| VMX has left the building: 0.

Kill the process
Run the following command on the Service Console to find out what process locks the swap file

ps aux | grep
If a rogue process locks the swap file, kill it and then try and start the VM.
Try this on every ESX server in the cluster.

Move the VM
Try to migrate the VM to a different ESX host and Datastore. It could resolve the lock on the swap file.

Check your LUN's
Check every ESX host in the cluster if all the LUNS are visible for the ESX host.
If a ESX host is missing a LUN, and the VM was running on that ESX host and LUN, the file is still locked after the ESX host lost connectivity to the LUN.
This was the case in our environment. After a SAN fabric problem, one ESX host was missing a LUN, and in a large environment you can easily miss that problem (12 node cluster, 40 LUNs')
Rescanned the LUNs, The rescan found the VM 'running'. Stopped the VM, and started it without a problem.

If everything fails
Shut down all the ESX hosts... This should definitely kill the file lock.

No comments: