Greetings. Here is my situation:
vSphere 5.1 U2 across all hosts and vCenter
Yesterday I got a report that some users were repeatedly losing connection to one of our server VMs. I go into vCenter and I see a couple dozen attempts to vMotion svr-xapp-a05 (DRS initiated) resullt in:
Cannot open the configuration file /vmfs/volumes/539f37d4-c1a3595b-af7a-002481e28f6e/svr-xapp-a05/svr-xapp-a05.vmx.
An error occurred while creating temporary file for /vmfs/volumes/539f37d4-c1a3595b-af7a-002481e28f6e/svr-xapp-a05/svr-xapp-a05.vmx: The file already exists.
The vMotion only makes it to 14% before this error occurs. The VM then looses network connectivity dropping 5-10 pings and of course disrupting services.
The last successful vMotion was July 16 @ 12:10PM. The first failed vMotion was July 16 @ 1:50PM. On that day, between those hours our vCenter server was patched with the latest MS patches and rebooted with a reboot of vCenter having happened at 1:48PM.
The first thing I did to prevent further user disruption was to "pin the VM" to the host it is running on currently.
Now when I browse any datastore (All FC SAN storage from our VNX) regardless of if I am browsing from vCenter or directly on a host all VMs have a VMNAME.vmx.lck except for svr-xapp-a05 (The VM in the error above).
When I go through the process of checking for locked files VMs that have the .vmx.lck do correctly show the MAC address of the host they are registered to. The VM that does not have a .vmx.lck file (svr-xapp-a05) shows a MAC address of all zeros (000000000000).
I have not run into any problems vMotioning any other VMs.
Also, most (not all) VMs have a blank icon for the .vmx file instead of the standard VMware Logo icon when browsing the datastores. Shutting down (not just rebooting) and powering on a VM will restore the correct icon. When I right click one of the unknown file type icons of a .vmx the option to "Add To Inventory" is greyed out but if I shut the VM down and power it on again, restoring the icon, the option to "Add Ton Inventory" comes back.
This is across all datastores so I hope it is not a corruption issue.
The screenshot attached shows the results of a check for locked files for a working VM vs the non-working VM. By non-working I mean I am not able to vMotion the VM. Otherwise the VM is working just fine.
Any thoughts?
