This problem cropped up when I started updating ESX host servers to 3.0.2. When I try to migrate VMs off a host, it fails with error message:
>>Unable to migrate.
>>Unable to access the virtual machine configuration file. A file was not found. *** .vswp
This has happened across 8 different ESX host servers. All were running 3.0.1 and to be upgraded to 3.0.2 Update 1.
Not all VMs are affected. Maybe about 15-25% of total VMs. Strange thing is, after I click on Enter Maintenance Mode, DRS generates migration tasks, and some number of those complete successfully. In the lastest case, about 20 migrations completed, then the process just stopped. Now every time I try to do a migration manually, It fails with the same message.
Opened a case with tech support. Was told this was caused by a file lock not being released. Recommendation was to reboot the ESX server, which would clear the file locks. We have had SAN problems recently, which I can imagine could potentially result in file lock problems.
I found by trial and error that I can power off the VM, and immediately restart it on a different host and the problem goes away.
Right now I have about 20 VMs with this problem on the last host to be updated. It will take a lot of time to contact the VM users and arrange to power off the VM.
Question is, what is the best thing to do? Is there a better way than negotiating downtime with many different users? Is there some way to remove file locks on .vswp files without taking down all the VMs?
