Hi all,
According to the most recent log files, it appears that one of my users ignored a "The specified virtual disk needs repair (14)" error long enough that the vdisk is now corrupt. The corruption appears to have occurred when his host froze and he powercycled the machine. On restart the VM no longer starts, complaining that "The file specified is not a virtual disk". First thing I did was confirm that the vmx is correctly pointing to the root vmdk and not one of the slices (we're using 2GB slices - non pre-allocated). The vmx looks clean. I then confirmed that the descriptor file is ok. It shows all 31 slices and seems to be ok. I think ran a batch file to scan the header on all 31 slices to confirm that each slice header was intact. They all start with "KDMV..." which I think means they're ok?
On a whim I extracted the last known configuration information from the most recent log file (vmware-0.log) which was written to the same day that the VM died. I created a new .vmx file from that data and tried to start the VM again but get the same error. That tells me that the the old vmx and the new vmx are the same and not really the issue. So then I moved on to the descriptor. When I created this VM, I made a "golden image" and simply copy it to each user, so I have an original vmdk file. I replaced the "broken" one with the "golden" one, and get the same error. So, I'm assuming that something is wrong with one of the slices, but I can't figure out what.
Additionally, there were lock files and vmss/vmem files that were left over from the crash. I deleted those, with no visible change. I also tried mounting the vmdk file in VMWare Disk Mount, in WinMount, in OSFMount and a few other tools I could find. I can't get vdk.exe to work in my x64 host though.
Any help would be appreciated.