I have a server with a purple screen of death that I can't seem to resolve. It is ESXi 4.1 installed on mirrored internal drives, with one datastore on the local drives containing one virtual server. I booted the server to a USB key that I have ESXi preinstalled on but get put into vmkdbg mode because of multiple ESXi installations. This to me validated that it isn't a hardware problem with the server, more than likely a corrupt install caused by a power outage.
So if I go through the installation again to the local disks will this wipe out my local datastores? If I just need to get this server running again and recover the virtual machine that is on the local storage.
You can get the USB install to boot. Have a look at the following KB article. http://kb.vmware.com/kb/1035107
So if I wipe out the installation on the hard drive, will I still be able to access the local VMFS datastore?
I have just changed the file system type from FAT to nothing using fdisk once the server has booted. If the FAT partitions aren't readable then it doesn't cause the duplicate UUID halt. By just changing the file system from fat to nothing allows me to change back at any time. I have also added the overrideDuplicateImageDetection option to the boot.cfg file. I'll see if I can find the entry for you.
After you boot from the USB device log in to the to the tech support console and modify the /bootbank/boot.cfg to include the entry
http://communities.vmware.com/message/1733410#1733410