I found a way to trash the partition table and I am looking for a way to prevent it. The pattern is consistent. I have a VM with Ubuntu 14. I have a solid state drive with Ubuntu that runs on a Minnowboard Max, and the main file system is locked readonly and needs repair. Plug the solid state drive into a Apricorn cabal so that it has a USB to disk drive. Plug the USB into the host, my mac. The VM will run for a minute or so and freeze. This was long enough to notice there was not /dev device created. You can then try to stop the VM, and it will spin forever and not complete. Kill Parallels and start it, goes back to spinning. Kill Parallels again, and start it, and you have a trashed VM file system. Run gparted and you will see that the partition table is hosed. What I thought should happen is when I plug in the disk drive with USB, it would simple make a /dev/sdbN or similar, and I could run fsck on it. I tried to do the same while booted on a gparted image, and it also freezes. But here I noticed doing a Stop got me out of trouble without trading the file system or partition table. Is there some general mistake I am making by trying to plug in a primary drive from another system to repair the drive?
