January 13, 20215 yr Hi, Need some inputs... I "converted" my personal server to Unraid last week. Earlier I was running Ubuntu 18.04 native, but I wanted to switch to Unraid and run Ubuntu in a VM instead. I created a new array from the disks, and moved over the data there. Then I started up my old server by passing the nvme-disk directly to a VM and booted it up, so far so good. But I really want the nvme as a cache partition, and have vdisks instead. So i ran qemu-img and made a vdisk from the old disk, which also worked as expected. After running it a few days i formatted the nvme so it's now in use as a cache instead. Then yesterday I needed to reboot unraid, but the VM refused to shut down. It shut down all processes and I came to the line "Halt initiated" (or whatever it says) just before it should switch off the "power". I waited like 15 minutes, but nothing happened. So finally I just did a "Force Stop" so I could halt the server and add the new physical disks. After that the VM did not boot. I just came to the uefi-boot which could not find a bootable harddrive... I messed around with qemu-img a bit and converted the disk to other formats, ran checks, etc, but it did not find anything wrong with the disk. Then I brought back a backup from the original vdisk and also mounted the current vdisk. The backup booted up as it should, and also found the current vdisk. But according to cfdisk it was an empty drive without partitions... I installed testdisk and scanned the vdisk, and it found a partition table. I restored it to the vdisk, then the VM booted up again Phew! ....but what happened? And how do i prevent it from happening again? Before I tried to shut down the VM, before this happened, I moved ~300 Gb of data from the vdisk to the array. I assume that this is not a very normal operation ;). Could this have caused this error? Ideally I should have reduced the size of the partition before creating the vdisk, but as I did not want to "risk" anything I just made a vdisk of the complete harddrive = 466Gb .img-file. I also forgot the dir with 300 Gb of data before I formatted the nvme, so that's why I had to move it from the vdisk. Perhaps the operation I had made was not "synced" to the array? So the disk / partition table got corrupted somehow? I dont know exactly how vdisks work. Does the VM write directly to the array, or are the write operations "buffered" somewhere and then KVM is writing it to the array afterwards? Larger vdisk = longer delay before changes are written to aray = bigger risk of failure? So a smaller vdisk is perhaps more safe to use? (The VMS are on the array with cache = no, in case that matters) This VM is only temporary. I'm installning a new VM and will move over the services there, and finally scrap this VM completely. The new VM have a 120G vdisk, is this a reasonable / safe size? I could make it smaller and then just extend when I need to. But as I have the space I would rather just let it be a little bit big to avoid extending partitions in the future. Or was this simply just "something that happened"? I could not find much info about others who have had similar issues... Might also have been an issue from when I made the .img from the physical disk as well of course? Just that it did not show up until now (~a week afterwards). Regards, Johan Edited January 13, 20215 yr by y0dase
Archived
This topic is now archived and is closed to further replies.