Server no longer booting

ITarrant

New Member
Mar 25, 2025
3
0
1
I have a server with 3x2TB in Raidz1 as setup during install on a Proxmox 5.2 installation. It seems in the last couple of days it has had an issue an now no longer boots, and I am only presented with the grub recovery menu. I tried using zfsbootmenu to see if any of it was intact and it presented me with a pool that could not be imported (last accessed by a different system) but one good drive out of 3.
If I boot with drives 1,2,3 connected I get the same result. If I boot with drives 1,2 or 1,3 connected I get the same result. If I bot with drives 2,3 connected I get no pool available.
This leads me to believe that drive 1 should be ok, but 2 and 3 are most likely done. My initial thought is that all is done and I should move on, but is there a chance I could get something to boot and potentially recover some data?
 
I have a server with 3x2TB in Raidz1 as setup during install on a Proxmox 5.2 installation. It seems in the last couple of days it has had an issue an now no longer boots, and I am only presented with the grub recovery menu. I tried using zfsbootmenu to see if any of it was intact and it presented me with a pool that could not be imported (last accessed by a different system) but one good drive out of 3.
If I boot with drives 1,2,3 connected I get the same result. If I boot with drives 1,2 or 1,3 connected I get the same result. If I bot with drives 2,3 connected I get no pool available.
This leads me to believe that drive 1 should be ok, but 2 and 3 are most likely done. My initial thought is that all is done and I should move on, but is there a chance I could get something to boot and potentially recover some data?
Could you try booting with only drive 1?
 
Do you have any other ways of verifying that drive 1 definitely is not the culprit? I know using a single drive doesn't make much sense for your setup but to make sure drives 2 and 3 are definitely the issue and are indeed "done", we need a way to test each drive in turn.
 
I have tried booting off a single drive. Tried each one separately and the system could not find an operating system. When I took all the drives out and placed them into a drive caddy, only drive 1 would spin up and be read by Linux disks on my PC. The other 2 I got nothing.
The strangest thing to me is with 2 or 3 disks installed I get a grub rescue menu, so there is some disk configuration still there but no way of accessing it that I can see.
From a live boot I can see a single drive using disk -l and lsblk is about the same.
 
Last edited:
my guess is that you have a failed disk in your volumeset, and its the one that has the boot partition.

You can use something like sysrescue livecd with zfs to boot the system and reassemble the volume from your survivors. If you can do that, you can replace the failed drive and rebuild your grub partition(s) on the rest of the disks. you can use the procedure here:

https://pve.proxmox.com/wiki/ZFS_on_Linux#_installation_as_root_file_system
https://pve.proxmox.com/wiki/Host_Bootloader#sysboot

Since this system was installed some time ago (pve5.2 you say?) its likely you are using legacy (bios) boot. make sure to set all three of your disks in your bios boot order, and once you regain access convert it to UEFI boot; if you're already on UEFI boot you should be good to go.
 
So systemrescue iso did not boot, so i built from source. still no boot. There was a big storm that hit the area, so maybe some components are a playing up. I am installing a new pve8.1 onto a spare drive I had to get these guys up and going again while doing a deep data search/rescue on the one drive I know works. I am not too hopeful.
As for backups, the client and I spoke about them but nothing eventuated, thus the search for data.
And yes pve5.2 is old, the system has been running without any issues for the last 7 years with security updates only. We had talked about migrating to new hardware and/or upgrading what was on site but the client did not see themselves needing it for more than another year or so. So we left it as it was. That was 2 years ago...