Logical volume failed somehow - restore of backup problems

[Thomas2]

Hi, i have proxmox VE running on an Intel NUC (256GB SSD) containing my Homeassistant installation. Last time i tried to reboot Homeassistant VM it got stuck somehow on boot and since i switched off my NUC hard (power button). Upon restart i logged in to the proxmox webfrontend and saw a lot of errors in the VM's console trying to boot Homeassistant:



Now, i do mention the following first repair attempt since i suspect it has something to do with not being able to restore a backup:
As i am no Linux pro i tried to google what might have happened andhow to fix it. Since i did not suspect a catastrophic hardware error (smart looks fine and the system is rather new) i've shut down the VM and tried

Since the command did not output if anything has been repaired at all and i take from the manual that it replaces the corrupted data (keeping a copy of the unrepaired version) i tried to boot the VM again to no further success (same behaviour).

Lastly i tried to restore a backup for the VM by selecting the (non-running) VM in the Proxmox Webinterface and then choosing "Backup" and selecting a .vma.zst file i have on an external network storage. However, upon pressing "Restore" button and confirming that the "VM100" will be permanently erased/overwritten i get the following problem, for which i have no clue what to do:


Has the first repair attempt somehow set a readonly flag on the metadata volume and now prevents the backup restoration from overwriting it ?
(...pure speculation - my knowledge ends here...)

 

[Dunuin]

Last error message sounds line your LVM-Thin pool is full. Did you check that? A LVM-Thin pool should never reach 100% or data might corrupt and nothing is preventing this from happening other than you always monitoring it and deleting data before that happens.

Also never try to repair the storage without doing a blocklevel backup first, so your trys to repair it can't damage it even more.

 

[Thomas2]

Thanks for your answer, Dunuin. I think i am not 100% sure how to actually check if the LVM-Thin Pool ist full. In the pve under the node "nuc" i have options to show Disks->LVM and Disks->LVM Thin.

Under LVM i have:

And under LVM-Thin:


lvs shows me quite some volumes, but how the sizes stack or are (partially) included in each other...?

 

[Dunuin]

The LVM-Thin pool is 19% filled. so that should be ok.

 

[Thomas2]

So any suggestions on how to fix the "Thin pool needs check and has read-only metadata" problem preventing me from restoring a backup?

 

[Thomas2]

There was no way that i've found within one week of daily research allowing me to perform the "check" of the pool and repair the system.
I have made a complete reinstall of proxmox and thankfully was able to restore an external backup of the VM. Possibly this was a bad sector (as smart shows 2 of them) - but no idea if they were there before. Keeping a close eye on that.

Thanks for your help anyway.