Boot the VM from a Linux ISO and mount the disk inside the VM and delete files to make space?
That's a good idea, of course. But I wonder how something like that could happen. Probably a PBS problem. That doesn't exactly increase my confidence in a (bare metal) PBS, though. I've been familiar with tricks for creating temporary free space by moving data blocks for a long time. Where is the necessary emergency brake on PBS to send an error message when x% remains? Annoying. Who wants to be fiddling with their backups all the time?
The disk was probably thin provisioned and the VM wanted to use space that wasn't actually available on the storage, which causes I/O errors. This happens to VMs of people every so often on this forum. Depending on the storage type, the available space might be less than expected because of (ZFS) overhead/padding or snapshots or discard not (fully) working.
I think this is a PVE over-committing sparse storage by the administrator.
I love running PBS in containers (and syncing to off-site installations). Maybe talk about your PBS doubts to people in the specialized sub-forum: https://forum.proxmox.com/forums/proxmox-backup-installation-and-configuration.24/ ?
