The node had some hardware issue and failed to use the hard drive completely, rendering it unusable and needing a hard-reboot from impi.
Found out because the VMs hosted in it had stopped running (sowed unowned by horizon logs).
The node was still running, just unable to read/write to hard drive, throwing many I/O error every time you tried anything and failing to find some binaries (ex. reboot).
After hard rebooting it it came up online ok, dmesg clean, and journal shows logs only until 5min before the VMs it hosted stopped (nothing suspicious before on a first pass).