OPNsense crashed at some point, which I suspect is due to overheating of the box (no fans, so purely passive cooling), which then in turn led to a degradation of the network connection (1Gps -> 100Mbps). This then triggered error messages for the wireguard interface in Suricata which filled up the log in /var/log/suricata:
2024-12-23T15:34:39+08:00 ... suricata 62911 - [meta sequenceId="..."] [100501] <Error> -- wg0: error reading netmap data via polling: No buffer space available
This went on for about 10 days, with 4TB of log files generated (and compressed on-the-fly by the underlying ZFS filesystem), until the system ran out of memory and crashed.
To manually handle the log files, one should enable SSH access via "System > Settings > Administration", and temporarily allow sudo access for the non-root user. SSH authorized keys are found under "System > Access > Users: edit".