That would be my guess also.
The nodes have all been pushed a fair bit lately and RaspiOS defaults to a very small amount of VM (just 100MB, I think) so it may well be that.
How much memory does your model have?
aaaaah, damn - it was my fault. topic can be closed: i have a maintenance script running manually - this includes shutting down the nodes for an automatic package update and rebooting them. the latest update of the script was copied to my health-check-script, which runs once per hour. haha, a noob issue.