The workstations had lost their NFS mounts, which makes them unhappy. While it was possible to log in directly on cdsfs0, there was a kernel stack trace in dmesg, so I rebooted it. It hung on shutdown (not terribly surprising), but power-cycling brought it back with no apparent issues.
Also, cdsfs0 was still trying to NFS mount h2boot which no longer exists, so I corrected this. 'df' shows that the filesystem usage on /ligo dropped from 66% to 64% after reboot, will have to look into that; not that I entirely trust the output from before the reboot based on the presence of a kernel stack trace.
The timestamp in the syslog for the start of the stack trace is Apr 27 05:08:27 (local). This is absent in the cut and paste from the dmesg (which is timeticks only) in the file attached.