Over the past few days and for the foreseeable future I'm running a python script which looks for ADC,TIM,IPC glitches on the end station computers. It scans roughly once a second (not synced to GPS but rather runs a loop and then waits one second). For all models running on the machines h1suse[x,y], h1seie[x,y] and h1isce[x,y] it reports any STATE_WORD which have non-zero values after the upper bits have been masked. For any model with an IPC error, it scans each IPC receiver channel and reports any non-zero error counts (plus IPC type and name of sender). Any latched errors are subsequently cleared by the once-a-minute issuing of a DIAG_RESET by a cronjob (actually happens at about 3 seconds in the minute).

Once I've accumulated some statistics, I'll post a report.