Our read-only FR10 status display sometimes stops updating while the log folder is busy. It started after I enabled detailed trend logging. Cell controls are separate. I suspect synchronous writes, but communication delay is still possible. What timing evidence should I collect before changing the recorder?
Can you time acquisition, decode, publication and file writes separately with a monotonic clock? Keep the timing trace outside the suspected write path, and record the age of the displayed sample.
A small memory trace caught one pause after the reply arrived and before publication, while a synchronous file write occupied the same thread. Other requests are slower too. We have evidence for one logging delay, not an explanation of every pause.
Will the recorder proposal define bounded buffering and visible overflow or storage failure? Moving writes off the display path needs a stated record-loss policy, and it will not answer those slower requests by itself.