Since I enabled detailed read-only logging, our FR10 telemetry display sometimes stalls while the log folder is busy; cell controls are separate, and although synchronous writes look suspicious I have not ruled out delayed acquisition, so what timing would separate them without adding more load to the same recorder?
Measure acquisition, decoding, publication and file-write time separately with a monotonic clock. Keep that small timing record away from the suspect write path, and compare equivalent observations with detailed logging on and off. Also record the age of the displayed reading; a frozen screen doesn't establish an offline controller.
I'll separate those stages and use a bounded timing capture outside the detailed logger; no measurements back yet, and the existing screen does not make the age of its last reading obvious enough.
Fix that presentation expectation alongside the diagnosis. If you later move writes into a queue, decide its limit and failure reporting before calling it solved. A responsive screen with silently missing diagnostics would answer only half your original problem.