In the "request completed" log message that we generate at the INFO level today, we should include our best guess of the response latency. I suspect it would be valuable to measure this in microseconds, though perhaps milliseconds would be fine enough. We should use Instant to track this latency in terms of CLOCK_HIGHRES.
In the "request completed" log message that we generate at the INFO level today, we should include our best guess of the response latency. I suspect it would be valuable to measure this in microseconds, though perhaps milliseconds would be fine enough. We should use
Instantto track this latency in terms ofCLOCK_HIGHRES.