Why is app_time fetched via CLOCK_MONOTONIC in addslave and active (for DC distributed clock initialization), yet switches to CLOCK_REALTIME during periodic operation? Slave hardware may not jump abruptly, but the master's app_time still changes. What’s the rationale for this design?
Why is app_time fetched via CLOCK_MONOTONIC in addslave and active (for DC distributed clock initialization), yet switches to CLOCK_REALTIME during periodic operation? Slave hardware may not jump abruptly, but the master's app_time still changes. What’s the rationale for this design?