Replanning happens when cached plan is found ineffective during a trial run and results in a multi-planning, which is expensive.

It's possible that replanning ends up selecting the same plan as the one that was discarded from the plan cache when replanning was triggered. In this sense, the replanning process was "wasteful". This ticket suggests adding a serverStatus counter which counts the number of wasteful replanning events for diagnostic purposes.