@vdusek The current behavior of purge was implemented during the storage refactoring. Perhaps there was a reason for keeping the handled_request_count' and total_request_count' counters?

Yeah, I initially thought total_request_count should represent the actual total number of requests this instance has processed (even if purge was executed), with only a drop resetting it. But if that's confusing or inconsistent with the JS version, then let's change it.