In the reaper thread we reset these metrics on the beginning of the reap thread. This can cause an issue where we are never able to see the actual reported metrics in T2.

The periodic job scheduler will attempt to run the function once per time interval, if the reaper thread takes longer than the scheduled interval the reaper function will be called immediately once the previous iteration has finished. When we reset the values here that means we are never able to report the previous iterations values in T2, which leads to an observability gap. We saw this recently in a HELP ticket which made debugging difficult

I am proposing that we instead opt out of resetting the metrics here, and report the previous runs values until we finish the next run. There is a metric transactionReaperJobCount that indicates when the reaper thread has started, so we still know when a new reap has begun.