During object-replicator runs, stats are available regarding partitions replicated, scanned, etc., but this information is only available in the logs (by default emitted every 5min) -- but not available through recon. As a result, it is only feasible to gather these metrics in a large cluster with many object nodes by harvesting, parsing, and indexing logs, and then querying against that.
It is very valuable to be able to perform automation under various conditions that look at replicator progress, as well as to be able to tell whether a replicator is busy or not. If we can make these stats visible through recon at least as often as we do in the logs, any operator-specific automation/alerting/detection/etc. that revolves around these metrics becomes substantially easier.
Fix proposed to branch: master
Review: https:/