Github user ptgoetz commented on the issue: https://github.com/apache/storm/pull/2203 @erikdw Of course! This needs to be documented, and if accepted (or likely to be), I will do so. But if there's community consensus around NOT accepting this patch, I'd rather not waste the effort. This only addresses the sub tasks of STORM-2153 that are marked "in progress." The other issues, particularly the RocksDb, is likely off the table for now, since RocksDb is now considered category x (on a side note, a number of Apache projects like Kafka and flink are in for some hurt in that respect, since they will likely have to migrate grate off of RocksDb very shortly). As far as your other question, aggregation is left to the destination metrics system. The individual metrics are namespaced with topology, host, port, etc. that uniquely identify each event down to the component/executor level. The namespace uses dot (".") notation to make it easy for tools to parse. You can effectively disable this feature by not registering any reporters. The metrics objects will still be there, but there will be no reporters doing anything with them. If would be a fairly simple optimization to only register the metrics objects if there are one or more reporters registered. (IIRC this was one for f the concerns you raised in one of the discussion threads...) @HeartSaVioR See above for some of your questions. This is an independent *alternative* to the existing metrics system. The goal is to make things simpler. I think a lot of users have struggled with storms built in metrics system, if they have approached it at all.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. ---