Sounds a good plan to me. Thanks Enrico. Regarding 4lw - my opinion is we could just leave it as is and don't expose any of the newly added metrics through 4lw endpoints. Instead, advocating admin server end points which will be supported moving forward.
On Tue, Sep 11, 2018 at 7:41 AM, Enrico Olivelli <eolive...@gmail.com> wrote: > Hi, > I would like to share the plan for the new "Pluggable metrics system for > ZooKeeper" [1] > > Currently on master we have the MetricsProvider API and the management for > the lifecycle of the MetricsProvider (start/stop/configuration....). > > Next immediate steps, which can run concurrently: > - Implement a first "provider" (Prometheus.io) > - Integrate the code with the new system > > There is an ongoing activity from Facebook guys to back port all of the > metrics they added in their private fork "Implement additional server > metrics" [2] > > My proposal is to wait for the integration of [2] and then proceed with the > integration with Metrics Providers. > I think that it would be better that Allan Liu or Joseph (or some other guy > from Facebook) drive this migration. > > In the meantime I will work on the implementation of the Prometheus > provider > > An important point is to decide how to support 'legacy' 4lw commands, my > idea would be to implement a "FourLettersWordsMetricsProvider" which > collects data and serves metrics to the legacy endpoint. > > Thoughts ? > > > [1] https://issues.apache.org/jira/browse/ZOOKEEPER-3092 > [2] https://github.com/apache/zookeeper/pull/580 > > Enrico >
