Hi Enis, Hecks yeah, there are so many different prefixes alone that make finding what you need nearly impossible without practice. I agree on the configurable metrics backend, that will allow to enable what is needed optionally. We just have to find a good approach. One of the things that FB has learned the "hard" way is that some metrics at scale are not sustainable. We could, even think about building in some reasonable heuristics to disable some high throughput metrics when a threshold is crossed.
Lars On Jun 16, 2012, at 1:47, Enis Söztutar <e...@hortonworks.com> wrote: > Big +1. > > We also need to rename some of the metrics to be consistent in naming, and > make some it more configurable (like enable/disable per-region / per-schema > metrics, etc) > > See: https://issues.apache.org/jira/browse/HBASE-5601. > > Enis > > On Fri, Jun 15, 2012 at 3:01 PM, Stack <st...@duboce.net> wrote: > >> On Fri, Jun 15, 2012 at 10:04 AM, Elliott Clark <ecl...@stumbleupon.com> >> wrote: >>> Right now they are very similar to >>> RegionServerMetrics/RegionServerStatistics and seems like there is a lot >> of >>> overlap. >>> >> >> This would be really great Elliott. You'd package the metrics into >> the region server report? protobuf it? Then these would go into >> ClusterStatus in master? Good one. >> >> St.Ack >>
