Thanks for the positive feedback! Seems like that the Apache JIRA is down. When they bring it back i will open a JIRA ticket to track the Geode/Grafana integration!
@Mike, I fully agree with that "we should start a discussion of exactly what stats should be pulled". At the moment the tool exposes all MBeans attributes and all Statistics metrics found in the archive file. But i doubt they are all needed. Furthermore in the case of the Statistic metrics i find very difficult to understand the semantics of all the metrics exposed. The documentation ( http://gemfire.docs.pivotal.io/geode/reference/statistics/statistics_list.html) quite modest. Are there any other internal/external documents on the subject? @Jens, if you refer to the jmxtrans tool (currently used in jmx-to-grafana) i think it might be better to get rid of it or heavy refactor it. We can discuss this on the JIRA thread (once i create the ticket ;) @Jens, @Udo, Regarding the packaging (deployment) i agree that we should aim at self-contained installation/deployment stack. One important question to address is which time-series DB should we use. Current prototype requires/depends on InfluxDB (MIT - license). There are some other option, but most interestingly would be to explore is we can make Geode itself a Grafana datasource (so eliminate the need of third-party db). Again we can continue this over JIRA. @Greg, I like the idea for "self-administering distributed system" ;) @Dan, indeed both external components (InfluxDB and Grafana) are with permissive licenses. Apache 2.0 for Grafana and MIT for InfluxDB ( MIT is compatible with Apache licenses). Cheers, Christian On 12 January 2017 at 01:29, Dan Smith <dsm...@pivotal.io> wrote: > +1 This looks awesome! Looks like grafana is apache licensed, so that > shouldn't be a problem with integrating this. > > -Dan > > On Wed, Jan 11, 2017 at 3:42 AM, Christian Tzolov <ctzo...@pivotal.io> > wrote: > > I've been experimenting with Geode-to-Grafana integration options. The > > geode-dashboard (https://github.com/tzolov/geode-dashboard) project uses > > Grafana dashboards for querying, visualizing and analysing Apache Geode > > (GemFire) historical and real-time metrics and statistics. > > > > An important goal was to provides an unified stack that can analyze BOTH > the > > real-time (JMX metrics) and the historical (archive files) Geode > > distributed-system statistics. > > > > The github documentation and the blogs below should explain the approach: > > https://www.linkedin.com/pulse/visualize-analyse- > apache-geode-gemfire-real-time-metrics-tzolov > > http://blog.tzolov.net/2017/01/visualize-and-analyse- > apache-geode.html?view=sidebar > > > > At the moment the tool uses InfluxDB as a time-series DB.But i've been > > considering adding support for Ambari Metrics Collector System as an > > alternative time-series DB. Later is supported by Grafana > > (https://grafana.net/plugins/praj-ams-datasource) and Ambari in-turn > > integrates with Grafana (http://bit.ly/2j34aIX). So if we add to the > mix the > > Geode Ambari service (http://bit.ly/2jd0MbS) It will make a decent > Hadoop > > friendly stack for Geode/Gemfire. > > > > Another even more interesting angle is to make Geode itself a Grafana > > compliant datasource (http://docs.grafana.org/plugins/datasources, > > https://grafana.net/plugins). So > > > > Do you think it would be worth bringing part of this work under Geode > > project umbrella? > > > > Cheers, > > Christian > > > > P.S. Note that this project focus on Geode metrics only but similar > approach > > can be used to explore business time-series. > > > -- Christian Tzolov <http://www.linkedin.com/in/tzolov> | Solution Architect, EMEA Practice Team | Pivotal <http://pivotal.io/> ctzo...@pivotal.io|+31610285517
