kamilwu commented on pull request #11555: URL: https://github.com/apache/beam/pull/11555#issuecomment-624088189
@Ardagan Thanks for your suggestions! > > I'd put graphs with similar arguments side-by-side, make graphs narrower. Goal is to see general sense of current status. Current layout is too sparse and it takes way too much time to scroll. > Set default dashboard interval to last month. If you feel month is not enough, choose value you feel more relevant. Done. I pushed modified version to the website (http://metrics.beam.apache.org) so that you can see what's changed. Also, I pushed some data to InfluxDB to make the review process easier. > Some space can be recovered by removing time scale. Do you mean X axis or Y axis? > I think that similar metrics can be put into single graph, ex: TextIO 1Gb [GCS, HDFS]. I tried this out, but four different data series (read_time x2 and write_time x2) in one single chart were not very readable. I think it's better to keep them separated. > I would add a short description of what tests are executed. We all can access code, but it would be much more handy. What kind of description do you expect? We have some documentation on what tests are executed in Beam website [1] and cwiki [2]. If something needs to improved, let's improve website/wiki content. I prefer to avoid repetitions between website/wiki and descriptions in Grafana, because it'd be hard to keep them in sync. [1] https://beam.apache.org/documentation/io/testing/#i-o-transform-integration-tests [2] https://cwiki.apache.org/confluence/display/BEAM/Contribution+Testing+Guide#ContributionTestingGuide-TestsofCoreApacheBeamOperations > Would be great if local docker-compose service can fetch some data. It will be much more easy to understand what would be shown on each of the graphs. This would also help a lot for debugging or modifying dashboards. I agree. Created a JIRA ticket to track the effort: https://issues.apache.org/jira/browse/BEAM-9889 > Might not be a part of this PR, but I'd be interested in seeing some summary window that shows only anomalies. This would be feasible if we introduced Kapacitor (a component responsible for detecting anomalies). We could write back Kapacitor alerts into InfluxDB and visualize them in a summary window in Grafana. This is not a part of this PR, but I have a plan to introduce anomaly detection this month. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
