Hi Jacky, Some performance improvements for Flame Graphs related to how the stack traces are collected were recently merged [1]. Since you are working on a real-life environment with high parallelism, could you maybe run a couple of tests to verify that everything still works as expected after the change? Ideally we should now be able to remove the feature flag for Flame Graphs, but I would like to collect some data points first.
[1] https://github.com/apache/flink/pull/19228 Thanks, Alexander Fedulov On Fri, Mar 11, 2022 at 12:41 PM Yufei Liu <liuyufei9...@gmail.com> wrote: > Hi Jacky, > I think it’s a very useful feature, can get flame graph easily and avoid > run script personally. > But I think can package async-profiler library into flink, and use JNI > execute the profiling, doesn’t depend on local environment. > We did this way in out internal version, and provide api to download > generated html file. > > Jacky Lau <281293...@qq.com.invalid> 于2022年1月24日周一 16:48写道： > > > Hi All, > > I would like to start the discussion on FLIP-213 < > > > https://cwiki.apache.org/confluence/display/FLINK/FLIP-213%3A+TaskManager%27s+Flame+Graphs> > > ; > > which aims to provide taskmanager level(process level) flame graph > > by async profiler, which is most popular tool in java performance. and > the > > arthas and intellij both use it. > > And we support it in our ant group company. > > And Flink supports FLIP-165: Operator's Flame Graphs > > now. and it draw flame graph by the front-end > > libraries d3-flame-graph, which has some problem in jobs > > of large of parallelism. > > Please be aware that the FLIP wiki area is not fully done > > since i don't konw whether it will accept by flink community. > > Feel free to add your thoughts to make this feature better! > i > > am looking forward to all your response. Thanks too much! > > > > > > > > > > Best Jacky Lau >
