-
Notifications
You must be signed in to change notification settings - Fork 10.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
add --perf_args usage to jenkins performance runner scripts #9038
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm good with this, but I'm a little rusty on how all this fits together. @jtattermusch, @nicolasnoble, @matt-kwong any thoughts?
btw I just saw that these were the scripts called by |
I think you covered all of the performance jobs we run, but I would let Nico or Jan confirm. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Who sets the $PERF_ARGS env variable? Are they set in jenkins job config? If so, I'm not sure we want that. We have the tools/jenkins/run_* scripts because we don't want to have too much configuration done in Jenkins UI (where it's not versioned).
Also, won't enabling the perf args influence the benchmark results?
@@ -67,6 +69,7 @@ tools/run_tests/run_performance_tests.py \ | |||
--bq_result_table performance_test.performance_experiment_windows \ | |||
--remote_worker_host grpc-performance-windows1 grpc-performance-windows2 \ | |||
--xml_report report_windows.xml \ | |||
--perf_args "$PERF_ARGS" \ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
these are running on windows, using "perf" command will probably break them.
For setting the From what I've seen, running under perf has been pretty low overhead, maybe 5-10% Maybe this should change to hard code the perf parameters? Or run perf on different jobs? |
Sorry I'm actually thinking this will not work out very well, mostly because the Thinking these flame graphs might want to stay on the |
Yeah, that sounds alright - I think it makes sense to make these benchmarks run where the sweep suite is running. We already have enough stuff running on the perf workers where multilang configurations are running. |
d36c7a8
to
dbce6e9
Compare
dbce6e9
to
66c6782
Compare
Latest updates add flamegraph directory name option to the driver, for running multiple PTAL, this should be ready for review For the new jenkins job: |
Is this still active? |
Yes this should be ready. Actually the jenkins job created along with it is still alive in https://grpc-testing.appspot.com/job/gRPC_performance_flamegraphs/ - the most recent job there was has flame graphs for c++ and go attached. (btw the new job calls the new |
This is amazing. Please merge and enable it for sweeps. |
failed tests were under log of failures in https://grpc-testing.appspot.com/job/gRPC_pull_requests_Bazel_full/38/console |
@ctiller, done, these reports should start showing up in the sweep job |
heads up on sweeps these aren't running yet - needed to re-run worker init scripts. |
needed to use the perf flame graphs on more jenkins performance jobs