parthchandra commented on PR #44021:
URL: https://github.com/apache/spark/pull/44021#issuecomment-1866836994
That was pretty cool stuff you did in Safari!
I think we may not have to do too much work ourselves.
The way I see it, async_profiler is doing the mapping of the java threads
and stack traces already for us (and we know that both async_profiler and
honest_profiler avoid the safepoint bias problem so this is as good as it
gets). In addition there is a filter API to filter on one or more threads so
async_profiler collects events only for the given thread(s). The API for
filtering takes a `java.lang.Thread` as input.
The way I see it potentially working is: when a user asks to profile a task,
we start profiling for only the task's thread similar to the way a task stack
trace is being done today. Then we ship over the the collected data and display
it.
I'll have to play around with this though. There might be some gotchas in
profiling multiple threads simultaneously and/or some APIs might be private
making it
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
