Bo Yang

Cool, thanks for willing to work on this! Several options I could think of for S3OutputReporter: 1. Batch data inside the profiler and upload to S3 2. Upload data to...

The reason for the relocation is to prevent potential conflict between jvm-profiler's dependencies and spark application's dependencies. User's spark application may have all kinds of dependency, and jvm-profiler is attached...

You could create a new profile in pom.xml and remove the relocation. So it won't impact others.

Yeah, this is interesting. We haven't got time to look into this. If you get any findings, welcome to submit any PR.

Normally we report all metrics and filter them during query time. If you would like to add some threshold inside profiler, please also go ahead. Probably control those thresholds by...

Thanks for reporting the issue. We haven't tried the profiler with spark client mode yet. You are welcome to submit a fix if you are interested to solve this issue...

I see. It just looks weird that Arguments.java contains a Reporter field. If you want to reuse Reporter, I would suggest change it somewhere else. e.g. cache the reporter returned...

Yeah, a lot of small index files will not work well in object storage like S3. Maybe good idea to store in other places. Or is it possible to serve...

> @liangrui1988 i'm considering trying to add this support to RemoteShuffleService, would you be interested for me to try and contribute it to this repo? Yeah, you are welcome to...

> Thank you. One more question.Spark. Shuffle. RSS. DataCenter = dc1 the dc1 is to point to?Does local storage mean disk storage?How do you specify multiple disk directories? "DataCenter =...