wankun
Hi, @mridulm Thanks for your review. In my opinion, I do not think we should add `ShuffleMergeId` in RemoveShuffleMerge. The push-based shuffle service will only keep the latest shuffleMergeId data...
> We should decouple current implementation details when making protocol changes, and make it extensible for future evolution. > > In this case though, it is much more straightforward -...
Thanks @mridulm , I will the the `shuffleMergeId` into the protocol later. This PR also has some other issues, I will fix it soon.
> Gentle ping @wankunde. Do you think you can update the PR soon? Please let us know if you need any help. Sorry for the late reply, I will update...
Hi, @otterc @mridulm If the spark application stops unexpectedly, there will be some leaked merge files. In our production cluster, yarn nodeManager will cleanup these leaked files if the disk...
I had signed the CLA,but the status is still cla-missing ? Can any one help me? [FWD__Spark_Cassandra_Connector_CLA.pdf](https://github.com/datastax/spark-cassandra-connector/files/2618376/FWD__Spark_Cassandra_Connector_CLA.pdf)
I'm sorry where is smj_leftIndex_0 reset to 0 ? https://gist.github.com/zml1206/a27350b8849951e6efac0fb6088e527f#file-full_outer_sort_merge_join_codegen_after-L280-L305
Hi, @mridulm , I've been working on some other issues recently. If @yabola can do all or part of this task in https://github.com/apache/spark/pull/38560, please go ahead.
Hi, @mridulm @yabola I have updated this PR, could you help to this PR again? Thanks.