MINOR: remove FetchResponse.AbortedTransaction and redundant construc…#9758
MINOR: remove FetchResponse.AbortedTransaction and redundant construc…#9758chia7712 merged 43 commits intoapache:trunkfrom
Conversation
0fe301e to
bae931b
Compare
2a3ce40 to
3b606da
Compare
ijuma
left a comment
There was a problem hiding this comment.
Thanks for the PR. A couple of initial comments.
86da652 to
1a5def5
Compare
There was a problem hiding this comment.
I don't batch the partitions again in this PR as it create a new FetchResponseData.
There was a problem hiding this comment.
Can you clarify what you mean here?
There was a problem hiding this comment.
Oh, I planed to remove all usages of this method from production (i.e KafkaApis should generate batched response directly). However, it can produce a big patch so I will keep this method in next commit.
|
fix conflicting files |
|
fix conflicting files again |
ijuma
left a comment
There was a problem hiding this comment.
Thanks for the updates. A few more comments below.
|
@ijuma Thanks for all your great comments. I have updated this PR. Please take a look. |
ijuma
left a comment
There was a problem hiding this comment.
Thanks for the updates. I think these are the last comments and then we should be good. :)
ijuma
left a comment
There was a problem hiding this comment.
I left one comment, the rest LGTM.
unrelated flaky |
|
Great to see this merged. @chia7712 Will you follow up with the items we split into separate JIRAs? I think the 3 key ones are:
|
https://issues.apache.org/jira/browse/KAFKA-12385
https://issues.apache.org/jira/browse/KAFKA-12387
ugh, I missed this one. opened: https://issues.apache.org/jira/browse/KAFKA-12410 |
thanks for this response. Only #10269 is in progress. I'm sorry that this PR brings a bunch of conflicting files to #9944. I'd like to merge #9944 before other minor refactor-related patches. |
|
Does the refactoring simplify #9944 or not? |
|
@ijuma on first glance, I'm not sure if it simplifies. But it will introduce changes. And these are likely changes we want to have eventually. |
|
So should we go forward with #10269 first still? |
Changes
INVALID_HIGHWATERMARKtoINVALID_HIGH_WATERMARKFetchResponse.AbortedTransactionbyFetchResponseData.AbortedTransactionFetchResponse.PartitionDatarecordSettorecordsPerformance Tests
loop 10 times and get average.
case 1: @parametrize(acks=1, topic=TOPIC_REP_ONE)
diff: -0.3994016685 %
case 2: @parametrize(acks=-1, topic=TOPIC_REP_THREE)
diff: +0.4917916785 %
Committer Checklist (excluded from commit message)