junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-689838971
@chia7712 : Thanks a lot for staying on this tricky issue and finding a
simpler solution!
This is an automated
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-689241869
@ijuma @hachikuji @rajinisivaram : I think this PR is ready to be merged.
Any further comments from you?
This is
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-685831526
@chia7712 It seems there are some compilation errors in jenkins?
https://ci-builds.apache.org/job/Kafka/job/kafka-pr/view/change-requests/job/PR-8657/1/consoleFull
`
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-683859366
@chia7712 : Thanks for the performance results. It seems that the average
across multiple runs doesn't change much?
Also, 1 failure in the latest system test run.
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-683052000
@ijuma @hachikuji @rajinisivaram : Do you want to take another look at the
latest solution from Chia-Ping? It (1) solves the known issues completely; (2)
doesn't require new
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-681361030
The following is my understand. The current PR introduces a new deadlock
through the following path.
path 1
hold group lock ->
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-681109748
@chia7712 : Thanks for the reply. I like your overall idea and I think it
can be used to solve the problem completely in a simpler way.
1. Instead of at `Partition`, we
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-679256967
@chia7712 : Sorry for the late response. I just realized there seems to be
another issue in addition to the above one that I mentioned. The second issue
is that we hold a group
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-671520491
@chia7712 : I agree mostly with your assessment. For most delayed
operations, the checking for the completeness of the operation and the calling
of onComplete() don't have to be
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-670113688
@chia7712 : If we could solve the issue by simplifying DelayedOperation, it
would be ideal. I am not sure how your proposal avoids the above potential
deadlock. Could you provide
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-669330984
@chia7712 : The following is my thought after thinking about this a bit
more. The changes that we made in DelayedJoin is complicated and it still
doesn't completely solve the
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-667233872
@chia7712 : So, it's just a rebase and there is no change in your PR?
This is an automated message from the Apache
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-666753554
@ijuma : I think the proposal is to complete all delayed operations in a
separate thread pool. My concerns for that approach are the following: (1)
Configuration: How many threads
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-664652669
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-663597341
3 system test failures with trunk.
http://confluent-kafka-system-test-results.s3-us-west-2.amazonaws.com/2020-07-23--001.1595551051--apache--trunk--0b181fdde/report.html
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-663109443
@chia7712: Only 6 test failures in the latest run with your PR.
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-662642708
@chia7712 : I just merged #9026. Could you rebase again? I will run the
system tests after that. Thanks.
This is
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-658281873
@chia7712 : Here is the latest system test result.
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-656365594
@chia7712 : All 3 PRs you fixed above have been merged. Do you want to
rebase again so that I can run system tests one more time?
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-653953935
@chia7712 : Thanks. Are the unit test failures also due to flaky tests?
This is an automated message from the
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-653227128
@chia7712 : Thanks for the investigation. You don't need to fix all those
flaky tests. It would be helpful if you could file separate jiras to track
them, if not already.
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-650786185
Latest system test results. Down to 15 failures.
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-650583070
@chia7712 I just merged the PR for KAFKA-10180. Perhaps you can rebase again.
This is an automated message from
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-650240109
For comparison, 29 test failures in trunk
http://testing.confluent.io/confluent-kafka-system-test-results/?prefix=2020-06-25--001.1593132363--apache--trunk--3348fc49d/
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-649615869
Latest test results with 31 failures.
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-648259565
Thanks. Triggering another round of system tests.
This is an automated message from the Apache Git Service.
To
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-648254407
@chia7712 : I think the client compatibility test failures are probably
because you haven't rebased the PR. #8841 was fixed 6 days ago. Could you
rebase your PR?
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-648249808
There were still lots of client compatibility related failures
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-647688705
It seems we just fixed a bunch of client compatibility related failures in
https://github.com/apache/kafka/pull/8841. Another 18 test failures were due to
TLSv1.3 and are tracked
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-647020292
This is a second run of the system tests.
junrao commented on pull request #8657:
URL: https://github.com/apache/kafka/pull/8657#issuecomment-646685855
@chia7712 : The system test results can be found in
31 matches
Mail list logo