[jira] [Commented] (KAFKA-15105) Flaky test FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable
[
https://issues.apache.org/jira/browse/KAFKA-15105?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17738081#comment-17738081
]
Josep Prat commented on KAFKA-15105:
Hi [~riedelmax] ,
Only maintainers + a subgroup of collaborators can rerun builds in CI, but even
for them, they can just run them as they are (no more detailed output). And
sorry, I just realized I copy pasted the wrong output. This is the right one:
[https://ci-builds.apache.org/job/Kafka/job/kafka-pr/job/PR-13865/2/testReport/junit/integration.kafka.server/FetchFromFollowerIntegrationTest/Build___JDK_11_and_Scala_2_13___testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable___2/]
I'm updating the description of the Jira issue.+
> Flaky test
> FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable
> -
>
> Key: KAFKA-15105
> URL: https://issues.apache.org/jira/browse/KAFKA-15105
> Project: Kafka
> Issue Type: Bug
> Components: core
>Affects Versions: 3.5.0
>Reporter: Josep Prat
>Assignee: Max Riedel
>Priority: Major
> Labels: flaky-test
>
> Test
> integration.kafka.server.FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable()
> became flaky. An example can be found here:
> [https://ci-builds.apache.org/job/Kafka/job/kafka-pr/job/PR-13865/2/testReport/junit/integration.kafka.server/FetchFromFollowerIntegrationTest/Build___JDK_11_and_Scala_2_13___testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable__/]
> The error might be caused because of a previous kafka cluster used for
> another test wasn't cleaned up properly before this one run.
>
> h3. Error Message
> {code:java}
> org.apache.kafka.common.errors.TopicExistsException: Topic
> '__consumer_offsets' already exists.{code}
> h3. Stacktrace
> {code:java}
> org.apache.kafka.common.errors.TopicExistsException: Topic
> '__consumer_offsets' already exists. {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
[jira] [Commented] (KAFKA-15105) Flaky test FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable
[
https://issues.apache.org/jira/browse/KAFKA-15105?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17738077#comment-17738077
]
Max Riedel commented on KAFKA-15105:
Hi [~josep.prat]
Thanks for giving me the necessary Jira rights. I was able to assign the ticket
to me now.
So far, all test runs I did on my local environment passed. But I will try the
option to run until failure and see what I can learn from that.
My question was about the CI. Is it correct that we are not able to rerun
builds there or get more detailed output? I couldn't find the error message
"Topic '__consumer_offsets' already exists." in the standard output of the test
run
> Flaky test
> FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable
> -
>
> Key: KAFKA-15105
> URL: https://issues.apache.org/jira/browse/KAFKA-15105
> Project: Kafka
> Issue Type: Bug
> Components: core
>Affects Versions: 3.5.0
>Reporter: Josep Prat
>Assignee: Max Riedel
>Priority: Major
> Labels: flaky-test
>
> Test
> integration.kafka.server.FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable()
> became flaky. An example can be found here:
> [https://ci-builds.apache.org/job/Kafka/job/kafka-pr/job/PR-13865/2/testReport/junit/integration.kafka.server/FetchFromFollowerIntegrationTest/Build___JDK_11_and_Scala_2_13___testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable__/]
> The error might be caused because of a previous kafka cluster used for
> another test wasn't cleaned up properly before this one run.
>
> h3. Error Message
> {code:java}
> org.apache.kafka.common.errors.TopicExistsException: Topic
> '__consumer_offsets' already exists.{code}
> h3. Stacktrace
> {code:java}
> org.apache.kafka.common.errors.TopicExistsException: Topic
> '__consumer_offsets' already exists. {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
[jira] [Commented] (KAFKA-15105) Flaky test FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable
[
https://issues.apache.org/jira/browse/KAFKA-15105?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17737696#comment-17737696
]
Josep Prat commented on KAFKA-15105:
Hi [~riedelmax], feel free to assign this issue to yourself :)
{quote}I'm still trying to understand how the build infrastructure works. Can
someone give me a hint, how to reproduce the behavior?
{quote}
In this rely part of the problem, many times these issues are not easily
reproducible on your machine. Usually they occur on CI.
The test itself is under the core module, so one way to try to reproduce it
would be running `./gradlew core:test` or `./gradlew core:test --tests
integration.kafka.server.FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable`repeatedly
until failure. IntelliJ (if you have it and use it) has a similar feature
where one can run a test until failure.
Is this answering your question? Or you wanted to know some other specific
details?
> Flaky test
> FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable
> -
>
> Key: KAFKA-15105
> URL: https://issues.apache.org/jira/browse/KAFKA-15105
> Project: Kafka
> Issue Type: Bug
> Components: core
>Affects Versions: 3.5.0
>Reporter: Josep Prat
>Priority: Major
> Labels: flaky-test
>
> Test
> integration.kafka.server.FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable()
> became flaky. An example can be found here:
> [https://ci-builds.apache.org/job/Kafka/job/kafka-pr/job/PR-13865/2/testReport/junit/integration.kafka.server/FetchFromFollowerIntegrationTest/Build___JDK_11_and_Scala_2_13___testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable__/]
> The error might be caused because of a previous kafka cluster used for
> another test wasn't cleaned up properly before this one run.
>
> h3. Error Message
> {code:java}
> org.apache.kafka.common.errors.TopicExistsException: Topic
> '__consumer_offsets' already exists.{code}
> h3. Stacktrace
> {code:java}
> org.apache.kafka.common.errors.TopicExistsException: Topic
> '__consumer_offsets' already exists. {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
[jira] [Commented] (KAFKA-15105) Flaky test FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable
[
https://issues.apache.org/jira/browse/KAFKA-15105?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17737186#comment-17737186
]
Max Riedel commented on KAFKA-15105:
I would like to work on this issue. I'm still trying to understand how the
build infrastructure works. Can someone give me a hint, how to reproduce the
behavior?
> Flaky test
> FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable
> -
>
> Key: KAFKA-15105
> URL: https://issues.apache.org/jira/browse/KAFKA-15105
> Project: Kafka
> Issue Type: Bug
> Components: core
>Affects Versions: 3.5.0
>Reporter: Josep Prat
>Priority: Major
> Labels: flaky-test
>
> Test
> integration.kafka.server.FetchFromFollowerIntegrationTest.testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable()
> became flaky. An example can be found here:
> [https://ci-builds.apache.org/job/Kafka/job/kafka-pr/job/PR-13865/2/testReport/junit/integration.kafka.server/FetchFromFollowerIntegrationTest/Build___JDK_11_and_Scala_2_13___testFetchFromLeaderWhilePreferredReadReplicaIsUnavailable__/]
> The error might be caused because of a previous kafka cluster used for
> another test wasn't cleaned up properly before this one run.
>
> h3. Error Message
> {code:java}
> org.apache.kafka.common.errors.TopicExistsException: Topic
> '__consumer_offsets' already exists.{code}
> h3. Stacktrace
> {code:java}
> org.apache.kafka.common.errors.TopicExistsException: Topic
> '__consumer_offsets' already exists. {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
