[jira] [Commented] (KAFKA-10362) When resuming Streams active task with EOS, the checkpoint file should be deleted
[ https://issues.apache.org/jira/browse/KAFKA-10362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17190178#comment-17190178 ] Matthias J. Sax commented on KAFKA-10362: - Thanks [~DOJI] and [~ipasynkov]! We should make sure that tickets are assigned properly to avoid such an overlap in the future. > When resuming Streams active task with EOS, the checkpoint file should be > deleted > - > > Key: KAFKA-10362 > URL: https://issues.apache.org/jira/browse/KAFKA-10362 > Project: Kafka > Issue Type: Bug > Components: streams >Affects Versions: 2.6.0 >Reporter: Guozhang Wang >Assignee: Sharath Bhat >Priority: Major > Labels: newbie++ > > Today when we suspend a task we commit and along with the commit we always > write checkpoint file even if we are eosEnabled (since the state is already > SUSPENDED). But the suspended task may later be resumed and in that case the > checkpoint file should be deleted since it should only be written when it is > cleanly closed. > With our latest rebalance protocol in KIP-429, resume would not be called > since all suspended tasks would be closed, but with the old eager protocol it > may still be called — I think that may be the reason we did not get it often. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10362) When resuming Streams active task with EOS, the checkpoint file should be deleted
[ https://issues.apache.org/jira/browse/KAFKA-10362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17189879#comment-17189879 ] Sharath Bhat commented on KAFKA-10362: -- Thank you [~ipasynkov] > When resuming Streams active task with EOS, the checkpoint file should be > deleted > - > > Key: KAFKA-10362 > URL: https://issues.apache.org/jira/browse/KAFKA-10362 > Project: Kafka > Issue Type: Bug > Components: streams >Affects Versions: 2.6.0 >Reporter: Guozhang Wang >Priority: Major > Labels: newbie++ > > Today when we suspend a task we commit and along with the commit we always > write checkpoint file even if we are eosEnabled (since the state is already > SUSPENDED). But the suspended task may later be resumed and in that case the > checkpoint file should be deleted since it should only be written when it is > cleanly closed. > With our latest rebalance protocol in KIP-429, resume would not be called > since all suspended tasks would be closed, but with the old eager protocol it > may still be called — I think that may be the reason we did not get it often. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10362) When resuming Streams active task with EOS, the checkpoint file should be deleted
[ https://issues.apache.org/jira/browse/KAFKA-10362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17189860#comment-17189860 ] Ilia Pasynkov commented on KAFKA-10362: --- Hello [~DOJI]. You can continue with this ticket since you've already made some progress and have a PR. I'll switch to another task then > When resuming Streams active task with EOS, the checkpoint file should be > deleted > - > > Key: KAFKA-10362 > URL: https://issues.apache.org/jira/browse/KAFKA-10362 > Project: Kafka > Issue Type: Bug > Components: streams >Affects Versions: 2.6.0 >Reporter: Guozhang Wang >Priority: Major > Labels: newbie++ > > Today when we suspend a task we commit and along with the commit we always > write checkpoint file even if we are eosEnabled (since the state is already > SUSPENDED). But the suspended task may later be resumed and in that case the > checkpoint file should be deleted since it should only be written when it is > cleanly closed. > With our latest rebalance protocol in KIP-429, resume would not be called > since all suspended tasks would be closed, but with the old eager protocol it > may still be called — I think that may be the reason we did not get it often. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10362) When resuming Streams active task with EOS, the checkpoint file should be deleted
[ https://issues.apache.org/jira/browse/KAFKA-10362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17189831#comment-17189831 ] Sharath Bhat commented on KAFKA-10362: -- Hi [~ipasynkov], Can I continue work on this ticket ? I am newbie to kafka community, so I thought this would be a place to start and have raised the PR. Please let me know your thoughts. > When resuming Streams active task with EOS, the checkpoint file should be > deleted > - > > Key: KAFKA-10362 > URL: https://issues.apache.org/jira/browse/KAFKA-10362 > Project: Kafka > Issue Type: Bug > Components: streams >Affects Versions: 2.6.0 >Reporter: Guozhang Wang >Priority: Major > Labels: newbie++ > > Today when we suspend a task we commit and along with the commit we always > write checkpoint file even if we are eosEnabled (since the state is already > SUSPENDED). But the suspended task may later be resumed and in that case the > checkpoint file should be deleted since it should only be written when it is > cleanly closed. > With our latest rebalance protocol in KIP-429, resume would not be called > since all suspended tasks would be closed, but with the old eager protocol it > may still be called — I think that may be the reason we did not get it often. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10362) When resuming Streams active task with EOS, the checkpoint file should be deleted
[ https://issues.apache.org/jira/browse/KAFKA-10362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17189711#comment-17189711 ] Matthias J. Sax commented on KAFKA-10362: - [~DOJI], even if the ticket was not assigned yet, it seems that [~ipasynkov] was actually working on it... Not sure how to proceed now. [~ipasynkov] do you already have a PR? I leave it up to both of you to figure it out you takes this ticket. > When resuming Streams active task with EOS, the checkpoint file should be > deleted > - > > Key: KAFKA-10362 > URL: https://issues.apache.org/jira/browse/KAFKA-10362 > Project: Kafka > Issue Type: Bug > Components: streams >Affects Versions: 2.6.0 >Reporter: Guozhang Wang >Priority: Major > Labels: newbie++ > > Today when we suspend a task we commit and along with the commit we always > write checkpoint file even if we are eosEnabled (since the state is already > SUSPENDED). But the suspended task may later be resumed and in that case the > checkpoint file should be deleted since it should only be written when it is > cleanly closed. > With our latest rebalance protocol in KIP-429, resume would not be called > since all suspended tasks would be closed, but with the old eager protocol it > may still be called — I think that may be the reason we did not get it often. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10362) When resuming Streams active task with EOS, the checkpoint file should be deleted
[ https://issues.apache.org/jira/browse/KAFKA-10362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17189629#comment-17189629 ] Sharath Bhat commented on KAFKA-10362: -- Hi [~guozhang], I worked on the bug and have raised a [PR|https://github.com/apache/kafka/pull/9247] could you please review the PR. and can I get this ticket on my name? > When resuming Streams active task with EOS, the checkpoint file should be > deleted > - > > Key: KAFKA-10362 > URL: https://issues.apache.org/jira/browse/KAFKA-10362 > Project: Kafka > Issue Type: Bug > Components: streams >Affects Versions: 2.6.0 >Reporter: Guozhang Wang >Priority: Major > Labels: newbie++ > > Today when we suspend a task we commit and along with the commit we always > write checkpoint file even if we are eosEnabled (since the state is already > SUSPENDED). But the suspended task may later be resumed and in that case the > checkpoint file should be deleted since it should only be written when it is > cleanly closed. > With our latest rebalance protocol in KIP-429, resume would not be called > since all suspended tasks would be closed, but with the old eager protocol it > may still be called — I think that may be the reason we did not get it often. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10362) When resuming Streams active task with EOS, the checkpoint file should be deleted
[ https://issues.apache.org/jira/browse/KAFKA-10362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17183802#comment-17183802 ] Ilia Pasynkov commented on KAFKA-10362: --- [~guozhang] Hello. I've read related TaskManager and StreamTask. Please correct me if I'm wrong, It seems that I have to make a call to OffsetCheckpoint's delete() method in StreamTask's resume() method [when task's state is SUSPENDED] > When resuming Streams active task with EOS, the checkpoint file should be > deleted > - > > Key: KAFKA-10362 > URL: https://issues.apache.org/jira/browse/KAFKA-10362 > Project: Kafka > Issue Type: Bug > Components: streams >Affects Versions: 2.6.0 >Reporter: Guozhang Wang >Priority: Major > Labels: newbie++ > > Today when we suspend a task we commit and along with the commit we always > write checkpoint file even if we are eosEnabled (since the state is already > SUSPENDED). But the suspended task may later be resumed and in that case the > checkpoint file should be deleted since it should only be written when it is > cleanly closed. > With our latest rebalance protocol in KIP-429, resume would not be called > since all suspended tasks would be closed, but with the old eager protocol it > may still be called — I think that may be the reason we did not get it often. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10362) When resuming Streams active task with EOS, the checkpoint file should be deleted
[ https://issues.apache.org/jira/browse/KAFKA-10362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17183104#comment-17183104 ] Ilia Pasynkov commented on KAFKA-10362: --- [~high.lee] Hello, yes I'm working on this task) > When resuming Streams active task with EOS, the checkpoint file should be > deleted > - > > Key: KAFKA-10362 > URL: https://issues.apache.org/jira/browse/KAFKA-10362 > Project: Kafka > Issue Type: Bug > Components: streams >Affects Versions: 2.6.0 >Reporter: Guozhang Wang >Priority: Major > Labels: newbie++ > > Today when we suspend a task we commit and along with the commit we always > write checkpoint file even if we are eosEnabled (since the state is already > SUSPENDED). But the suspended task may later be resumed and in that case the > checkpoint file should be deleted since it should only be written when it is > cleanly closed. > With our latest rebalance protocol in KIP-429, resume would not be called > since all suspended tasks would be closed, but with the old eager protocol it > may still be called — I think that may be the reason we did not get it often. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10362) When resuming Streams active task with EOS, the checkpoint file should be deleted
[ https://issues.apache.org/jira/browse/KAFKA-10362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17183075#comment-17183075 ] highluck commented on KAFKA-10362: -- [~ipasynkov] Are you working on it? If not, can I do PR? > When resuming Streams active task with EOS, the checkpoint file should be > deleted > - > > Key: KAFKA-10362 > URL: https://issues.apache.org/jira/browse/KAFKA-10362 > Project: Kafka > Issue Type: Bug > Components: streams >Affects Versions: 2.6.0 >Reporter: Guozhang Wang >Priority: Major > Labels: newbie++ > > Today when we suspend a task we commit and along with the commit we always > write checkpoint file even if we are eosEnabled (since the state is already > SUSPENDED). But the suspended task may later be resumed and in that case the > checkpoint file should be deleted since it should only be written when it is > cleanly closed. > With our latest rebalance protocol in KIP-429, resume would not be called > since all suspended tasks would be closed, but with the old eager protocol it > may still be called — I think that may be the reason we did not get it often. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10362) When resuming Streams active task with EOS, the checkpoint file should be deleted
[ https://issues.apache.org/jira/browse/KAFKA-10362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17181498#comment-17181498 ] Guozhang Wang commented on KAFKA-10362: --- Hi [~ipasynkov], sure! If you have read through the related code of Kafka Streams around TaskManager and StreamTask, then feel free to start preparing a PR and ping me whenever it is ready for reviews. > When resuming Streams active task with EOS, the checkpoint file should be > deleted > - > > Key: KAFKA-10362 > URL: https://issues.apache.org/jira/browse/KAFKA-10362 > Project: Kafka > Issue Type: Bug > Components: streams >Affects Versions: 2.6.0 >Reporter: Guozhang Wang >Priority: Major > Labels: newbie++ > > Today when we suspend a task we commit and along with the commit we always > write checkpoint file even if we are eosEnabled (since the state is already > SUSPENDED). But the suspended task may later be resumed and in that case the > checkpoint file should be deleted since it should only be written when it is > cleanly closed. > With our latest rebalance protocol in KIP-429, resume would not be called > since all suspended tasks would be closed, but with the old eager protocol it > may still be called — I think that may be the reason we did not get it often. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Commented] (KAFKA-10362) When resuming Streams active task with EOS, the checkpoint file should be deleted
[ https://issues.apache.org/jira/browse/KAFKA-10362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17181231#comment-17181231 ] Ilia Pasynkov commented on KAFKA-10362: --- Hello, can I pick this task? > When resuming Streams active task with EOS, the checkpoint file should be > deleted > - > > Key: KAFKA-10362 > URL: https://issues.apache.org/jira/browse/KAFKA-10362 > Project: Kafka > Issue Type: Bug > Components: streams >Affects Versions: 2.6.0 >Reporter: Guozhang Wang >Priority: Major > Labels: newbie++ > > Today when we suspend a task we commit and along with the commit we always > write checkpoint file even if we are eosEnabled (since the state is already > SUSPENDED). But the suspended task may later be resumed and in that case the > checkpoint file should be deleted since it should only be written when it is > cleanly closed. > With our latest rebalance protocol in KIP-429, resume would not be called > since all suspended tasks would be closed, but with the old eager protocol it > may still be called — I think that may be the reason we did not get it often. -- This message was sent by Atlassian Jira (v8.3.4#803005)
