[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16739909#comment-16739909 ] ASF GitHub Bot commented on KAFKA-7040: --- apovzner commented on pull request #6122: KAFKA-7040:

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16738587#comment-16738587 ] Lucas Wang commented on KAFKA-7040: --- Hi [~apovzner], thanks for the info. Please go ahead and take over

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16738578#comment-16738578 ] Anna Povzner commented on KAFKA-7040: - Hi [~luwang], we found a related (with the same root cause)

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16590634#comment-16590634 ] Lucas Wang commented on KAFKA-7040: --- [~lindong] Sure, I will own this ticket and work on the fix. >

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16590545#comment-16590545 ] Dong Lin commented on KAFKA-7040: - [~luwang] It looks like we need to fix this in Kafka to further reduce

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16589401#comment-16589401 ] Anna Povzner commented on KAFKA-7040: - Hi Lucas, You are right. Looks like this may cause message

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16520899#comment-16520899 ] Lucas Wang commented on KAFKA-7040: --- Hi [~apovzner] Sorry for the delayed response. Regarding Case 1

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16514458#comment-16514458 ] Anna Povzner commented on KAFKA-7040: - The offset should not be lost if min.insync.replicas > 1 and

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16514401#comment-16514401 ] Anna Povzner commented on KAFKA-7040: - Hi [~luwang], Did you actually see data loss (or log

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16508907#comment-16508907 ] Lucas Wang commented on KAFKA-7040: --- To be more specific, I think the following sequence of events may

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16508867#comment-16508867 ] Lucas Wang commented on KAFKA-7040: --- [~lindong] Thanks for pointing out the truncation may happen below

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16508847#comment-16508847 ] Dong Lin commented on KAFKA-7040: - [~luwang] Nice catch. It is indeed problematic (with some probability

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16508771#comment-16508771 ] Ismael Juma commented on KAFKA-7040: cc [~apovzner] > The replica fetcher thread may truncate

[jira] [Commented] (KAFKA-7040) The replica fetcher thread may truncate accepted messages during multiple fast leadership transitions

[ https://issues.apache.org/jira/browse/KAFKA-7040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16508763#comment-16508763 ] Dong Lin commented on KAFKA-7040: - [~luwang] Say message m0 is truncated in step 5. Is that message acked