You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "Jiangtao Liu (Jira)" <ji...@apache.org> on 2020/04/07 21:43:00 UTC
[jira] [Updated] (KAFKA-8270) Kafka timestamp-based retention
policy is not working when Kafka client has out of sync system clock issue
[ https://issues.apache.org/jira/browse/KAFKA-8270?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jiangtao Liu updated KAFKA-8270:
--------------------------------
Summary: Kafka timestamp-based retention policy is not working when Kafka client has out of sync system clock issue (was: Kafka retention hour is not working)
> Kafka timestamp-based retention policy is not working when Kafka client has out of sync system clock issue
> ----------------------------------------------------------------------------------------------------------
>
> Key: KAFKA-8270
> URL: https://issues.apache.org/jira/browse/KAFKA-8270
> Project: Kafka
> Issue Type: Bug
> Components: log
> Reporter: Jiangtao Liu
> Priority: Major
> Labels: storage
> Attachments: Screen Shot 2019-04-20 at 10.57.59 PM.png
>
>
> Kafka l cannot be deleted after the configured retention hours (12 hours for log retention).
> What's our Kafka cluster look like?
> There are 6 brokers deployed with Kafka version 1.1.1.
>
> Is it reproducible?
> I am not sure since our Kafka cluster is working well over 1.5 years without retention issue until 4/13/2019 ~ 4/20/2019.
>
> is it related to the active segment?
> as I know Kafka will not delete an active segment, my case those old logs are not activated, they should be inactivated.
>
> What's the current status?
> Those old logs have been deleted after I manually ran rolling restart Kafka servers with retention hours adjustment (Ideally I tried this solution aimed to{color:#333333} force retention hours work, not really want to adjust the retention hours, finally the solution it's working, but not immediately, I remember the retention start work after couples of hours after applying the change and rolling restart Kafka servers{color}.), now our Kafka storage is back to normal, please check the screenshot attached with this ticket.
> A sample old log added here for better understanding of the retention hours not working issue.
> ```
> // it has been there from 4/12
> -rw-r--r-- 1 root root 136866627{color:#d04437} Apr 12{color} 04:33 00000000002581377820.log
> // It was still being opened by Kafka when I check it with the tool lsof on {color:#d04437}4/19/2019 before server rolling restart with retention hours adjustment{color}.
> java 20281 0 1678u REG 202,32 136866627 1074562295 /kafka/data/.../00000000002581377820.log
> ```
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)