You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "Kai Tian (Jira)" <ji...@apache.org> on 2021/07/06 02:28:00 UTC

[jira] [Commented] (KAFKA-7260) Failures while attempting to move partitions with kafka-reassign-partitions.sh

    [ https://issues.apache.org/jira/browse/KAFKA-7260?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17375124#comment-17375124 ] 

Kai Tian commented on KAFKA-7260:
---------------------------------

I had a similar problem to yours, but we had a disk failure due to the Kafka cluster, and the bad disk configuration was not removed from the Kafka configuration.Causes the replica on the bad disk to be unable to delete when repartition, when viewing the partition number, some partition replica will be greater than the replication-factor, at most will be twice the replication-factor.Remove the configuration of the bad disk and restart the broker to recover.

> Failures while attempting to move partitions with kafka-reassign-partitions.sh
> ------------------------------------------------------------------------------
>
>                 Key: KAFKA-7260
>                 URL: https://issues.apache.org/jira/browse/KAFKA-7260
>             Project: Kafka
>          Issue Type: Bug
>          Components: controller, tools
>    Affects Versions: 1.0.0
>            Reporter: Kristopher Browne
>            Priority: Major
>
> We're having issues with some clusters where a given broker shows as assigned partitions, but doesn't currently host them:
> {code:java}
> Topic: <redacted>    Partition: 108  Leader: 5       Replicas: 5,3,4 Isr: 5,4{code}
> When I tried to use {{kafka-reassign-partitions.sh}} to push around the partitions showing that issue, it failed on most of the new assignments...
> {code:java}
> ERROR: Assigned replicas (5,2,3) don't match the list of replicas for reassignment (2,5,1) for partition <redacted>-108{code}
> I'm working through that, but when I try to do further manipulations I'm getting errors that some actions are still in progress, despite nothing showing in zookeeper.
> {code:java}
> Reassignment of partition <redacted>-40 is still in progress{code}
>  
> {code:java}
> $ zookeeper-shell.sh localhost:2181 <<< 'ls /admin/reassign_partitions'
> []
> {code}
> I don't see other tickets that relate to these, and do not see similar issues/fixes googling around.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)