You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@kafka.apache.org by "Joel Koshy (JIRA)" <ji...@apache.org> on 2014/08/22 22:00:14 UTC

[jira] [Updated] (KAFKA-1609) New producer metadata response handling should only exclude a PartitionInfo when its error is LEADER_NOT_AVAILABLE

     [ https://issues.apache.org/jira/browse/KAFKA-1609?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Joel Koshy updated KAFKA-1609:
------------------------------

    Summary: New producer metadata response handling should only exclude a PartitionInfo when its error is LEADER_NOT_AVAILABLE  (was: Metadata response should only exclude a PartitionInfo when its error is LEADER_NOT_AVAILABLE)

> New producer metadata response handling should only exclude a PartitionInfo when its error is LEADER_NOT_AVAILABLE
> ------------------------------------------------------------------------------------------------------------------
>
>                 Key: KAFKA-1609
>                 URL: https://issues.apache.org/jira/browse/KAFKA-1609
>             Project: Kafka
>          Issue Type: Bug
>            Reporter: Dong Lin
>            Assignee: Dong Lin
>         Attachments: KAFKA-1609.patch
>
>
> The new producer is not able to produce data when it sees ReplicaNotAvailable error, even when the leader of the replica is still available. 
> This behavior is different from old producer, which says "any other error code (e.g. ReplicaNotAvailable) can be ignored since the producer does not need to access the replica and isr metadata".
> To reproduce the error:
> 1) Start 4 brokers
> 2) Create a topic of 1 partition using a replication factor of 3.
> 3) Start producerPerformance with new producer
> 4) Kill the leader of the topic/partition
> 5) Kill the new leader of the topic/partition
> 6) Observe that the producerPerformance stops producing data.



--
This message was sent by Atlassian JIRA
(v6.2#6252)