[jira] [Updated] (KAFKA-575) Partition.makeFollower() reads broker info from ZK

["Swapnil Ghike (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/KAFKA-575?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Swapnil Ghike updated KAFKA-575:
--------------------------------

    Attachment: kafka-575-v1.patch

1. Live broker info included in LeaderIsrRequest. 

ControllerBrokerRequestBatch.sendrequestsToBrokers() passes the live brokers list to LeaderIsrRequest. 

Thus Partition.makeFollower no longer needs to read the broker info from zookeeper. Only the controller reads this info from zookeeper with this patch.

2. Partition.makeLeaderOrFollower feels unnecessary because it reuses only a couple lines of code, and introduces an extra step in understanding the logic. 

Hence, got rid of makeLeaderOrFollower and copied the lock synchronization and discarding the incoming request part to makeLeader and makeFollower separately.
                
> Partition.makeFollower() reads broker info from ZK
> --------------------------------------------------
>
>                 Key: KAFKA-575
>                 URL: https://issues.apache.org/jira/browse/KAFKA-575
>             Project: Kafka
>          Issue Type: Bug
>          Components: core
>    Affects Versions: 0.8
>            Reporter: Jun Rao
>            Assignee: Swapnil Ghike
>            Priority: Blocker
>              Labels: bugs
>         Attachments: kafka-575-v1.patch
>
>   Original Estimate: 48h
>  Remaining Estimate: 48h
>
> To follow a new leader, Partition.makeFollower() has to obtain the broker info of the new leader. Currently, it reads that info from ZK for every affected partition. This increases the time for a leader to truly available. 

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira