[jira] [Updated] (KAFKA-532) Multiple controllers can co-exist during soft failures

["Neha Narkhede (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/KAFKA-532?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Neha Narkhede updated KAFKA-532:
--------------------------------

    Attachment: kafka-532-v1.patch

Introduced a controller generation/epoch that increments after a successful controller election

Changes include -
1. Guard zookeeper writes by the controller with the controller epoch. This includes initializing a leader/isr path, e
lecting leader for a partition and shrinking the isr for a partition
2. Include controller epoch in the state change requests sent to the broker
3. Include logic to discard state change requests with a stale controller epoch on the brokers

Testing
LeaderElectionTest: Added a unit test to send leader/isr request with a stale controller epoch and check if the bro
ker discards the request and sends back the appropriate error code (StaleControllerEpochCode)

                
> Multiple controllers can co-exist during soft failures
> ------------------------------------------------------
>
>                 Key: KAFKA-532
>                 URL: https://issues.apache.org/jira/browse/KAFKA-532
>             Project: Kafka
>          Issue Type: Bug
>    Affects Versions: 0.8
>            Reporter: Neha Narkhede
>            Assignee: Neha Narkhede
>            Priority: Blocker
>              Labels: bugs
>         Attachments: kafka-532-v1.patch
>
>   Original Estimate: 48h
>  Remaining Estimate: 48h
>
> If the current controller experiences an intermittent soft failure (GC pause) in the middle of leader election or partition reassignment, a new controller might get elected and start communicating new state change decisions to the brokers. After recovering from the soft failure, the old controller might continue sending some stale state change decisions to the brokers, resulting in unexpected failures. We need to introduce a controller generation id that increments with controller election. The brokers should reject any state change requests by a controller with an older generation id.

--
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