[
https://issues.apache.org/jira/browse/KAFKA-3761?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15988745#comment-15988745
]
ASF GitHub Bot commented on KAFKA-3761:
---------------------------------------
GitHub user ijuma opened a pull request:
https://github.com/apache/kafka/pull/2935
MINOR: onControllerResignation should be invoked if triggerControllerMove
is called
This fixes a transient test failure due to a NPE in
ControllerFailoverTest.testMetadataUpdate:
```text
Caused by: java.lang.NullPointerException
at
kafka.controller.ControllerBrokerRequestBatch.addUpdateMetadataRequestForBrokers(ControllerChannelManager.scala:338)
at
kafka.controller.KafkaController.sendUpdateMetadataRequest(KafkaController.scala:975)
at
kafka.controller.ControllerFailoverTest.testMetadataUpdate(ControllerFailoverTest.scala:141)
```
The underlying issue is that setting `activeControllerId.set(-1)` in
`triggerControllerMove`
causes `Reelect` not to invoke `onControllerResignation`. Among other
things, this
causes an IllegalStateException to be thrown when `KafkaScheduler.startup`
is invoked
for the second time without the corresponding `shutdown`.
I also updated the test so that we can trigger this issue deterministically
instead of
transiently.
Finally, I included a few clean-ups:
1. No longer update the broker state in `onControllerFailover`. This is no
longer needed
since we removed the `RunningAsController` state (KAFKA-3761).
2. Trivial clean-ups in KafkaController
3. Removed unused parameter in `ZkUtils.getPartitionLeaderAndIsrForTopics`
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/ijuma/kafka
on-controller-resignation-if-trigger-controller-move
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/kafka/pull/2935.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #2935
----
commit c01d29b3a95e7ddffb91550397a5505d9711d5c8
Author: Ismael Juma <ism...@juma.me.uk>
Date: 2017-04-28T12:30:03Z
MINOR: onControllerResignation should be invoked if triggerControllerMove
is called
commit f28de697f7d109893537652e8b8216c4d06677a7
Author: Ismael Juma <ism...@juma.me.uk>
Date: 2017-04-28T12:30:47Z
Remove remnant broker state update in `onControllerFailover`
commit 898b88b59cffbfdb7df864d0b070ed7a4960601e
Author: Ismael Juma <ism...@juma.me.uk>
Date: 2017-04-28T12:31:15Z
A few trivial clean-ups in KafkaController
commit 241b9890ab47b4670e61f3f9d3b51c6aa92a8a94
Author: Ismael Juma <ism...@juma.me.uk>
Date: 2017-04-28T12:31:38Z
Remove unused parameter in `getPartitionLeaderAndIsrForTopics`
----
> Controller has RunningAsBroker instead of RunningAsController state
> -------------------------------------------------------------------
>
> Key: KAFKA-3761
> URL: https://issues.apache.org/jira/browse/KAFKA-3761
> Project: Kafka
> Issue Type: Bug
> Reporter: Ismael Juma
> Assignee: Roger Hoover
> Fix For: 0.10.1.0
>
>
> In `KafkaServer.start`, we start `KafkaController`:
> {code}
> /* start kafka controller */
> kafkaController = new KafkaController(config, zkUtils, brokerState,
> kafkaMetricsTime, metrics, threadNamePrefix)
> kafkaController.startup()
> {code}
> Which sets the state to `RunningAsController` in
> `KafkaController.onControllerFailover`:
> `brokerState.newState(RunningAsController)`
> And this later gets set to `RunningAsBroker`.
> This doesn't match the diagram in `BrokerStates`. [~junrao] suggested that we
> should start the controller after we register the broker in ZK, but this
> seems tricky as we need to controller in `KafkaApis`.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
