Skip to content

KAFKA-9866: Avoid election for topics where preferred leader is not in ISR #8524

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
junrao merged 6 commits into from
Apr 27, 2020
Merged

KAFKA-9866: Avoid election for topics where preferred leader is not in ISR #8524

Show file tree
Hide file tree
Changes from 5 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
44eea04
Avoid starting election for topics where preferred leader is not in s…
leonardge Apr 21, 2020
29632f7
Refactored the code and fixed bug in the integration test.
leonardge Apr 22, 2020
172960f
Refactored according to comments.
leonardge Apr 23, 2020
24c6d42
Kept the code to be consistent.
leonardge Apr 24, 2020
dc67b99
Reverted style change.
leonardge Apr 24, 2020
2fa3cee
Renamed helper function after review.
leonardge Apr 25, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 12 additions & 1 deletion core/src/main/scala/kafka/controller/KafkaController.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1068,12 +1068,23 @@ class KafkaController(val config: KafkaConfig,
val candidatePartitions = topicsNotInPreferredReplica.keys.filter(tp => controllerContext.isReplicaOnline(leaderBroker, tp) &&
controllerContext.partitionsBeingReassigned.isEmpty &&
!topicDeletionManager.isTopicQueuedUpForDeletion(tp.topic) &&
controllerContext.allTopics.contains(tp.topic))
controllerContext.allTopics.contains(tp.topic) &&
isPreferredLeaderInSync(tp)
)
onReplicaElection(candidatePartitions.toSet, ElectionType.PREFERRED, AutoTriggered)
}
}
}

private def isPreferredLeaderInSync(tp: TopicPartition): Boolean = {

@junrao junrao Apr 24, 2020

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Perhaps a more accurate name is canPreferredReplicaBeLeader()?

@leonardge leonardge Apr 25, 2020

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done!

val assignment = controllerContext.partitionReplicaAssignment(tp)
val liveReplicas = assignment.filter(replica => controllerContext.isReplicaOnline(replica, tp))

@ijuma ijuma Apr 27, 2020

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We also do controllerContext.isReplicaOnline(leaderBroker, tp) in the caller. Do we also need it here?

@junrao junrao Apr 27, 2020

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good point. We can get rid of the isReplicaOnline() check in the caller. @leonardge : Could you summit a followup minor PR?

@leonardge leonardge Apr 27, 2020

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sure thing!

val isr = controllerContext.partitionLeadershipInfo(tp).leaderAndIsr.isr
PartitionLeaderElectionAlgorithms
.preferredReplicaPartitionLeaderElection(assignment, isr, liveReplicas.toSet)
.nonEmpty
}

private def processAutoPreferredReplicaLeaderElection(): Unit = {
if (!isActive) return
try {
Expand Down
4 changes: 2 additions & 2 deletions core/src/test/scala/unit/kafka/controller/ControllerIntegrationTest.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -433,8 +433,8 @@ class ControllerIntegrationTest extends ZooKeeperTestHarness {
TestUtils.createTopic(zkClient, tp.topic, partitionReplicaAssignment = assignment, servers = servers)
waitForPartitionState(tp, firstControllerEpoch, otherBrokerId, LeaderAndIsr.initialLeaderEpoch,
"failed to get expected partition state upon topic creation")
servers(1).shutdown()
servers(1).awaitShutdown()
servers(otherBrokerId).shutdown()
servers(otherBrokerId).awaitShutdown()
TestUtils.waitUntilTrue(() => {
val leaderIsrAndControllerEpochMap = zkClient.getTopicPartitionStates(Seq(tp))
leaderIsrAndControllerEpochMap.contains(tp) &&
Expand Down