Skip to content
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

feat(metadata): fix data race when close and reopen coordinator partition #1255

Merged
merged 3 commits into from
May 14, 2024
Merged

feat(metadata): fix data race when close and reopen coordinator partition #1255

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
bfb205b
feat(metadata): fix data race when close and reopen coordinator parti…
ShadowySpirits May 14, 2024
0645857
fix(metadata): fix potential data race
ShadowySpirits May 14, 2024
d2558fb
feat(spotbugs): make spotbugs happy
ShadowySpirits May 14, 2024
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
25 changes: 21 additions & 4 deletions core/src/main/scala/kafka/server/ReplicaManager.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,7 +57,7 @@ import org.apache.kafka.common.requests.FetchRequest.PartitionData
import org.apache.kafka.common.requests.ProduceResponse.PartitionResponse
import org.apache.kafka.common.requests._
import org.apache.kafka.common.utils.{ThreadUtils, Time}
import org.apache.kafka.image.{LocalReplicaChanges, MetadataImage, TopicsDelta}
import org.apache.kafka.image.{LocalReplicaChanges, MetadataImage, TopicDelta, TopicsDelta}
import org.apache.kafka.metadata.LeaderConstants.NO_LEADER
import org.apache.kafka.server.common.MetadataVersion._
import org.apache.kafka.server.metrics.s3stream.S3StreamKafkaMetricsManager
Expand Down Expand Up @@ -2682,7 +2682,15 @@ class ReplicaManager(val config: KafkaConfig,
}

def applyDelta(delta: TopicsDelta, newImage: MetadataImage): Unit = {
asyncApplyDelta(delta, newImage).get()
asyncApplyDelta(delta, newImage, (_, _) => {}).get()
}

def getTopicDelta(topicName: String, newImage: MetadataImage, delta: TopicsDelta): Option[TopicDelta] = {
Option(newImage.topics().getTopic(topicName)).flatMap {
topicImage => Option(delta).flatMap {
topicDelta => Option(topicDelta.changedTopic(topicImage.id()))
}
}
}

// AutoMQ for Kafka inject start
Expand All @@ -2692,7 +2700,7 @@ class ReplicaManager(val config: KafkaConfig,
* @param delta The delta to apply.
* @param newImage The new metadata image.
*/
def asyncApplyDelta(delta: TopicsDelta, newImage: MetadataImage): CompletableFuture[Void] = {
def asyncApplyDelta(delta: TopicsDelta, newImage: MetadataImage, callback: (TopicDelta, Int) => Unit): CompletableFuture[Void] = {
// Before taking the lock, compute the local changes
val localChanges = delta.localChanges(config.nodeId)

Expand All @@ -2709,7 +2717,7 @@ class ReplicaManager(val config: KafkaConfig,
def doPartitionDeletion(): Unit = {
stateChangeLogger.info(s"Deleting ${deletes.size} partition(s).")
deletes.forKeyValue((tp, _) => {
val opCf = doPartitionDeletionAsyncLocked(tp)
val opCf = doPartitionDeletionAsyncLocked(tp, delta, newImage, callback)
opCfList.add(opCf)
})
}
Expand Down Expand Up @@ -2738,6 +2746,8 @@ class ReplicaManager(val config: KafkaConfig,
val leader = mutable.Map[TopicPartition, LocalReplicaChanges.PartitionInfo]()
leader += (tp -> info)
applyLocalLeadersDelta(changedPartitions, delta, lazyOffsetCheckpoints, leader)
// Apply the delta before elect leader.
getTopicDelta(tp.topic(), newImage, delta).foreach(callback(_, tp.partition()))
} catch {
case t: Throwable => stateChangeLogger.error(s"Transitioning partition(s) fail: $localChanges", t)
} finally {
Expand Down Expand Up @@ -2788,6 +2798,10 @@ class ReplicaManager(val config: KafkaConfig,
* @return A future which completes when the partition has been deleted.
*/
private def doPartitionDeletionAsyncLocked(tp: TopicPartition): CompletableFuture[Void] = {
doPartitionDeletionAsyncLocked(tp, null, null, (_, _) => {})
}

private def doPartitionDeletionAsyncLocked(tp: TopicPartition, delta: TopicsDelta, newImage: MetadataImage, callback: (TopicDelta, Int) => Unit): CompletableFuture[Void] = {
val prevOp = partitionOpMap.getOrDefault(tp, CompletableFuture.completedFuture(null))
val opCf = new CompletableFuture[Void]()
partitionOpMap.put(tp, opCf)
Expand All @@ -2806,6 +2820,9 @@ class ReplicaManager(val config: KafkaConfig,
s"we got an unexpected ${e.getClass.getName} exception: ${e.getMessage}")
}
}
if (newImage != null && delta != null) {
getTopicDelta(tp.topic(), newImage, delta).foreach(callback(_, tp.partition()))
}
} finally {
opCf.complete(null)
partitionOpMap.remove(tp, opCf)
Expand Down
123 changes: 83 additions & 40 deletions core/src/main/scala/kafka/server/metadata/BrokerMetadataPublisher.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,8 +17,6 @@

package kafka.server.metadata

import java.util.Properties
import java.util.concurrent.atomic.AtomicLong
import kafka.coordinator.group.GroupCoordinator
import kafka.coordinator.transaction.TransactionCoordinator
import kafka.log.{LogManager, UnifiedLog}
Expand All @@ -27,11 +25,13 @@ import kafka.utils.Logging
import org.apache.kafka.common.TopicPartition
import org.apache.kafka.common.internals.Topic
import org.apache.kafka.common.utils.ThreadUtils
import org.apache.kafka.image.{MetadataDelta, MetadataImage, TopicDelta, TopicsDelta, TopicsImage}
import org.apache.kafka.image._
import org.apache.kafka.metadata.authorizer.ClusterMetadataAuthorizer
import org.apache.kafka.server.authorizer.Authorizer
import org.apache.kafka.server.fault.FaultHandler

import java.util.Properties
import java.util.concurrent.atomic.AtomicLong
import scala.collection.mutable


Expand Down Expand Up @@ -233,51 +233,72 @@ class BrokerMetadataPublisher(
}

def handleTopicsDelta(deltaName: String, topicsDelta: TopicsDelta, delta: MetadataDelta, newImage: MetadataImage): Unit = {
// Notify the replica manager about changes to topics.
val cf = replicaManager.asyncApplyDelta(topicsDelta, newImage)
cf.whenCompleteAsync((nil, ex) => {
if (ex != null) {
metadataPublishingFaultHandler.handleFault("Error applying topics " + s"delta in ${deltaName}", ex)
}
try {
// Update the group coordinator of local changes
updateCoordinator(newImage,
delta,
Topic.GROUP_METADATA_TOPIC_NAME,
groupCoordinator.onElection,
groupCoordinator.onResignation)
} catch {
case t: Throwable => metadataPublishingFaultHandler.handleFault("Error updating group " +
s"coordinator with local changes in ${deltaName}", t)
// Callback for each topic delta.
def callback(topicDelta: TopicDelta, partition: Int): Unit = {
if (Topic.GROUP_METADATA_TOPIC_NAME.equals(topicDelta.name())) {
try {
// Handle the case where the group metadata topic was deleted
if (topicsDelta.topicWasDeleted(Topic.GROUP_METADATA_TOPIC_NAME)) {
val partitionRegistration = topicsDelta.image.getTopic(Topic.GROUP_METADATA_TOPIC_NAME).partitions.get(partition)
if (partitionRegistration != null && partitionRegistration.leader == brokerId) {
groupCoordinator.onResignation(partition, Option(partitionRegistration.leaderEpoch))
}
}

// Update the group coordinator of local changes
updateCoordinator(
topicDelta,
partition,
groupCoordinator.onElection,
(partitionIndex, leaderEpochOpt) => groupCoordinator.onResignation(partitionIndex, leaderEpochOpt)
)
} catch {
case t: Throwable => metadataPublishingFaultHandler.handleFault("Error updating group " +
s"coordinator with local changes in $deltaName", t)
}
}
try {
// Update the transaction coordinator of local changes
updateCoordinator(newImage,
delta,
Topic.TRANSACTION_STATE_TOPIC_NAME,
txnCoordinator.onElection,
txnCoordinator.onResignation)
} catch {
case t: Throwable => metadataPublishingFaultHandler.handleFault("Error updating txn " +
s"coordinator with local changes in ${deltaName}", t)

if (Topic.TRANSACTION_STATE_TOPIC_NAME.equals(topicDelta.name())) {
try {
// Handle the case where the transaction state topic was deleted
if (topicsDelta.topicWasDeleted(Topic.TRANSACTION_STATE_TOPIC_NAME)) {
val partitionRegistration = topicsDelta.image.getTopic(Topic.TRANSACTION_STATE_TOPIC_NAME).partitions.get(partition)
if (partitionRegistration != null && partitionRegistration.leader == brokerId) {
groupCoordinator.onResignation(partition, Option(partitionRegistration.leaderEpoch))
}
}

// Update the transaction coordinator of local changes
updateCoordinator(
topicDelta,
partition,
txnCoordinator.onElection,
txnCoordinator.onResignation)
} catch {
case t: Throwable => metadataPublishingFaultHandler.handleFault("Error updating txn " +
s"coordinator with local changes in $deltaName", t)
}
}

try {
// Notify the group coordinator about deleted topics.
val deletedTopicPartitions = new mutable.ArrayBuffer[TopicPartition]()
topicsDelta.deletedTopicIds().forEach { id =>
val topicImage = topicsDelta.image().getTopic(id)
topicImage.partitions().keySet().forEach {
id => deletedTopicPartitions += new TopicPartition(topicImage.name(), id)
}
}
if (deletedTopicPartitions.nonEmpty) {
groupCoordinator.handleDeletedPartitions(deletedTopicPartitions, RequestLocal.NoCaching)
if (topicsDelta.deletedTopicIds().contains(topicDelta.id())) {
groupCoordinator.handleDeletedPartitions(
mutable.Seq.apply(new TopicPartition(topicDelta.name(), partition)), RequestLocal.NoCaching)
}
} catch {
case t: Throwable => metadataPublishingFaultHandler.handleFault("Error updating group " +
s"coordinator with deleted partitions in ${deltaName}", t)
s"coordinator with deleted partitions in $deltaName", t)
}
}

// Notify the replica manager about changes to topics.
val cf = replicaManager.asyncApplyDelta(topicsDelta, newImage, callback)
cf.whenComplete((_, ex) => {
if (ex != null) {
metadataPublishingFaultHandler.handleFault("Error applying topics " + s"delta in ${deltaName}", ex)
}
}, partitionOpCallbackExecutor)
})
}

override def publishedOffset: Long = publishedOffsetAtomic.get()
Expand Down Expand Up @@ -331,6 +352,28 @@ class BrokerMetadataPublisher(
}
}

// AutoMQ inject start
def updateCoordinator(
topicDelta: TopicDelta,
partition: Int,
election: (Int, Int) => Unit,
resignation: (Int, Option[Int]) => Unit
): Unit = {
val topicPartition = new TopicPartition(topicDelta.name(), partition)
val changes = topicDelta.localChanges(brokerId)

if (changes.deletes.contains(topicPartition)) {
resignation(partition, None)
}
Option(changes.leaders.get(topicPartition)).foreach { partitionInfo =>
election(partition, partitionInfo.partition.leaderEpoch)
}
Option(changes.followers().get(topicPartition)).foreach { partitionInfo =>
resignation(partition, Some(partitionInfo.partition.leaderEpoch))
}
}
// AutoMQ inject end

private def initializeManagers(): Unit = {
try {
// Start log manager, which will perform (potentially lengthy)
Expand Down
Loading