cmccabe commented on a change in pull request #10070: URL: https://github.com/apache/kafka/pull/10070#discussion_r579463797
########## File path: metadata/src/main/java/org/apache/kafka/controller/ReplicationControlManager.java ########## @@ -0,0 +1,894 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.kafka.controller; + +import org.apache.kafka.clients.admin.AlterConfigOp.OpType; +import org.apache.kafka.common.ElectionType; +import org.apache.kafka.common.Uuid; +import org.apache.kafka.common.config.ConfigResource; +import org.apache.kafka.common.errors.BrokerIdNotRegisteredException; +import org.apache.kafka.common.errors.InvalidReplicationFactorException; +import org.apache.kafka.common.errors.InvalidRequestException; +import org.apache.kafka.common.errors.InvalidTopicException; +import org.apache.kafka.common.internals.Topic; +import org.apache.kafka.common.message.AlterIsrRequestData; +import org.apache.kafka.common.message.AlterIsrResponseData; +import org.apache.kafka.common.message.BrokerHeartbeatRequestData; +import org.apache.kafka.common.message.CreateTopicsRequestData.CreatableReplicaAssignment; +import org.apache.kafka.common.message.CreateTopicsRequestData.CreatableTopic; +import org.apache.kafka.common.message.CreateTopicsRequestData.CreatableTopicCollection; +import org.apache.kafka.common.message.CreateTopicsRequestData; +import org.apache.kafka.common.message.CreateTopicsResponseData.CreatableTopicResult; +import org.apache.kafka.common.message.CreateTopicsResponseData; +import org.apache.kafka.common.message.ElectLeadersRequestData.TopicPartitions; +import org.apache.kafka.common.message.ElectLeadersRequestData; +import org.apache.kafka.common.message.ElectLeadersResponseData.PartitionResult; +import org.apache.kafka.common.message.ElectLeadersResponseData.ReplicaElectionResult; +import org.apache.kafka.common.message.ElectLeadersResponseData; +import org.apache.kafka.common.metadata.FenceBrokerRecord; +import org.apache.kafka.common.metadata.PartitionChangeRecord; +import org.apache.kafka.common.metadata.PartitionRecord; +import org.apache.kafka.common.metadata.TopicRecord; +import org.apache.kafka.common.metadata.UnfenceBrokerRecord; +import org.apache.kafka.common.metadata.UnregisterBrokerRecord; +import org.apache.kafka.common.protocol.Errors; +import org.apache.kafka.common.requests.ApiError; +import org.apache.kafka.common.utils.LogContext; +import org.apache.kafka.controller.BrokersToIsrs.TopicPartition; +import org.apache.kafka.metadata.ApiMessageAndVersion; +import org.apache.kafka.metadata.BrokerHeartbeatReply; +import org.apache.kafka.metadata.BrokerRegistration; +import org.apache.kafka.timeline.SnapshotRegistry; +import org.apache.kafka.timeline.TimelineHashMap; +import org.slf4j.Logger; + +import java.util.AbstractMap.SimpleImmutableEntry; +import java.util.ArrayList; +import java.util.Arrays; +import java.util.Collections; +import java.util.HashMap; +import java.util.HashSet; +import java.util.Iterator; +import java.util.List; +import java.util.Map; +import java.util.Map.Entry; +import java.util.Objects; +import java.util.Random; + +import static org.apache.kafka.clients.admin.AlterConfigOp.OpType.SET; +import static org.apache.kafka.common.config.ConfigResource.Type.TOPIC; + + +/** + * The ReplicationControlManager is the part of the controller which deals with topics + * and partitions. It is responsible for managing the in-sync replica set and leader + * of each partition, as well as administrative tasks like creating or deleting topics. + */ +public class ReplicationControlManager { + static class TopicControlInfo { + private final Uuid id; + private final TimelineHashMap<Integer, PartitionControlInfo> parts; + + TopicControlInfo(SnapshotRegistry snapshotRegistry, Uuid id) { + this.id = id; + this.parts = new TimelineHashMap<>(snapshotRegistry, 0); + } + } + + static class PartitionControlInfo { + private final int[] replicas; + private final int[] isr; + private final int[] removingReplicas; + private final int[] addingReplicas; + private final int leader; + private final int leaderEpoch; + private final int partitionEpoch; + + PartitionControlInfo(PartitionRecord record) { + this(Replicas.toArray(record.replicas()), + Replicas.toArray(record.isr()), + Replicas.toArray(record.removingReplicas()), + Replicas.toArray(record.addingReplicas()), + record.leader(), + record.leaderEpoch(), + record.partitionEpoch()); + } + + PartitionControlInfo(int[] replicas, int[] isr, int[] removingReplicas, + int[] addingReplicas, int leader, int leaderEpoch, + int partitionEpoch) { + this.replicas = replicas; + this.isr = isr; + this.removingReplicas = removingReplicas; + this.addingReplicas = addingReplicas; + this.leader = leader; + this.leaderEpoch = leaderEpoch; + this.partitionEpoch = partitionEpoch; + } + + PartitionControlInfo merge(PartitionChangeRecord record) { + int[] newIsr = (record.isr() == null) ? isr : Replicas.toArray(record.isr()); + int newLeader; + int newLeaderEpoch; + if (record.leader() == Integer.MIN_VALUE) { + newLeader = leader; + newLeaderEpoch = leaderEpoch; + } else { + newLeader = record.leader(); + newLeaderEpoch = leaderEpoch + 1; Review comment: Hmm... ReplicationControlManager should not allow this to happen during an alter isr request. There is some code that checks if the alter isr request is attempting to remove the current leader from the isr, and returns an error if so. So the leader should not be changed by an alter isr request and therefore the leader epoch will not be. ``` if (!Replicas.contains(newIsr, partition.leader)) { // An alterIsr request can't remove the current leader. responseTopicData.partitions().add(new AlterIsrResponseData.PartitionData(). setPartitionIndex(partitionData.partitionIndex()). setErrorCode(Errors.INVALID_REQUEST.code())); continue; } ``` ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org