jsancio commented on code in PR #16626: URL: https://github.com/apache/kafka/pull/16626#discussion_r1684658784
########## raft/src/main/java/org/apache/kafka/raft/internals/AddVoterHandler.java: ########## @@ -0,0 +1,383 @@ +/* + * 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.raft.internals; + +import org.apache.kafka.common.Node; +import org.apache.kafka.common.feature.SupportedVersionRange; +import org.apache.kafka.common.message.AddRaftVoterResponseData; +import org.apache.kafka.common.message.ApiVersionsRequestData; +import org.apache.kafka.common.message.ApiVersionsResponseData; +import org.apache.kafka.common.protocol.Errors; +import org.apache.kafka.common.requests.ApiVersionsRequest; +import org.apache.kafka.common.utils.AppInfoParser; +import org.apache.kafka.common.utils.LogContext; +import org.apache.kafka.common.utils.Time; +import org.apache.kafka.raft.Endpoints; +import org.apache.kafka.raft.LeaderState; +import org.apache.kafka.raft.LogOffsetMetadata; +import org.apache.kafka.raft.RaftUtil; +import org.apache.kafka.server.common.KRaftVersion; + +import org.slf4j.Logger; + +import java.util.Optional; +import java.util.OptionalLong; +import java.util.concurrent.CompletableFuture; + +/** + * This type implements the protocol for adding a voter to a KRaft partition. + * + * The general algorithm for adding a voter to the voter set is: + * + * 1. Check that the leader has fenced the previous leader(s) by checking that the HWM is known, + * otherwise return the REQUEST_TIMED_OUT error. + * 2. Check that the cluster supports kraft.version 1, otherwise return the UNSUPPORTED_VERSION error. + * 3. Check that there are no uncommitted voter changes, otherwise return the REQUEST_TIMED_OUT error. + * 4. Check that the new voter's id is not part of the existing voter set, otherwise return the + * DUPLICATE_VOTER error. + * 5. Send an API_VERSIONS RPC to the first (default) listener to discover the supported + * kraft.version of the new voter. + * 6. Check that the new voter supports the current kraft.version, otherwise return the + * INVALID_REQUEST error. + * 7. Check that the new voter is caught up to the log end offset of the leader, otherwise return + * a REQUEST_TIMED_OUT error. + * 8. Append the updated VotersRecord to the log. The KRaft internal listener will read this + * uncommitted record from the log and add the new voter to the set of voters. + * 9. Wait for the VotersRecord to commit using the majority of the new set of voters. Return a + * REQUEST_TIMED_OUT error if it doesn't commit in time. + * 10. Send the AddVoter successful response to the client. + * + * The algorithm above could be improved as part of KAFKA-17147. Instead of returning an error + * immediately for 1., 2. and 7., KRaft can wait with a timeout until those invariants are true. + */ +public final class AddVoterHandler { + private final KRaftControlRecordStateMachine partitionState; + private final RequestSender requestSender; + private final Time time; + private final Logger logger; + + public AddVoterHandler( + KRaftControlRecordStateMachine partitionState, + RequestSender requestSender, + Time time, + LogContext logContext + ) { + this.partitionState = partitionState; + this.requestSender = requestSender; + this.time = time; + this.logger = logContext.logger(AddVoterHandler.class); + } + + public CompletableFuture<AddRaftVoterResponseData> handleAddVoterRequest( + LeaderState<?> leaderState, + ReplicaKey voterKey, + Endpoints voterEndpoints, + long currentTimeMs + ) { + // Check if there are any pending add voter requests + if (isOperationPending(leaderState, currentTimeMs)) { + return CompletableFuture.completedFuture( + RaftUtil.addVoterResponse( + Errors.REQUEST_TIMED_OUT, + "Request timed out waiting for leader to handle previous add voter request" + ) + ); + } + + Optional<LogHistory.Entry<VoterSet>> votersEntry = partitionState.lastVoterSetEntry(); + + // Check that the leader has established a HWM and committed the current epoch + Optional<Long> highWatermark = leaderState.highWatermark().map(LogOffsetMetadata::offset); + if (!highWatermark.isPresent()) { + return CompletableFuture.completedFuture( + RaftUtil.addVoterResponse( + Errors.REQUEST_TIMED_OUT, + "Request timed out waiting for leader to establish HWM and fence previous voter changes" + ) + ); + } + + // Check that the cluster supports kraft.version >= 1 + KRaftVersion kraftVersion = partitionState.lastKraftVersion(); + if (!kraftVersion.isReconfigSupported()) { + return CompletableFuture.completedFuture( + RaftUtil.addVoterResponse( + Errors.UNSUPPORTED_VERSION, + String.format( + "Cluster doesn't support adding voter because the %s feature is %s", + kraftVersion.featureName(), + kraftVersion.featureLevel() + ) + ) + ); + } + + // Check that there are no uncommitted VotersRecord + if (!votersEntry.isPresent() || votersEntry.get().offset() >= highWatermark.get()) { + return CompletableFuture.completedFuture( + RaftUtil.addVoterResponse( + Errors.REQUEST_TIMED_OUT, + String.format( + "Request timed out waiting for voters to commit the latest voter change at %s with HWM %d", + votersEntry.map(LogHistory.Entry::offset), + highWatermark.get() + ) + ) + ); + } + + // Check that the new voter id is not part of the current voter set + VoterSet voters = votersEntry.get().value(); + if (voters.voterIds().contains(voterKey.id())) { + return CompletableFuture.completedFuture( + RaftUtil.addVoterResponse( + Errors.DUPLICATE_VOTER, + String.format( + "The voter id for %s is already part of the set of voters %s.", + voterKey, + voters.voterKeys() + ) + ) + ); + } + + // Send API_VERSIONS request to new voter to discover their supported kraft.version range + OptionalLong timeout = requestSender.send( + voterEndpoints + .address(requestSender.listenerName()) + .map(address -> new Node(voterKey.id(), address.getHostName(), address.getPort())) + .orElseThrow( + () -> new IllegalArgumentException( + String.format( + "Provided listeners %s do not contain a listener for %s", + voterEndpoints, + requestSender.listenerName() + ) + ) + ), + this::buildApiVersionsRequest, + currentTimeMs + ); + if (!timeout.isPresent()) { + return CompletableFuture.completedFuture( + RaftUtil.addVoterResponse( + Errors.REQUEST_TIMED_OUT, + String.format("New voter %s is not ready to receive requests", voterKey) + ) + ); + } + + AddVoterHandlerState state = new AddVoterHandlerState( + voterKey, + voterEndpoints, + time.timer(timeout.getAsLong()) + ); + leaderState.resetAddVoterHandlerState( + Errors.UNKNOWN_SERVER_ERROR, + null, + Optional.of(state) + ); + + return state.future(); + } + + public boolean handleApiVersionsResponse( + LeaderState<?> leaderState, + Node source, + Errors error, + Optional<ApiVersionsResponseData.SupportedFeatureKey> supportedKraftVersions, + long currentTimeMs + ) { + Optional<AddVoterHandlerState> handlerState = leaderState.addVoterHandlerState(); + if (!handlerState.isPresent()) { + // There are no pending add operation just ignore the api response + return true; + } + + // Check that the API_VERSIONS response matches the id of the voter getting added + AddVoterHandlerState current = handlerState.get(); + if (!current.expectingApiResponse(source.id())) { + logger.info( + "API_VERSIONS response is not expected from {}: voterKey is {}, lastOffset is {}", + source, + current.voterKey(), + current.lastOffset() + ); + + return true; + } + + // Abort operation if the API_VERSIONS returned an error + if (error != Errors.NONE) { + logger.info( + "Aborting add voter operation for {} at {} since API_VERSIONS returned an error {}", + current.voterKey(), + current.voterEndpoints(), + error + ); + + abortAddVoter( + leaderState, + Errors.REQUEST_TIMED_OUT, + String.format( + "Aborted add voter operation for since API_VERSIONS returned an error %s", + error + ) + ); + + return false; + } + + // Check that the new voter supports the kraft.verion for reconfiguration + KRaftVersion kraftVersion = partitionState.lastKraftVersion(); + if (!validVersionRange(kraftVersion, supportedKraftVersions)) { + logger.info( + "Aborting add voter operation for {} at {} since kraft.version range {} doesn't " + + "support reconfiguration", + current.voterKey(), + current.voterEndpoints(), + supportedKraftVersions + ); + + abortAddVoter( + leaderState, + Errors.INVALID_REQUEST, + String.format( + "Aborted add voter operation for %s since the %s range %s doesn't " + + "support the finalized version %s", + current.voterKey(), + KRaftVersion.FEATURE_NAME, + supportedKraftVersions + .map( + range -> String.format( + "(min: %s, max: %s", + range.minVersion(), + range.maxVersion() + ) + ) + .orElse("(min: 0, max: 0)"), + kraftVersion.featureLevel() + ) + ); + + return true; + } + + // Check that the new voter is caught up to the LEO to avoid delays in HWM increases + if (!leaderState.isReplicaCaughtUp(current.voterKey(), currentTimeMs)) { + logger.info( + "Aborting add voter operation for {} at {} since it is lagging behind: {}", + current.voterKey(), + current.voterEndpoints(), + leaderState.getReplicaState(current.voterKey()) + ); + + abortAddVoter( + leaderState, + Errors.REQUEST_TIMED_OUT, + String.format( + "Aborted add voter operation for %s since it is lagging behind", + current.voterKey() + ) + ); + + return true; + } + + // Add the new voter to the set of voters and append the record to the log + VoterSet newVoters = partitionState + .lastVoterSet() + .addVoter( + VoterSet.VoterNode.of( + current.voterKey(), + current.voterEndpoints(), + new SupportedVersionRange( + supportedKraftVersions.get().minVersion(), + supportedKraftVersions.get().maxVersion() + ) + ) + ) + .orElseThrow(() -> + new IllegalStateException( + String.format( + "Unable to add %s to the set of voters %s", + current.voterKey(), + partitionState.lastVoterSet() + ) + ) + ); + current.setLastOffset(leaderState.appendVotersRecord(newVoters, currentTimeMs)); + + return true; + } + + public void highWatermarkUpdated(LeaderState<?> leaderState) { + leaderState.addVoterHandlerState().ifPresent(current -> { + leaderState.highWatermark().ifPresent(highWatermark -> { + current.lastOffset().ifPresent(lastOffset -> { + if (highWatermark.offset() > lastOffset) { + // VotersRecord with the added voter was committed; complete the RPC + completeAddVoter(leaderState); + } + }); + }); + }); + } + + public long maybeExpirePendingOperation(LeaderState<?> leaderState, long currentTimeMs) { + long timeUntilOperationExpiration = leaderState + .addVoterHandlerState() + .map(state -> state.timeUntilOperationExpiration(currentTimeMs)) + .orElse(Long.MAX_VALUE); + + if (timeUntilOperationExpiration == 0) { + abortAddVoter(leaderState, Errors.REQUEST_TIMED_OUT, null); + return Long.MAX_VALUE; + } else { + return timeUntilOperationExpiration; + } + + } + + private ApiVersionsRequestData buildApiVersionsRequest() { Review Comment: The problem with the builder is that it requires a `version`. At this layer we don't know the versions supported by the destination since it is possible that the leader never created a connection, etc. Just like other requests sent by `KafkaRaftClient` it is better to let `NetworkClient` negotiate the version. One solution is to make `ApiVersionRequest.Builder.DATA` public and AddVoterHandler just sends that. What do you think? -- 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. To unsubscribe, e-mail: jira-unsubscr...@kafka.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
