chia7712 commented on code in PR #23383: URL: https://github.com/apache/kafka/pull/23383#discussion_r3944374125
########## server/src/main/java/org/apache/kafka/server/purgatory/DelayedFetch.java: ########## @@ -0,0 +1,257 @@ +/* + * 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.server.purgatory; + +import org.apache.kafka.common.TopicIdPartition; +import org.apache.kafka.common.errors.FencedLeaderEpochException; +import org.apache.kafka.common.errors.KafkaStorageException; +import org.apache.kafka.common.errors.NotLeaderOrFollowerException; +import org.apache.kafka.common.errors.UnknownTopicOrPartitionException; +import org.apache.kafka.common.message.OffsetForLeaderEpochResponseData.EpochEndOffset; +import org.apache.kafka.common.metrics.internals.MetricsUtils; +import org.apache.kafka.common.protocol.Errors; +import org.apache.kafka.common.requests.FetchRequest.PartitionData; +import org.apache.kafka.server.log.remote.TopicPartitionLog; +import org.apache.kafka.server.metrics.KafkaMetricsGroup; +import org.apache.kafka.server.quota.ReplicaQuota; +import org.apache.kafka.server.storage.log.FetchParams; +import org.apache.kafka.server.storage.log.FetchPartitionData; +import org.apache.kafka.storage.internals.log.FetchPartitionStatus; +import org.apache.kafka.storage.internals.log.LogOffsetMetadata; +import org.apache.kafka.storage.internals.log.LogOffsetSnapshot; +import org.apache.kafka.storage.internals.log.LogReadResult; + +import com.yammer.metrics.core.Meter; + +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import java.util.LinkedHashMap; +import java.util.Map; +import java.util.Optional; +import java.util.concurrent.TimeUnit; +import java.util.function.Consumer; + +import static org.apache.kafka.common.requests.OffsetsForLeaderEpochResponse.UNDEFINED_EPOCH; +import static org.apache.kafka.common.requests.OffsetsForLeaderEpochResponse.UNDEFINED_EPOCH_OFFSET; + +/** + * A delayed fetch operation that can be created by the replica manager and watched + * in the fetch operation purgatory. + */ +public class DelayedFetch extends DelayedOperation { + private static final Logger LOG = LoggerFactory.getLogger(DelayedFetch.class); + + // Changing the package or class name may cause incompatibility with existing code and metrics configuration. + private static final KafkaMetricsGroup METRICS_GROUP = + new KafkaMetricsGroup("kafka.server", "DelayedFetchMetrics"); + private static final String FETCHER_TYPE_KEY = "fetcherType"; + private static final Meter FOLLOWER_EXPIRED_REQUEST_METER = METRICS_GROUP.newMeter( + "ExpiresPerSec", + "requests", + TimeUnit.SECONDS, + MetricsUtils.getTags(FETCHER_TYPE_KEY, "follower") + ); + private static final Meter CONSUMER_EXPIRED_REQUEST_METER = METRICS_GROUP.newMeter( + "ExpiresPerSec", + "requests", + TimeUnit.SECONDS, + MetricsUtils.getTags(FETCHER_TYPE_KEY, "consumer") + ); + + private final FetchParams params; + private final LinkedHashMap<TopicIdPartition, FetchPartitionStatus> fetchPartitionStatus; + private final ReplicaManagerAdapter replicaManager; + private final ReplicaQuota quota; + private final Consumer<LinkedHashMap<TopicIdPartition, FetchPartitionData>> responseCallback; + + public DelayedFetch(FetchParams params, + LinkedHashMap<TopicIdPartition, FetchPartitionStatus> fetchPartitionStatus, + ReplicaManagerAdapter replicaManager, + ReplicaQuota quota, + Consumer<LinkedHashMap<TopicIdPartition, FetchPartitionData>> responseCallback) { + super(params.maxWaitMs); + this.params = params; + this.fetchPartitionStatus = fetchPartitionStatus; + this.replicaManager = replicaManager; + this.quota = quota; + this.responseCallback = responseCallback; + } + + @Override + public String toString() { + return "DelayedFetch(params=" + params + + ", numPartitions=" + fetchPartitionStatus.size() + + ")"; + } + + /** + * The operation can be completed if: + * + * Case A: This broker is no longer the leader for some partitions it tries to fetch + * Case B: The replica is no longer available on this broker + * Case C: This broker does not know of some partitions it tries to fetch + * Case D: The partition is in an offline log directory on this broker + * Case E: This broker is the leader, but the requested epoch is now fenced + * Case F: The fetch offset locates not on the last segment of the log + * Case G: The accumulated bytes from all the fetching partitions exceeds the minimum bytes + * Case H: A diverging epoch was found, return response to trigger truncation + * Upon completion, should return whatever data is available for each valid partition. + */ + @SuppressWarnings({"CyclomaticComplexity", "NPathComplexity"}) + @Override + public boolean tryComplete() { + int accumulatedSize = 0; + for (Map.Entry<TopicIdPartition, FetchPartitionStatus> entry : fetchPartitionStatus.entrySet()) { + TopicIdPartition topicIdPartition = entry.getKey(); + FetchPartitionStatus fetchStatus = entry.getValue(); + LogOffsetMetadata fetchOffset = fetchStatus.startOffsetMetadata(); + Optional<Integer> fetchLeaderEpoch = fetchStatus.fetchInfo().currentLeaderEpoch; + + try { + if (!fetchOffset.equals(LogOffsetMetadata.UNKNOWN_OFFSET_METADATA)) { + TopicPartitionLog partition = replicaManager.getPartitionOrException(topicIdPartition.topicPartition()); + LogOffsetSnapshot offsetSnapshot = partition.fetchOffsetSnapshot(fetchLeaderEpoch, params.fetchOnlyLeader()); + + LogOffsetMetadata endOffset = switch (params.isolation) { + case LOG_END -> offsetSnapshot.logEndOffset(); + case HIGH_WATERMARK -> offsetSnapshot.highWatermark(); + case TXN_COMMITTED -> offsetSnapshot.lastStableOffset(); + }; + + // Go directly to the check for Case G if the message offsets are the same. If the log segment + // has just rolled, then the high watermark offset will remain the same but be on the old segment, + // which would incorrectly be seen as an instance of Case F. + if (fetchOffset.messageOffset > endOffset.messageOffset) { + // Case F, this can happen when the new fetch operation is on a truncated leader. + LOG.debug("Satisfying fetch {} since it is fetching later segments of partition {}.", this, topicIdPartition); + return forceComplete(); + } else if (fetchOffset.messageOffset < endOffset.messageOffset) { + if (fetchOffset.onOlderSegment(endOffset)) { + // Case F, this can happen when the fetch operation is falling behind the current segment + // or the partition has just rolled a new segment. + LOG.debug("Satisfying fetch {} immediately since it is fetching older segments.", this); + // We will not force complete the fetch request if a replica should be throttled. + if (!params.isFromFollower() || !replicaManager.shouldLeaderThrottle(quota, partition, params.replicaId)) { + return forceComplete(); + } + } else if (fetchOffset.onSameSegment(endOffset)) { + // We take the partition fetch size as upper bound when accumulating the bytes + // (skip if a throttled partition). + int bytesAvailable = Math.min(endOffset.positionDiff(fetchOffset), fetchStatus.fetchInfo().maxBytes); + if (!params.isFromFollower() || !replicaManager.shouldLeaderThrottle(quota, partition, params.replicaId)) { + accumulatedSize += bytesAvailable; + } + } + } + + // Case H: If truncation has caused diverging epoch while this request was in purgatory, + // return to trigger truncation. + Optional<Integer> lastFetchedEpoch = fetchStatus.fetchInfo().lastFetchedEpoch; + if (lastFetchedEpoch.isPresent()) { + int fetchEpoch = lastFetchedEpoch.get(); + EpochEndOffset epochEndOffset = partition.lastOffsetForLeaderEpoch( + fetchLeaderEpoch, + fetchEpoch, + false + ); + if (epochEndOffset.errorCode() != Errors.NONE.code() + || epochEndOffset.endOffset() == UNDEFINED_EPOCH_OFFSET + || epochEndOffset.leaderEpoch() == UNDEFINED_EPOCH) { + LOG.debug( + "Could not obtain last offset for leader epoch for partition {}, epochEndOffset={}.", + topicIdPartition, + epochEndOffset + ); + return forceComplete(); + } else if (epochEndOffset.leaderEpoch() < fetchEpoch + || epochEndOffset.endOffset() < fetchStatus.fetchInfo().fetchOffset) { + LOG.debug( + "Satisfying fetch {} since it has diverging epoch requiring truncation for partition " + + "{} epochEndOffset={} fetchEpoch={} fetchOffset={}.", + this, + topicIdPartition, + epochEndOffset, + fetchEpoch, + fetchStatus.fetchInfo().fetchOffset + ); + return forceComplete(); + } + } + } + } catch (NotLeaderOrFollowerException e) { // Case A or Case B + LOG.debug("Broker is no longer the leader or follower of {}, satisfy {} immediately", topicIdPartition, this); + return forceComplete(); + } catch (UnknownTopicOrPartitionException e) { // Case C + LOG.debug("Broker no longer knows of partition {}, satisfy {} immediately", topicIdPartition, this); + return forceComplete(); + } catch (KafkaStorageException e) { // Case D + LOG.debug("Partition {} is in an offline log directory, satisfy {} immediately", topicIdPartition, this); + return forceComplete(); + } catch (FencedLeaderEpochException e) { // Case E + LOG.debug( + "Broker is the leader of partition {}, but the requested epoch {} is fenced by the latest leader epoch, " + + "satisfy {} immediately", + topicIdPartition, + fetchLeaderEpoch, + this + ); + return forceComplete(); + } + } + + // Case G + if (accumulatedSize >= params.minBytes) { + return forceComplete(); + } + return false; + } + + @Override + public void onExpiration() { + if (params.isFromFollower()) { + FOLLOWER_EXPIRED_REQUEST_METER.mark(); + } else { + CONSUMER_EXPIRED_REQUEST_METER.mark(); + } + } + + /** + * Upon completion, read whatever data is available and pass to the complete callback. + */ + @Override + public void onComplete() { + LinkedHashMap<TopicIdPartition, PartitionData> fetchInfos = new LinkedHashMap<>(); + fetchPartitionStatus.forEach((topicIdPartition, status) -> fetchInfos.put(topicIdPartition, status.fetchInfo())); + + LinkedHashMap<TopicIdPartition, LogReadResult> logReadResults = replicaManager.readFromLogByPurgatory( Review Comment: We get the `logReadResults and then convert them to `FetchPartitionData` immediately, so maybe we could let `readFromLogByPurgatory` return `FetchPartitionData` directly? -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
