LoganZhuZzz commented on code in PR #20089: URL: https://github.com/apache/kafka/pull/20089#discussion_r2191735677
########## coordinator-common/src/main/java/org/apache/kafka/coordinator/common/runtime/CoordinatorLoaderImpl.java: ########## @@ -0,0 +1,338 @@ +/* + * 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.coordinator.common.runtime; + +import org.apache.kafka.common.TopicPartition; +import org.apache.kafka.common.errors.NotLeaderOrFollowerException; +import org.apache.kafka.common.record.ControlRecordType; +import org.apache.kafka.common.record.FileRecords; +import org.apache.kafka.common.record.MemoryRecords; +import org.apache.kafka.common.record.MutableRecordBatch; +import org.apache.kafka.common.record.Record; +import org.apache.kafka.common.record.Records; +import org.apache.kafka.common.requests.TransactionResult; +import org.apache.kafka.common.utils.Time; +import org.apache.kafka.server.storage.log.FetchIsolation; +import org.apache.kafka.server.util.KafkaScheduler; +import org.apache.kafka.storage.internals.log.FetchDataInfo; +import org.apache.kafka.storage.internals.log.UnifiedLog; + +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import java.io.IOException; +import java.nio.ByteBuffer; +import java.util.Optional; +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.ScheduledFuture; +import java.util.concurrent.atomic.AtomicBoolean; +import java.util.function.Function; + +/** + * Coordinator loader which reads records from a partition and replays them + * to a group coordinator. + * + * @param <T> The record type. + */ +public class CoordinatorLoaderImpl<T> implements CoordinatorLoader<T> { + + private static final Logger LOG = LoggerFactory.getLogger(CoordinatorLoaderImpl.class); + + private final Time time; + private final Function<TopicPartition, Optional<UnifiedLog>> partitionLogSupplier; + private final Function<TopicPartition, Optional<Long>> partitionLogEndOffsetSupplier; + private final Deserializer<T> deserializer; + private final int loadBufferSize; + + private final AtomicBoolean isRunning = new AtomicBoolean(true); + private final KafkaScheduler scheduler = new KafkaScheduler(1); + + public CoordinatorLoaderImpl( + Time time, + Function<TopicPartition, Optional<UnifiedLog>> partitionLogSupplier, + Function<TopicPartition, Optional<Long>> partitionLogEndOffsetSupplier, + Deserializer<T> deserializer, + int loadBufferSize + ) { + this.time = time; + this.partitionLogSupplier = partitionLogSupplier; + this.partitionLogEndOffsetSupplier = partitionLogEndOffsetSupplier; + this.deserializer = deserializer; + this.loadBufferSize = loadBufferSize; + this.scheduler.startup(); + } + + /** + * Loads the coordinator by reading all the records from the TopicPartition + * and applying them to the Replayable object. + * + * @param tp The TopicPartition to read from. + * @param coordinator The object to apply records to. + */ + @Override + public CompletableFuture<LoadSummary> load(TopicPartition tp, CoordinatorPlayback<T> coordinator) { + final CompletableFuture<LoadSummary> future = new CompletableFuture<>(); + long startTimeMs = time.milliseconds(); + try { + ScheduledFuture<?> result = scheduler.scheduleOnce(String.format("Load coordinator from %s", tp), + () -> doLoad(tp, coordinator, future, startTimeMs)); + if (result.isCancelled()) { + future.completeExceptionally(new RuntimeException("Coordinator loader is closed.")); + } + } catch (Exception e) { + future.completeExceptionally(e); + } + return future; + } + + private void doLoad( + TopicPartition tp, + CoordinatorPlayback<T> coordinator, + CompletableFuture<LoadSummary> future, + long startTimeMs + ) { + long schedulerQueueTimeMs = time.milliseconds() - startTimeMs; + try { + Optional<UnifiedLog> logOpt = partitionLogSupplier.apply(tp); + if (logOpt.isEmpty()) { + future.completeExceptionally(new NotLeaderOrFollowerException( + "Could not load records from " + tp + " because the log does not exist.")); + return; + } + + UnifiedLog log = logOpt.get(); + + // Buffer may not be needed if records are read from memory. + ByteBuffer buffer = ByteBuffer.allocate(0); + long currentOffset = log.logStartOffset(); + LoadStats stats = new LoadStats(); + + long previousHighWatermark = -1L; + while (shouldFetchNextBatch(currentOffset, logEndOffset(tp), stats.readAtLeastOneRecord)) { + FetchDataInfo fetchDataInfo = log.read(currentOffset, loadBufferSize, FetchIsolation.LOG_END, true); + + stats.readAtLeastOneRecord = fetchDataInfo.records.sizeInBytes() > 0; + + MemoryRecords memoryRecords = toReadableMemoryRecords(tp, fetchDataInfo.records, buffer); + + ReplayResult replayResult = processMemoryRecords(tp, log, memoryRecords, coordinator, stats, currentOffset, previousHighWatermark); + currentOffset = replayResult.nextOffset; + previousHighWatermark = replayResult.highWatermark; + } + + completeLoadFuture(tp, future, startTimeMs, schedulerQueueTimeMs, stats); + } catch (Throwable ex) { + future.completeExceptionally(ex); + } + } + + private long logEndOffset(TopicPartition tp) { + return partitionLogEndOffsetSupplier.apply(tp).orElse(-1L); + } + + /** + * Returns true if it's still valid to fetch the next batch of records. + * <p> + * This method ensures fetching continues only under safe and meaningful conditions: + * <ul> + * <li>The current offset is less than the log end offset.</li> + * <li>At least one record was read in the previous fetch. This ensures that fetching stops even if the + * current offset remains smaller than the log end offset but the log is empty. This could happen with compacted topics.</li> + * <li>The log end offset is not -1L, which ensures the partition is online and is still the leader.</li> + * <li>The loader is still running.</li> + * </ul> + */ + private boolean shouldFetchNextBatch(long currentOffset, long logEndOffset, boolean readAtLeastOneRecord) { + return currentOffset < logEndOffset && readAtLeastOneRecord && isRunning.get(); + } + + private MemoryRecords toReadableMemoryRecords(TopicPartition tp, Records records, ByteBuffer buffer) throws IOException { + if (records instanceof MemoryRecords memoryRecords) { + return memoryRecords; + } else if (records instanceof FileRecords fileRecords) { + int sizeInBytes = fileRecords.sizeInBytes(); + int bytesNeeded = Math.max(loadBufferSize, sizeInBytes); + + // "minOneMessage = true in the above log.read() means that the buffer may need to + // be grown to ensure progress can be made. + if (buffer.capacity() < bytesNeeded) { + if (loadBufferSize < bytesNeeded) { + LOG.warn("Loaded metadata from {} with buffer larger ({} bytes) than" + + " configured buffer size ({} bytes).", tp, bytesNeeded, loadBufferSize); + } + + buffer = ByteBuffer.allocate(bytesNeeded); + } else { + buffer.clear(); + } + + fileRecords.readInto(buffer, 0); + return MemoryRecords.readableRecords(buffer); + } else { + throw new IllegalArgumentException("Unsupported record type: " + records.getClass()); + } + } + + private ReplayResult processMemoryRecords( + TopicPartition tp, + UnifiedLog log, + MemoryRecords memoryRecords, + CoordinatorPlayback<T> coordinator, + LoadStats loadStats, + long currentOffset, + long previousHighWatermark + ) { + + for (MutableRecordBatch batch : memoryRecords.batches()) { + if (batch.isControlBatch()) { + for (Record record : batch) { + ControlRecordType controlRecord = ControlRecordType.parse(record.key()); + if (controlRecord == ControlRecordType.COMMIT) { + if (LOG.isTraceEnabled()) { + LOG.trace("Replaying end transaction marker from {} at offset {} to commit" + + " transaction with producer id {} and producer epoch {}.", + tp, record.offset(), batch.producerId(), batch.producerEpoch()); + } + coordinator.replayEndTransactionMarker( + batch.producerId(), + batch.producerEpoch(), + TransactionResult.COMMIT + ); + } else if (controlRecord == ControlRecordType.ABORT) { + if (LOG.isTraceEnabled()) { + LOG.trace("Replaying end transaction marker from {} at offset {} to abort" + + " transaction with producer id {} and producer epoch {}.", + tp, record.offset(), batch.producerId(), batch.producerEpoch()); + } + coordinator.replayEndTransactionMarker( + batch.producerId(), + batch.producerEpoch(), + TransactionResult.ABORT + ); + } + } + } else { + for (Record record : batch) { + loadStats.numRecords++; + + Optional<T> coordinatorRecordOpt = Optional.empty(); + try { + coordinatorRecordOpt = Optional.ofNullable(deserializer.deserialize(record.key(), record.value())); + } catch (Deserializer.UnknownRecordTypeException ex) { + LOG.warn("Unknown record type {} while loading offsets and group metadata from {}." + + " Ignoring it. It could be a left over from an aborted upgrade.", ex.unknownType(), tp); + } catch (RuntimeException ex) { + String msg = String.format("Deserializing record %s from %s failed due to: %s", record, tp, ex.getMessage()); + LOG.error(msg); + throw new RuntimeException(msg, ex); + } + + if (coordinatorRecordOpt.isPresent()) { + T coordinatorRecord = coordinatorRecordOpt.get(); + try { + if (LOG.isTraceEnabled()) { + LOG.trace("Replaying record {} from {} at offset {} with producer id {}" + + " and producer epoch {}.", coordinatorRecord, tp, record.offset(), batch.producerId(), batch.producerEpoch()); + } + coordinator.replay( + record.offset(), + batch.producerId(), + batch.producerEpoch(), + coordinatorRecord + ); + } catch (RuntimeException ex) { + String msg = String.format("Replaying record %s from %s at offset %d with producer id %d and" + + " producer epoch %d failed due to: %s.", coordinatorRecord, tp, record.offset(), + batch.producerId(), batch.producerEpoch(), ex.getMessage()); + LOG.error(msg); + throw new RuntimeException(msg, ex); + } + } + } + } + + // Note that the high watermark can be greater than the current offset but as we load more records + // the current offset will eventually surpass the high watermark. Also note that the high watermark + // will continue to advance while loading. + currentOffset = batch.nextOffset(); + long currentHighWatermark = log.highWatermark(); + if (currentOffset >= currentHighWatermark) { + coordinator.updateLastWrittenOffset(currentOffset); + + if (currentHighWatermark > previousHighWatermark) { + coordinator.updateLastCommittedOffset(currentHighWatermark); + previousHighWatermark = currentHighWatermark; + } + } + } + loadStats.numBytes += memoryRecords.sizeInBytes(); + return new ReplayResult(currentOffset, previousHighWatermark); + } + + private void completeLoadFuture( Review Comment: Thanks for the feedback. I’ve fixed these issues. PTAL -- 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
