mumrah commented on a change in pull request #10085: URL: https://github.com/apache/kafka/pull/10085#discussion_r605171450
########## File path: raft/src/main/java/org/apache/kafka/raft/KafkaRaftClient.java ########## @@ -326,6 +336,14 @@ private void updateListenersProgress(List<ListenerContext> listenerContexts, lon } } + private Optional<SnapshotReader<T>> latestSnapshot() { + return log.latestSnapshotId().flatMap(snapshoId -> { + return log Review comment: nit: can omit the curly braces and return here ########## File path: raft/src/main/java/org/apache/kafka/raft/ReplicatedCounter.java ########## @@ -68,20 +70,65 @@ public synchronized void increment() { @Override public synchronized void handleCommit(BatchReader<Integer> reader) { try { - int initialValue = this.committed; + int initialValue = committed; while (reader.hasNext()) { BatchReader.Batch<Integer> batch = reader.next(); log.debug("Handle commit of batch with records {} at base offset {}", batch.records(), batch.baseOffset()); for (Integer value : batch.records()) { - if (value != this.committed + 1) { - throw new AssertionError("Expected next committed value to be " + - (this.committed + 1) + ", but instead found " + value + " on node " + nodeId); + if (value != committed + 1) { + throw new AssertionError( + String.format( + "Expected next committed value to be %s, but instead found %s on node %s", + committed + 1, + value, + nodeId + ) + ); } - this.committed = value; + committed = value; } + + nextReadOffset = batch.lastOffset() + 1; + readEpoch = batch.epoch(); } log.debug("Counter incremented from {} to {}", initialValue, committed); + + if (lastSnapshotEndOffset + 10 < nextReadOffset) { Review comment: Where does `10` come from? Should it be configurable? ########## File path: raft/src/main/java/org/apache/kafka/raft/KafkaRaftClient.java ########## @@ -326,6 +336,14 @@ private void updateListenersProgress(List<ListenerContext> listenerContexts, lon } } + private Optional<SnapshotReader<T>> latestSnapshot() { + return log.latestSnapshotId().flatMap(snapshoId -> { Review comment: typo: snapshoId ########## File path: raft/src/main/java/org/apache/kafka/raft/ReplicatedCounter.java ########## @@ -68,20 +70,65 @@ public synchronized void increment() { @Override public synchronized void handleCommit(BatchReader<Integer> reader) { try { - int initialValue = this.committed; + int initialValue = committed; while (reader.hasNext()) { BatchReader.Batch<Integer> batch = reader.next(); log.debug("Handle commit of batch with records {} at base offset {}", batch.records(), batch.baseOffset()); for (Integer value : batch.records()) { - if (value != this.committed + 1) { - throw new AssertionError("Expected next committed value to be " + - (this.committed + 1) + ", but instead found " + value + " on node " + nodeId); + if (value != committed + 1) { Review comment: nit: not really part of this PR, but can we use variable names besides "value" and "initialValue"? ########## File path: raft/src/main/java/org/apache/kafka/raft/internals/SerdeRecordsIterator.java ########## @@ -0,0 +1,231 @@ +/* + * 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 java.io.IOException; +import java.nio.ByteBuffer; +import java.util.ArrayList; +import java.util.Iterator; +import java.util.List; +import java.util.NoSuchElementException; +import java.util.Optional; +import org.apache.kafka.common.protocol.DataInputStreamReadable; +import org.apache.kafka.common.protocol.Readable; +import org.apache.kafka.common.record.DefaultRecordBatch; +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.Records; +import org.apache.kafka.common.utils.BufferSupplier; +import org.apache.kafka.raft.BatchReader.Batch; +import org.apache.kafka.raft.RecordSerde; + +public final class SerdeRecordsIterator<T> implements Iterator<Batch<T>>, AutoCloseable { + private final Records records; + private final RecordSerde<T> serde; + private final BufferSupplier bufferSupplier; + private final int maxBatchSize; + + private Optional<Iterator<MutableRecordBatch>> nextBatches = Optional.empty(); + private Optional<Batch<T>> nextBatch = Optional.empty(); + // Buffer used to as the backing store for nextBatches if needed + private Optional<ByteBuffer> allocatedBuffer = Optional.empty(); + // Number of bytes from records that read + private int bytesRead = 0; + private boolean isClosed = false; + + public SerdeRecordsIterator( + Records records, + RecordSerde<T> serde, + BufferSupplier bufferSupplier, + int maxBatchSize + ) { + this.records = records; + this.serde = serde; + this.bufferSupplier = bufferSupplier; + this.maxBatchSize = maxBatchSize; + } + + @Override + public boolean hasNext() { + checkIfClosed(); + + if (!nextBatch.isPresent()) { + nextBatch = nextBatch(); + } + + return nextBatch.isPresent(); + } + + @Override + public Batch<T> next() { + if (!hasNext()) { + throw new NoSuchElementException("Batch iterator doesn't have any more elements"); + } + + Batch<T> batch = nextBatch.get(); + nextBatch = Optional.empty(); + + return batch; + } + + @Override + public void close() { + isClosed = true; + allocatedBuffer.ifPresent(bufferSupplier::release); + allocatedBuffer = Optional.empty(); + } + + private void checkIfClosed() { + if (isClosed) { + throw new IllegalStateException("Serde record batch itererator was closed"); + } + } + + private Optional<Iterator<MutableRecordBatch>> nextBatches() { Review comment: Is the Optional necessary here? Could we instead return an empty Iterator? ########## File path: raft/src/main/java/org/apache/kafka/raft/internals/SerdeRecordsIterator.java ########## @@ -0,0 +1,231 @@ +/* + * 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 java.io.IOException; +import java.nio.ByteBuffer; +import java.util.ArrayList; +import java.util.Iterator; +import java.util.List; +import java.util.NoSuchElementException; +import java.util.Optional; +import org.apache.kafka.common.protocol.DataInputStreamReadable; +import org.apache.kafka.common.protocol.Readable; +import org.apache.kafka.common.record.DefaultRecordBatch; +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.Records; +import org.apache.kafka.common.utils.BufferSupplier; +import org.apache.kafka.raft.BatchReader.Batch; +import org.apache.kafka.raft.RecordSerde; + +public final class SerdeRecordsIterator<T> implements Iterator<Batch<T>>, AutoCloseable { + private final Records records; + private final RecordSerde<T> serde; + private final BufferSupplier bufferSupplier; + private final int maxBatchSize; + + private Optional<Iterator<MutableRecordBatch>> nextBatches = Optional.empty(); + private Optional<Batch<T>> nextBatch = Optional.empty(); + // Buffer used to as the backing store for nextBatches if needed + private Optional<ByteBuffer> allocatedBuffer = Optional.empty(); + // Number of bytes from records that read + private int bytesRead = 0; + private boolean isClosed = false; + + public SerdeRecordsIterator( + Records records, + RecordSerde<T> serde, + BufferSupplier bufferSupplier, + int maxBatchSize + ) { + this.records = records; + this.serde = serde; + this.bufferSupplier = bufferSupplier; + this.maxBatchSize = maxBatchSize; + } + + @Override + public boolean hasNext() { + checkIfClosed(); + + if (!nextBatch.isPresent()) { + nextBatch = nextBatch(); + } + + return nextBatch.isPresent(); + } + + @Override + public Batch<T> next() { + if (!hasNext()) { + throw new NoSuchElementException("Batch iterator doesn't have any more elements"); + } + + Batch<T> batch = nextBatch.get(); + nextBatch = Optional.empty(); + + return batch; + } + + @Override + public void close() { + isClosed = true; + allocatedBuffer.ifPresent(bufferSupplier::release); + allocatedBuffer = Optional.empty(); + } + + private void checkIfClosed() { + if (isClosed) { + throw new IllegalStateException("Serde record batch itererator was closed"); + } + } + + private Optional<Iterator<MutableRecordBatch>> nextBatches() { + int recordSize = records.sizeInBytes(); + if (bytesRead < recordSize) { + final MemoryRecords memoryRecords; + if (records instanceof MemoryRecords) { + bytesRead = recordSize; + memoryRecords = (MemoryRecords) records; + } else if (records instanceof FileRecords) { + final ByteBuffer buffer; + if (allocatedBuffer.isPresent()) { + buffer = allocatedBuffer.get(); + buffer.compact(); + + if (!buffer.hasRemaining()) { + // The buffer is not big enough to read an entire batch + throw new IllegalStateException( + String.format( + "Unable to read batch from file records buffer %s with maximum batch %s and record size %s", + buffer, + maxBatchSize, + records.sizeInBytes() + ) + ); + } + } else { + buffer = bufferSupplier.get(Math.min(maxBatchSize, records.sizeInBytes())); + allocatedBuffer = Optional.of(buffer); + } + + int start = buffer.position(); + try { + ((FileRecords) records).readInto(buffer, bytesRead); + } catch (IOException e) { + throw new RuntimeException("Failed to read records into memory", e); + } + + bytesRead += buffer.limit() - start; + memoryRecords = MemoryRecords.readableRecords(buffer.slice()); + } else { + throw new IllegalStateException(String.format("Unexpected Records type %s", records.getClass())); + } + + return Optional.of(memoryRecords.batchIterator()); + } else { + return Optional.empty(); + } + } + + private Optional<Batch<T>> nextBatch() { + if (!nextBatches.isPresent()) { + nextBatches = nextBatches(); + } + + while (nextBatches.isPresent()) { + if (nextBatches.get().hasNext()) { + MutableRecordBatch nextBatch = nextBatches.get().next(); + + // Update the buffer position to reflect the read batch + allocatedBuffer.ifPresent(buffer -> buffer.position(buffer.position() + nextBatch.sizeInBytes())); + + if (!(nextBatch instanceof DefaultRecordBatch)) { + throw new IllegalStateException( + String.format("DefaultRecordBatch expected by record type was %s", nextBatch.getClass()) + ); + } + return Optional.of(readBatch((DefaultRecordBatch) nextBatch)); + } + + nextBatches = nextBatches(); + } + + return Optional.empty(); + } + + private Batch<T> readBatch(DefaultRecordBatch batch) { + final Batch<T> result; + if (batch.isControlBatch()) { + result = Batch.empty(batch.baseOffset(), batch.partitionLeaderEpoch(), batch.lastOffset()); + } else { + Integer numRecords = batch.countOrNull(); Review comment: Maybe use OptionalInt here? We're pretty consistently using Optional everywhere else in this class ########## File path: raft/src/main/java/org/apache/kafka/raft/internals/SerdeRecordsIterator.java ########## @@ -0,0 +1,231 @@ +/* + * 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 java.io.IOException; +import java.nio.ByteBuffer; +import java.util.ArrayList; +import java.util.Iterator; +import java.util.List; +import java.util.NoSuchElementException; +import java.util.Optional; +import org.apache.kafka.common.protocol.DataInputStreamReadable; +import org.apache.kafka.common.protocol.Readable; +import org.apache.kafka.common.record.DefaultRecordBatch; +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.Records; +import org.apache.kafka.common.utils.BufferSupplier; +import org.apache.kafka.raft.BatchReader.Batch; +import org.apache.kafka.raft.RecordSerde; + +public final class SerdeRecordsIterator<T> implements Iterator<Batch<T>>, AutoCloseable { + private final Records records; + private final RecordSerde<T> serde; + private final BufferSupplier bufferSupplier; + private final int maxBatchSize; + + private Optional<Iterator<MutableRecordBatch>> nextBatches = Optional.empty(); + private Optional<Batch<T>> nextBatch = Optional.empty(); + // Buffer used to as the backing store for nextBatches if needed + private Optional<ByteBuffer> allocatedBuffer = Optional.empty(); + // Number of bytes from records that read + private int bytesRead = 0; + private boolean isClosed = false; + + public SerdeRecordsIterator( + Records records, + RecordSerde<T> serde, + BufferSupplier bufferSupplier, + int maxBatchSize + ) { + this.records = records; + this.serde = serde; + this.bufferSupplier = bufferSupplier; + this.maxBatchSize = maxBatchSize; + } + + @Override + public boolean hasNext() { + checkIfClosed(); + + if (!nextBatch.isPresent()) { + nextBatch = nextBatch(); + } + + return nextBatch.isPresent(); + } + + @Override + public Batch<T> next() { + if (!hasNext()) { + throw new NoSuchElementException("Batch iterator doesn't have any more elements"); + } + + Batch<T> batch = nextBatch.get(); + nextBatch = Optional.empty(); + + return batch; + } + + @Override + public void close() { + isClosed = true; + allocatedBuffer.ifPresent(bufferSupplier::release); + allocatedBuffer = Optional.empty(); + } + + private void checkIfClosed() { + if (isClosed) { + throw new IllegalStateException("Serde record batch itererator was closed"); + } + } + + private Optional<Iterator<MutableRecordBatch>> nextBatches() { + int recordSize = records.sizeInBytes(); + if (bytesRead < recordSize) { + final MemoryRecords memoryRecords; + if (records instanceof MemoryRecords) { + bytesRead = recordSize; + memoryRecords = (MemoryRecords) records; + } else if (records instanceof FileRecords) { Review comment: I'm wondering if it makes sense to have two implementation for the two types of records we support here. It might make the code easier to follow -- especially if we start have lots of `instanceof` checks down the line (as the complexity of this code grows). -- 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