You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ozone.apache.org by GitBox <gi...@apache.org> on 2020/08/24 09:44:02 UTC
[GitHub] [hadoop-ozone] adoroszlai commented on a change in pull request #1336: HDDS-4119. Improve performance of the BufferPool management of Ozone client
adoroszlai commented on a change in pull request #1336:
URL: https://github.com/apache/hadoop-ozone/pull/1336#discussion_r475439432
##########
File path: hadoop-hdds/common/src/main/java/org/apache/hadoop/ozone/common/ChunkBuffer.java
##########
@@ -88,6 +86,12 @@ default ChunkBuffer put(byte[] b) {
return put(ByteBuffer.wrap(b));
}
+ /** Similar to {@link ByteBuffer#put(byte[])}. */
+ default ChunkBuffer put(byte b) {
+ byte[] buf = new byte[1];
+ buf[0] = (byte) b;
+ return put(buf, 0, 1); }
Review comment:
```suggestion
return put(buf, 0, 1);
}
```
##########
File path: hadoop-hdds/client/src/main/java/org/apache/hadoop/hdds/scm/storage/BlockOutputStream.java
##########
@@ -154,6 +159,16 @@ public BlockOutputStream(BlockID blockID,
this.bufferPool = bufferPool;
this.bytesPerChecksum = bytesPerChecksum;
+ //number of buffers used before doing a flush
+ currentBuffer = bufferPool.getCurrentBuffer();
+ currentBufferRemaining =
+ currentBuffer != null ? currentBuffer.remaining() : 0;
Review comment:
Can you please extract these 2 lines to a separate method (and replace other 2 duplicate fragments, too)?
##########
File path: hadoop-hdds/client/src/test/java/org/apache/hadoop/hdds/scm/storage/TestBlockOutputStreamCorrectness.java
##########
@@ -0,0 +1,232 @@
+/*
+ * 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.hadoop.hdds.scm.storage;
+
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Random;
+import java.util.concurrent.CompletableFuture;
+import java.util.concurrent.ExecutionException;
+import java.util.concurrent.TimeoutException;
+import java.util.concurrent.atomic.AtomicInteger;
+
+import org.apache.hadoop.hdds.client.BlockID;
+import org.apache.hadoop.hdds.protocol.DatanodeDetails;
+import org.apache.hadoop.hdds.protocol.MockDatanodeDetails;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.ChecksumType;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.ContainerCommandRequestProto;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.ContainerCommandResponseProto;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.GetCommittedBlockLengthResponseProto;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.PutBlockResponseProto;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.Result;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.Type;
+import org.apache.hadoop.hdds.protocol.proto.HddsProtos.ReplicationFactor;
+import org.apache.hadoop.hdds.protocol.proto.HddsProtos.ReplicationType;
+import org.apache.hadoop.hdds.scm.XceiverClientManager;
+import org.apache.hadoop.hdds.scm.XceiverClientReply;
+import org.apache.hadoop.hdds.scm.XceiverClientSpi;
+import org.apache.hadoop.hdds.scm.pipeline.Pipeline;
+import org.apache.hadoop.hdds.scm.pipeline.Pipeline.Builder;
+import org.apache.hadoop.hdds.scm.pipeline.Pipeline.PipelineState;
+import org.apache.hadoop.hdds.scm.pipeline.PipelineID;
+
+import org.apache.ratis.thirdparty.com.google.protobuf.ByteString;
+import org.jetbrains.annotations.NotNull;
+import org.junit.Assert;
+import org.junit.Test;
+import org.mockito.Mockito;
+
+/**
+ * UNIT test for BlockOutputStream.
+ * <p>
+ * Compares bytes written to the stream and received in the ChunkWriteRequests.
+ */
+public class TestBlockOutputStreamCorrectness {
+
+ private static final long SEED = 18480315L;
Review comment:
ðŸ‡ðŸ‡º
##########
File path: hadoop-hdds/client/src/test/java/org/apache/hadoop/hdds/scm/storage/TestBlockOutputStreamCorrectness.java
##########
@@ -0,0 +1,232 @@
+/*
+ * 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.hadoop.hdds.scm.storage;
+
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Random;
+import java.util.concurrent.CompletableFuture;
+import java.util.concurrent.ExecutionException;
+import java.util.concurrent.TimeoutException;
+import java.util.concurrent.atomic.AtomicInteger;
+
+import org.apache.hadoop.hdds.client.BlockID;
+import org.apache.hadoop.hdds.protocol.DatanodeDetails;
+import org.apache.hadoop.hdds.protocol.MockDatanodeDetails;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.ChecksumType;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.ContainerCommandRequestProto;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.ContainerCommandResponseProto;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.GetCommittedBlockLengthResponseProto;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.PutBlockResponseProto;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.Result;
+import org.apache.hadoop.hdds.protocol.datanode.proto.ContainerProtos.Type;
+import org.apache.hadoop.hdds.protocol.proto.HddsProtos.ReplicationFactor;
+import org.apache.hadoop.hdds.protocol.proto.HddsProtos.ReplicationType;
+import org.apache.hadoop.hdds.scm.XceiverClientManager;
+import org.apache.hadoop.hdds.scm.XceiverClientReply;
+import org.apache.hadoop.hdds.scm.XceiverClientSpi;
+import org.apache.hadoop.hdds.scm.pipeline.Pipeline;
+import org.apache.hadoop.hdds.scm.pipeline.Pipeline.Builder;
+import org.apache.hadoop.hdds.scm.pipeline.Pipeline.PipelineState;
+import org.apache.hadoop.hdds.scm.pipeline.PipelineID;
+
+import org.apache.ratis.thirdparty.com.google.protobuf.ByteString;
+import org.jetbrains.annotations.NotNull;
+import org.junit.Assert;
+import org.junit.Test;
+import org.mockito.Mockito;
+
+/**
+ * UNIT test for BlockOutputStream.
+ * <p>
+ * Compares bytes written to the stream and received in the ChunkWriteRequests.
+ */
+public class TestBlockOutputStreamCorrectness {
+
+ private static final long SEED = 18480315L;
+
+ private int writeUnitSize = 1;
+
+ @Test
+ public void test() throws IOException {
+
+ final BufferPool bufferPool = new BufferPool(4 * 1024 * 1024, 32 / 4);
+
+ for (int block = 0; block < 10; block++) {
+ BlockOutputStream outputStream =
+ createBlockOutputStream(bufferPool);
+
+ Random random = new Random(SEED);
+
+ int max = 256 * 1024 * 1024 / writeUnitSize;
+
+ byte[] writeBuffer = new byte[writeUnitSize];
+ for (int t = 0; t < max; t++) {
+ if (writeUnitSize > 1) {
+ for (int i = 0; i < writeBuffer.length; i++) {
+ writeBuffer[i] = (byte) random.nextInt();
+ }
+ outputStream.write(writeBuffer, 0, writeBuffer.length);
+ } else {
+ outputStream.write((byte) random.nextInt());
+ }
+ }
+ outputStream.close();
+ }
+ }
+
+ @NotNull
+ private BlockOutputStream createBlockOutputStream(BufferPool bufferPool)
+ throws IOException {
+ List<DatanodeDetails> nodes = new ArrayList<>();
+ nodes.add(MockDatanodeDetails.randomDatanodeDetails());
+ nodes.add(MockDatanodeDetails.randomDatanodeDetails());
+ nodes.add(MockDatanodeDetails.randomDatanodeDetails());
+
+ final Pipeline pipeline = new Builder()
+ .setFactor(ReplicationFactor.THREE)
+ .setType(ReplicationType.RATIS)
+ .setState(PipelineState.OPEN)
+ .setId(PipelineID.randomId())
+ .setNodes(nodes)
+ .build();
Review comment:
Can we move this logic to `MockPipeline` for reuse?
##########
File path: hadoop-hdds/common/src/main/java/org/apache/hadoop/ozone/common/ChunkBuffer.java
##########
@@ -44,9 +45,6 @@ static ChunkBuffer allocate(int capacity) {
* When increment <= 0, entire buffer is allocated in the beginning.
*/
static ChunkBuffer allocate(int capacity, int increment) {
- if (increment > 0 && increment < capacity) {
- return new IncrementalChunkBuffer(capacity, increment, false);
Review comment:
Can you please also change `TestChunkBuffer#runTestIncrementalChunkBuffer` to explicitly create `IncrementalChunkBuffer`? Currently it uses this factory method, and so with this patch it really tests `ChunkBufferImplWithByteBuffer`.
##########
File path: hadoop-hdds/client/src/main/java/org/apache/hadoop/hdds/scm/storage/BlockOutputStream.java
##########
@@ -209,9 +224,16 @@ public IOException getIoException() {
@Override
public void write(int b) throws IOException {
checkOpen();
- byte[] buf = new byte[1];
- buf[0] = (byte) b;
- write(buf, 0, 1);
+ if (currentBufferRemaining == 0) {
+ allocateNewBuffer();
+ }
+ currentBuffer.put((byte) b);
+ currentBufferRemaining--;
+ if (currentBufferRemaining == 0) {
+ writeChunk(currentBuffer);
+ }
Review comment:
For consistency, I think we should move the condition to a new method `writeChunkIfNeeded`.
##########
File path: hadoop-hdds/client/src/main/java/org/apache/hadoop/hdds/scm/storage/BlockOutputStream.java
##########
@@ -209,9 +224,16 @@ public IOException getIoException() {
@Override
public void write(int b) throws IOException {
checkOpen();
- byte[] buf = new byte[1];
- buf[0] = (byte) b;
- write(buf, 0, 1);
+ if (currentBufferRemaining == 0) {
+ allocateNewBuffer();
+ }
Review comment:
For consistency, I think we should move the condition to `allocateNewBuffer` and rename it to `allocateNewBufferIfNeeded`.
##########
File path: hadoop-hdds/client/src/main/java/org/apache/hadoop/hdds/scm/storage/BlockOutputStream.java
##########
@@ -481,7 +516,7 @@ private void handleFlush(boolean close)
checkOpen();
// flush the last chunk data residing on the currentBuffer
if (totalDataFlushedLength < writtenDataLength) {
- final ChunkBuffer currentBuffer = bufferPool.getCurrentBuffer();
+ currentBuffer = bufferPool.getCurrentBuffer();
Review comment:
Should we also update `currentBufferRemaining`?
----------------------------------------------------------------
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:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: ozone-issues-help@hadoop.apache.org