mridulm commented on a change in pull request #28618:
URL: https://github.com/apache/spark/pull/28618#discussion_r487421724
##########
File path:
core/src/main/java/org/apache/spark/shuffle/sort/io/LocalDiskShuffleExecutorComponents.java
##########
@@ -17,69 +17,64 @@
package org.apache.spark.shuffle.sort.io;
-import java.util.Map;
import java.util.Optional;
import com.google.common.annotations.VisibleForTesting;
+import com.google.common.base.Supplier;
+import com.google.common.base.Suppliers;
import org.apache.spark.SparkConf;
import org.apache.spark.SparkEnv;
+import org.apache.spark.shuffle.IndexShuffleBlockResolver;
import org.apache.spark.shuffle.api.ShuffleExecutorComponents;
import org.apache.spark.shuffle.api.ShuffleMapOutputWriter;
-import org.apache.spark.shuffle.IndexShuffleBlockResolver;
import org.apache.spark.shuffle.api.SingleSpillShuffleMapOutputWriter;
-import org.apache.spark.storage.BlockManager;
public class LocalDiskShuffleExecutorComponents implements
ShuffleExecutorComponents {
private final SparkConf sparkConf;
- private BlockManager blockManager;
- private IndexShuffleBlockResolver blockResolver;
+ private final Supplier<IndexShuffleBlockResolver> blockResolver;
Review comment:
The flow I see (ignoring tests) is :
`MemoizingShuffleDataIO._executor` ->
`LocalDiskShuffleDataIO.initializeShuffleExecutorComponents` ->
`LocalDiskShuffleExecutorComponents ctor`
`_executor` is a `lazy val`, but its construction requires `SparkEnv` to
have been set (and so initialized).
For this path, we will have block manager from `SparkEnv` available to
create `IndexShuffleBlockResolver` in `LocalDiskShuffleExecutorComponents`.
Is there any other path or flow which requires this to be lazy ?
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it in for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
Alternative would be to do this in subsequent or other efforts to complete
the api.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
Review comment:
`unregisterShuffle` currently does not lock ?
##########
File path:
core/src/main/java/org/apache/spark/shuffle/sort/io/LocalDiskShuffleExecutorComponents.java
##########
@@ -17,69 +17,64 @@
package org.apache.spark.shuffle.sort.io;
-import java.util.Map;
import java.util.Optional;
import com.google.common.annotations.VisibleForTesting;
+import com.google.common.base.Supplier;
+import com.google.common.base.Suppliers;
import org.apache.spark.SparkConf;
import org.apache.spark.SparkEnv;
+import org.apache.spark.shuffle.IndexShuffleBlockResolver;
import org.apache.spark.shuffle.api.ShuffleExecutorComponents;
import org.apache.spark.shuffle.api.ShuffleMapOutputWriter;
-import org.apache.spark.shuffle.IndexShuffleBlockResolver;
import org.apache.spark.shuffle.api.SingleSpillShuffleMapOutputWriter;
-import org.apache.spark.storage.BlockManager;
public class LocalDiskShuffleExecutorComponents implements
ShuffleExecutorComponents {
private final SparkConf sparkConf;
- private BlockManager blockManager;
- private IndexShuffleBlockResolver blockResolver;
+ private final Supplier<IndexShuffleBlockResolver> blockResolver;
Review comment:
The flow I see (ignoring tests) is :
`MemoizingShuffleDataIO._executor` ->
`LocalDiskShuffleDataIO.initializeShuffleExecutorComponents` ->
`LocalDiskShuffleExecutorComponents ctor`
`_executor` is a `lazy val`, but its construction requires `SparkEnv` to
have been set (and so initialized).
For this path, we will have block manager from `SparkEnv` available to
create `IndexShuffleBlockResolver` in `LocalDiskShuffleExecutorComponents`.
Is there any other path or flow which requires this to be lazy ?
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it in for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
Alternative would be to do this in subsequent or other efforts to complete
the api.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
Review comment:
`unregisterShuffle` currently does not lock ?
##########
File path:
core/src/main/java/org/apache/spark/shuffle/sort/io/LocalDiskShuffleExecutorComponents.java
##########
@@ -17,69 +17,64 @@
package org.apache.spark.shuffle.sort.io;
-import java.util.Map;
import java.util.Optional;
import com.google.common.annotations.VisibleForTesting;
+import com.google.common.base.Supplier;
+import com.google.common.base.Suppliers;
import org.apache.spark.SparkConf;
import org.apache.spark.SparkEnv;
+import org.apache.spark.shuffle.IndexShuffleBlockResolver;
import org.apache.spark.shuffle.api.ShuffleExecutorComponents;
import org.apache.spark.shuffle.api.ShuffleMapOutputWriter;
-import org.apache.spark.shuffle.IndexShuffleBlockResolver;
import org.apache.spark.shuffle.api.SingleSpillShuffleMapOutputWriter;
-import org.apache.spark.storage.BlockManager;
public class LocalDiskShuffleExecutorComponents implements
ShuffleExecutorComponents {
private final SparkConf sparkConf;
- private BlockManager blockManager;
- private IndexShuffleBlockResolver blockResolver;
+ private final Supplier<IndexShuffleBlockResolver> blockResolver;
Review comment:
The flow I see (ignoring tests) is :
`MemoizingShuffleDataIO._executor` ->
`LocalDiskShuffleDataIO.initializeShuffleExecutorComponents` ->
`LocalDiskShuffleExecutorComponents ctor`
`_executor` is a `lazy val`, but its construction requires `SparkEnv` to
have been set (and so initialized).
For this path, we will have block manager from `SparkEnv` available to
create `IndexShuffleBlockResolver` in `LocalDiskShuffleExecutorComponents`.
Is there any other path or flow which requires this to be lazy ?
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it in for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
Alternative would be to do this in subsequent or other efforts to complete
the api.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
Review comment:
`unregisterShuffle` currently does not lock ?
##########
File path:
core/src/main/java/org/apache/spark/shuffle/sort/io/LocalDiskShuffleExecutorComponents.java
##########
@@ -17,69 +17,64 @@
package org.apache.spark.shuffle.sort.io;
-import java.util.Map;
import java.util.Optional;
import com.google.common.annotations.VisibleForTesting;
+import com.google.common.base.Supplier;
+import com.google.common.base.Suppliers;
import org.apache.spark.SparkConf;
import org.apache.spark.SparkEnv;
+import org.apache.spark.shuffle.IndexShuffleBlockResolver;
import org.apache.spark.shuffle.api.ShuffleExecutorComponents;
import org.apache.spark.shuffle.api.ShuffleMapOutputWriter;
-import org.apache.spark.shuffle.IndexShuffleBlockResolver;
import org.apache.spark.shuffle.api.SingleSpillShuffleMapOutputWriter;
-import org.apache.spark.storage.BlockManager;
public class LocalDiskShuffleExecutorComponents implements
ShuffleExecutorComponents {
private final SparkConf sparkConf;
- private BlockManager blockManager;
- private IndexShuffleBlockResolver blockResolver;
+ private final Supplier<IndexShuffleBlockResolver> blockResolver;
Review comment:
The flow I see (ignoring tests) is :
`MemoizingShuffleDataIO._executor` ->
`LocalDiskShuffleDataIO.initializeShuffleExecutorComponents` ->
`LocalDiskShuffleExecutorComponents ctor`
`_executor` is a `lazy val`, but its construction requires `SparkEnv` to
have been set (and so initialized).
For this path, we will have block manager from `SparkEnv` available to
create `IndexShuffleBlockResolver` in `LocalDiskShuffleExecutorComponents`.
Is there any other path or flow which requires this to be lazy ?
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it in for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
Alternative would be to do this in subsequent or other efforts to complete
the api.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
Review comment:
`unregisterShuffle` currently does not lock ?
##########
File path:
core/src/main/java/org/apache/spark/shuffle/sort/io/LocalDiskShuffleExecutorComponents.java
##########
@@ -17,69 +17,64 @@
package org.apache.spark.shuffle.sort.io;
-import java.util.Map;
import java.util.Optional;
import com.google.common.annotations.VisibleForTesting;
+import com.google.common.base.Supplier;
+import com.google.common.base.Suppliers;
import org.apache.spark.SparkConf;
import org.apache.spark.SparkEnv;
+import org.apache.spark.shuffle.IndexShuffleBlockResolver;
import org.apache.spark.shuffle.api.ShuffleExecutorComponents;
import org.apache.spark.shuffle.api.ShuffleMapOutputWriter;
-import org.apache.spark.shuffle.IndexShuffleBlockResolver;
import org.apache.spark.shuffle.api.SingleSpillShuffleMapOutputWriter;
-import org.apache.spark.storage.BlockManager;
public class LocalDiskShuffleExecutorComponents implements
ShuffleExecutorComponents {
private final SparkConf sparkConf;
- private BlockManager blockManager;
- private IndexShuffleBlockResolver blockResolver;
+ private final Supplier<IndexShuffleBlockResolver> blockResolver;
Review comment:
The flow I see (ignoring tests) is :
`MemoizingShuffleDataIO._executor` ->
`LocalDiskShuffleDataIO.initializeShuffleExecutorComponents` ->
`LocalDiskShuffleExecutorComponents ctor`
`_executor` is a `lazy val`, but its construction requires `SparkEnv` to
have been set (and so initialized).
For this path, we will have block manager from `SparkEnv` available to
create `IndexShuffleBlockResolver` in `LocalDiskShuffleExecutorComponents`.
Is there any other path or flow which requires this to be lazy ?
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it in for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
+ */
+public interface ShuffleOutputTracker {
+
+ /**
+ * Called when a new shuffle stage is going to be run.
+ */
+ void registerShuffle(int shuffleId);
+
Review comment:
This is as a gap in the proposed api - which limits the effectiveness to
leverage it for other shuffle usecases.
+CC @otterc PTAL if without this, shuffle changes can be made for your
patches.
Alternative would be to do this in subsequent or other efforts to complete
the api.
##########
File path:
core/src/main/java/org/apache/spark/shuffle/api/metadata/ShuffleOutputTracker.java
##########
@@ -0,0 +1,56 @@
+/*
+ * 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.spark.shuffle.api.metadata;
+
+/**
+ * :: Private ::
+ *
+ * A plugin that can monitor the storage of shuffle data from map tasks, and
can provide
+ * metadata to shuffle readers to aid their reading of shuffle blocks in
reduce tasks.
+ * <p>
+ * {@link MapOutputMetadata} instances provided from the plugin tree's
implementation of
+ * {@link org.apache.spark.shuffle.api.ShuffleMapOutputWriter} are sent to
+ * <p>
+ * Implementations MUST be thread-safe. Spark will invoke methods in this
module in parallel.
Review comment:
`unregisterShuffle` currently does not lock ?
----------------------------------------------------------------
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:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
