[
https://issues.apache.org/jira/browse/FLINK-5041?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15725659#comment-15725659
]
ASF GitHub Bot commented on FLINK-5041:
---------------------------------------
Github user StephanEwen commented on a diff in the pull request:
https://github.com/apache/flink/pull/2781#discussion_r91058148
--- Diff:
flink-runtime/src/main/java/org/apache/flink/migration/runtime/checkpoint/savepoint/SavepointV0Serializer.java
---
@@ -0,0 +1,390 @@
+/*
+ * 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.flink.migration.runtime.checkpoint.savepoint;
+
+import org.apache.flink.annotation.VisibleForTesting;
+import org.apache.flink.core.fs.Path;
+import org.apache.flink.migration.runtime.checkpoint.KeyGroupState;
+import org.apache.flink.migration.runtime.checkpoint.SubtaskState;
+import org.apache.flink.migration.runtime.checkpoint.TaskState;
+import org.apache.flink.migration.runtime.state.KvStateSnapshot;
+import org.apache.flink.migration.runtime.state.StateHandle;
+import
org.apache.flink.migration.runtime.state.filesystem.AbstractFileStateHandle;
+import
org.apache.flink.migration.runtime.state.memory.SerializedStateHandle;
+import org.apache.flink.migration.state.MigrationKeyGroupStateHandle;
+import org.apache.flink.migration.state.MigrationStreamStateHandle;
+import org.apache.flink.migration.streaming.runtime.tasks.StreamTaskState;
+import
org.apache.flink.migration.streaming.runtime.tasks.StreamTaskStateList;
+import org.apache.flink.migration.util.SerializedValue;
+import org.apache.flink.runtime.checkpoint.savepoint.SavepointSerializer;
+import org.apache.flink.runtime.checkpoint.savepoint.SavepointV1;
+import org.apache.flink.runtime.jobgraph.JobVertexID;
+import org.apache.flink.runtime.state.ChainedStateHandle;
+import org.apache.flink.runtime.state.CheckpointStreamFactory;
+import org.apache.flink.runtime.state.KeyGroupRangeOffsets;
+import org.apache.flink.runtime.state.KeyGroupsStateHandle;
+import org.apache.flink.runtime.state.MultiStreamStateHandle;
+import org.apache.flink.runtime.state.OperatorStateHandle;
+import org.apache.flink.runtime.state.StreamStateHandle;
+import org.apache.flink.runtime.state.filesystem.FileStateHandle;
+import org.apache.flink.runtime.state.memory.ByteStreamStateHandle;
+import org.apache.flink.runtime.state.memory.MemCheckpointStreamFactory;
+import org.apache.flink.util.InstantiationUtil;
+
+import java.io.DataInputStream;
+import java.io.DataOutputStream;
+import java.io.IOException;
+import java.io.Serializable;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Collection;
+import java.util.HashMap;
+import java.util.List;
+
+/**
+ * <p>
+ * <p>In contrast to previous savepoint versions, this serializer makes
sure
+ * that no default Java serialization is used for serialization.
Therefore, we
+ * don't rely on any involved Java classes to stay the same.
+ */
+public class SavepointV0Serializer implements
SavepointSerializer<SavepointV1> {
+
+ public static final SavepointV0Serializer INSTANCE = new
SavepointV0Serializer();
+ private static final StreamStateHandle SIGNAL_0 = new
ByteStreamStateHandle("SIGNAL_0", new byte[]{0});
+ private static final StreamStateHandle SIGNAL_1 = new
ByteStreamStateHandle("SIGNAL_1", new byte[]{1});
+
+ private static final int MAX_SIZE = 4 * 1024 * 1024;
+
+ private ClassLoader userClassLoader;
+ private long checkpointID;
+
+ private SavepointV0Serializer() {
+ }
+
+
+ @Override
+ public void serialize(SavepointV1 savepoint, DataOutputStream dos)
throws IOException {
+ throw new UnsupportedOperationException("This serializer is
read-only and only exists for backwards compatibility");
+ }
+
+ @Override
+ public SavepointV1 deserialize(DataInputStream dis, ClassLoader
userClassLoader) throws IOException {
+
+ this.checkpointID = dis.readLong();
+ this.userClassLoader = userClassLoader;
+
+ // Task states
+ int numTaskStates = dis.readInt();
+ List<TaskState> taskStates = new ArrayList<>(numTaskStates);
+
+ for (int i = 0; i < numTaskStates; i++) {
+ JobVertexID jobVertexId = new
JobVertexID(dis.readLong(), dis.readLong());
+ int parallelism = dis.readInt();
+
+ // Add task state
+ TaskState taskState = new TaskState(jobVertexId,
parallelism);
+ taskStates.add(taskState);
+
+ // Sub task states
+ int numSubTaskStates = dis.readInt();
+ for (int j = 0; j < numSubTaskStates; j++) {
+ int subtaskIndex = dis.readInt();
+
+ int length = dis.readInt();
+
+ SerializedValue<StateHandle<?>> serializedValue;
+ if (length == -1) {
+ serializedValue = new
SerializedValue<>(null);
+ } else {
+ byte[] serializedData = new
byte[length];
+ dis.readFully(serializedData, 0,
length);
+ serializedValue =
SerializedValue.fromBytes(serializedData);
+ }
+
+ long stateSize = dis.readLong();
+ long duration = dis.readLong();
+
+ SubtaskState subtaskState = new SubtaskState(
+ serializedValue,
+ stateSize,
+ duration);
+
+ taskState.putState(subtaskIndex, subtaskState);
+ }
+
+ // Key group states
+ int numKvStates = dis.readInt();
+ for (int j = 0; j < numKvStates; j++) {
+ int keyGroupIndex = dis.readInt();
+
+ int length = dis.readInt();
+
+ SerializedValue<StateHandle<?>> serializedValue;
+ if (length == -1) {
+ serializedValue = new
SerializedValue<>(null);
+ } else {
+ byte[] serializedData = new
byte[length];
+ dis.readFully(serializedData, 0,
length);
+ serializedValue =
SerializedValue.fromBytes(serializedData);
+ }
+
+ long stateSize = dis.readLong();
+ long duration = dis.readLong();
+
+ KeyGroupState keyGroupState = new KeyGroupState(
+ serializedValue,
+ stateSize,
+ duration);
+
+ taskState.putKvState(keyGroupIndex,
keyGroupState);
+ }
+ }
+
+ try {
+ return convertSavepoint(taskStates);
+ } catch (Exception e) {
+ throw new IOException(e);
+ }
+ }
+
+ private SavepointV1 convertSavepoint(List<TaskState> taskStates) throws
Exception {
+
+ List<org.apache.flink.runtime.checkpoint.TaskState>
newTaskStates = new ArrayList<>(taskStates.size());
+
+ for (TaskState taskState : taskStates) {
+ newTaskStates.add(convertTaskState(taskState));
+ }
+
+ return new SavepointV1(checkpointID, newTaskStates);
+ }
+
+ private org.apache.flink.runtime.checkpoint.TaskState
convertTaskState(TaskState taskState) throws Exception {
+
+ JobVertexID jobVertexID = taskState.getJobVertexID();
+ int parallelism = taskState.getParallelism();
+ int chainLength = determineOperatorChainLength(taskState);
+
+ org.apache.flink.runtime.checkpoint.TaskState newTaskState =
+ new
org.apache.flink.runtime.checkpoint.TaskState(
+ jobVertexID,
+ parallelism,
+ parallelism,
+ chainLength);
+
+ if (chainLength > 0) {
+
+ int parallelInstanceIdx = 0;
+ Collection<SubtaskState> subtaskStates =
taskState.getStates();
--- End diff --
This looks like it iterates in undefined order over the previous parallel
subtasks, but maps them to the new parallel subtask indexes based on that
order. Probably safer to iterate over the entry set and reuse the old subtask
index.
> Implement savepoint backwards compatibility 1.1 -> 1.2
> ------------------------------------------------------
>
> Key: FLINK-5041
> URL: https://issues.apache.org/jira/browse/FLINK-5041
> Project: Flink
> Issue Type: New Feature
> Components: State Backends, Checkpointing
> Affects Versions: 1.2.0
> Reporter: Stefan Richter
> Assignee: Stefan Richter
>
> This issue tracks the implementation of backwards compatibility between Flink
> 1.1 and 1.2 releases.
> This task subsumes:
> - Converting old savepoints to new savepoints, including a conversion of
> state handles to their new replacement.
> - Converting keyed state from old backend implementations to their new
> counterparts.
> - Converting operator and function state for all changed operators.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
