Many resources are needed to download a project. Please understand that we have to compensate our server costs. Thank you in advance. Project price only 1 $
You can buy this project and download/modify it how often you want.
/*
* 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.runtime.executiongraph.failover;
import org.apache.flink.runtime.executiongraph.Execution;
import org.apache.flink.runtime.executiongraph.ExecutionEdge;
import org.apache.flink.runtime.executiongraph.ExecutionGraph;
import org.apache.flink.runtime.executiongraph.ExecutionVertex;
import org.apache.flink.runtime.executiongraph.IntermediateResult;
import org.apache.flink.runtime.executiongraph.IntermediateResultPartition;
import org.apache.flink.runtime.io.network.partition.DataConsumptionException;
import org.apache.flink.runtime.io.network.partition.ResultPartitionID;
import org.apache.flink.util.ExceptionUtils;
import org.apache.flink.util.FlinkException;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import java.util.ArrayDeque;
import java.util.ArrayList;
import java.util.IdentityHashMap;
import java.util.List;
import java.util.Optional;
import java.util.Queue;
import java.util.Set;
import java.util.concurrent.Executor;
/**
* A strict version of {@link RestartPipelinedRegionStrategy} which ensures data consistency
* even if there are non-deterministic shuffles(rebalance, rescale..).
*/
public class StrictRestartPipelinedRegionStrategy extends RestartPipelinedRegionStrategy {
/** The log object used for debugging. */
private static final Logger LOG = LoggerFactory.getLogger(StrictRestartPipelinedRegionStrategy.class);
/**
* Creates a new failover strategy to restart pipelined regions that works on the given
* execution graph and uses the execution graph's future executor to call restart actions.
*
* @param executionGraph The execution graph on which this FailoverStrategy will work
* @param regionFailLimit The max number a region can fail
*/
public StrictRestartPipelinedRegionStrategy(ExecutionGraph executionGraph, int regionFailLimit) {
this(executionGraph, executionGraph.getFutureExecutor(), regionFailLimit);
}
/**
* Creates a new failover strategy to restart pipelined regions that works on the given
* execution graph and uses the given executor to call restart actions.
*
* @param executionGraph The execution graph on which this FailoverStrategy will work
* @param executor The executor used for future actions
* @param regionFailLimit The max number a region can fail
*/
public StrictRestartPipelinedRegionStrategy(ExecutionGraph executionGraph, Executor executor, int regionFailLimit) {
super(executionGraph, executor, regionFailLimit);
}
// ------------------------------------------------------------------------
// failover implementation
// ------------------------------------------------------------------------
@Override
public void onTaskFailure(Execution taskExecution, Throwable cause) {
final ExecutionVertex ev = taskExecution.getVertex();
final FailoverRegion failoverRegion = vertexToRegion.get(ev);
if (failoverRegion == null) {
executionGraph.failGlobal(new FlinkException(
"Can not find a failover region for the execution " + ev.getTaskNameWithSubtaskIndex(), cause));
return;
}
List sortedRegionsToRestart = sortRegionsTopologically(
getRegionsToRestart(taskExecution.getVertex(), cause));
// Cancel and restart the region of the target vertex
LOG.info("Recovering task failure for {} #{} ({}) via restarting {} failover regions",
ev.getTaskNameWithSubtaskIndex(),
taskExecution.getAttemptNumber(),
taskExecution.getAttemptId(),
sortedRegionsToRestart.size());
for (FailoverRegion regionToRestart : sortedRegionsToRestart) {
regionToRestart.onExecutionFail(taskExecution.getGlobalModVersion(), cause);
}
}
/**
* All 'involved' regions are proposed to be restarted.
* The 'involved' regions are calculated with rules below:
* 1. The region containing the failed task is always involved
* 2. If an input result partition of an involved region is not available, i.e. Missing or Corrupted,
* the region containing the partition producer task is involved
* 3. If a region is involved, all of its consumer regions are involved
*/
private Set getRegionsToRestart(ExecutionVertex failedVertex, Throwable cause) {
IdentityHashMap regionsToRestart = new IdentityHashMap<>();
IdentityHashMap visitedRegions = new IdentityHashMap<>();
Queue regionsToVisit = new ArrayDeque<>();
FailoverRegion rootFailedRegion = getRootFailedRegion(failedVertex, cause);
visitedRegions.put(rootFailedRegion, null);
regionsToVisit.add(rootFailedRegion);
// start from the failed region to visit all involved regions
while (!regionsToVisit.isEmpty()) {
FailoverRegion regionToRestart = regionsToVisit.poll();
// an involved region should be restarted
regionsToRestart.put(regionToRestart, null);
// all consumer regions of an involved region should be involved
for (ExecutionVertex vertex : regionToRestart.getAllExecutionVertices()) {
for (IntermediateResultPartition resultPartition : vertex.getProducedPartitions().values()) {
for (List edges : resultPartition.getConsumers()) {
for (ExecutionEdge edge : edges) {
ExecutionVertex consumerVertex = edge.getTarget();
FailoverRegion consumerRegion = vertexToRegion.get(consumerVertex);
if (!visitedRegions.containsKey(consumerRegion)) {
visitedRegions.put(consumerRegion, null);
regionsToVisit.add(consumerRegion);
}
}
}
}
}
}
return regionsToRestart.keySet();
}
private FailoverRegion getRootFailedRegion(ExecutionVertex failedVertex, Throwable cause) {
Optional dataConsumptionException = ExceptionUtils.findThrowable(
cause, DataConsumptionException.class);
if (dataConsumptionException.isPresent()) {
LOG.info("Try restarting producer of {} due to DataConsumptionException", failedVertex);
ResultPartitionID predecessorResultPartition = dataConsumptionException.get().getResultPartitionId();
Execution producer = executionGraph.getRegisteredExecutions().get(predecessorResultPartition.getProducerId());
if (producer == null) {
// If the producer has finished, it is removed from registeredExecutions and we need to locate it via the
// ResultPartitionID and the down-stream task.
for (IntermediateResult intermediateResult : failedVertex.getJobVertex().getInputs()) {
IntermediateResultPartition resultPartition = intermediateResult.getPartitionOrNullById(
predecessorResultPartition.getPartitionId());
if (resultPartition != null) {
Execution producerVertexCurrentAttempt = resultPartition.getProducer().getCurrentExecutionAttempt();
if (producerVertexCurrentAttempt.getAttemptId().equals(predecessorResultPartition.getProducerId())) {
producer = producerVertexCurrentAttempt;
} else {
LOG.warn("partition {} has already been disposed, skip restarting the producer.",
predecessorResultPartition);
}
break;
}
}
}
return vertexToRegion.get(producer.getVertex());
} else {
return vertexToRegion.get(failedVertex);
}
}
private List sortRegionsTopologically(Set regions) {
final List regionsSorted = new ArrayList<>();
for (FailoverRegion region : sortedRegions) {
if (regions.contains(region)) {
regionsSorted.add(region);
}
}
return regionsSorted;
}
@Override
public String getStrategyName() {
return "Strict Pipelined Region Failover";
}
// ------------------------------------------------------------------------
// factory
// ------------------------------------------------------------------------
/**
* Factory that instantiates the StrictRestartPipelinedRegionStrategy.
*/
public static class Factory implements FailoverStrategy.Factory {
private int regionFailLimit = 100;
@Override
public FailoverStrategy create(ExecutionGraph executionGraph) {
return new StrictRestartPipelinedRegionStrategy(executionGraph, regionFailLimit);
}
public void setRegionFailLimit(int regionFailLimit) {
this.regionFailLimit = regionFailLimit;
}
}
}