GJL commented on a change in pull request #8922: [FLINK-12876][runtime] Add an adapter of region failover NG for legacy scheduler URL: https://github.com/apache/flink/pull/8922#discussion_r298538576
########## File path: flink-runtime/src/test/java/org/apache/flink/runtime/executiongraph/AdaptedRestartPipelinedRegionStrategyNGFailoverTest.java ########## @@ -0,0 +1,513 @@ +/* + * 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; + +import org.apache.flink.api.common.JobID; +import org.apache.flink.api.common.time.Time; +import org.apache.flink.runtime.akka.AkkaUtils; +import org.apache.flink.runtime.clusterframework.types.SlotProfile; +import org.apache.flink.runtime.concurrent.ComponentMainThreadExecutor; +import org.apache.flink.runtime.concurrent.FutureUtils; +import org.apache.flink.runtime.concurrent.ManuallyTriggeredScheduledExecutor; +import org.apache.flink.runtime.execution.ExecutionState; +import org.apache.flink.runtime.executiongraph.failover.AdaptedRestartPipelinedRegionStrategyNG; +import org.apache.flink.runtime.executiongraph.restart.InfiniteDelayRestartStrategy; +import org.apache.flink.runtime.executiongraph.restart.NoRestartStrategy; +import org.apache.flink.runtime.executiongraph.restart.RestartStrategy; +import org.apache.flink.runtime.executiongraph.utils.SimpleSlotProvider; +import org.apache.flink.runtime.instance.SlotSharingGroupId; +import org.apache.flink.runtime.io.network.partition.ResultPartitionID; +import org.apache.flink.runtime.io.network.partition.ResultPartitionType; +import org.apache.flink.runtime.io.network.partition.consumer.PartitionConnectionException; +import org.apache.flink.runtime.jobgraph.DistributionPattern; +import org.apache.flink.runtime.jobgraph.JobGraph; +import org.apache.flink.runtime.jobgraph.JobStatus; +import org.apache.flink.runtime.jobgraph.JobVertex; +import org.apache.flink.runtime.jobgraph.ScheduleMode; +import org.apache.flink.runtime.jobgraph.tasks.AbstractInvokable; +import org.apache.flink.runtime.jobmanager.scheduler.ScheduledUnit; +import org.apache.flink.runtime.jobmaster.LogicalSlot; +import org.apache.flink.runtime.jobmaster.SlotRequestId; +import org.apache.flink.runtime.jobmaster.slotpool.SlotProvider; +import org.apache.flink.runtime.scheduler.strategy.ExecutionVertexID; +import org.apache.flink.runtime.testingUtils.TestingUtils; +import org.apache.flink.util.FlinkException; +import org.apache.flink.util.TestLogger; + +import org.junit.Before; +import org.junit.Test; + +import javax.annotation.Nullable; + +import java.util.Arrays; +import java.util.Iterator; +import java.util.List; +import java.util.Set; +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.TimeoutException; + +import static org.apache.flink.util.Preconditions.checkNotNull; +import static org.hamcrest.Matchers.containsInAnyOrder; +import static org.junit.Assert.assertEquals; +import static org.junit.Assert.assertFalse; +import static org.junit.Assert.assertNotEquals; +import static org.junit.Assert.assertThat; + +/** + * Tests for {@link AdaptedRestartPipelinedRegionStrategyNG} failover handling. + */ +public class AdaptedRestartPipelinedRegionStrategyNGFailoverTest extends TestLogger { + + private static final JobID TEST_JOB_ID = new JobID(); + + private ComponentMainThreadExecutor componentMainThreadExecutor; + + private FailingSlotProviderDecorator slotProvider; + + private ManuallyTriggeredScheduledExecutor manualMainThreadExecutor; + + @Before + public void setUp() { + manualMainThreadExecutor = new ManuallyTriggeredScheduledExecutor(); + componentMainThreadExecutor = new ScheduledExecutorToComponentMainThreadExecutorAdapter(manualMainThreadExecutor, Thread.currentThread()); + slotProvider = new FailingSlotProviderDecorator(new SimpleSlotProvider(TEST_JOB_ID, 14)); + } + + /** + * Tests for region failover for job in EAGER mode. + * This applies to streaming job, with no BLOCKING edge. + * <pre> + * (v11) ---> (v21) + * + * (v12) ---> (v22) + * + * ^ + * | + * (pipelined) + * </pre> + */ + @Test + public void testRegionFailoverInEagerMode() throws Exception { + // create a streaming job graph with EAGER schedule mode + final JobGraph jobGraph = createStreamingJobGraph(); + final ExecutionGraph eg = createExecutionGraph(jobGraph); + + final Iterator<ExecutionVertex> vertexIterator = eg.getAllExecutionVertices().iterator(); + final ExecutionVertex ev11 = vertexIterator.next(); + final ExecutionVertex ev12 = vertexIterator.next(); + final ExecutionVertex ev21 = vertexIterator.next(); + final ExecutionVertex ev22 = vertexIterator.next(); + + // trigger task failure of ev11 + // vertices { ev11, ev21 } should be affected + ev11.getCurrentExecutionAttempt().fail(new Exception("Test Exception")); + manualMainThreadExecutor.triggerAll(); + + // verify vertex states and complete cancellation + assertVertexInState(ExecutionState.FAILED, ev11); + assertVertexInState(ExecutionState.DEPLOYING, ev12); + assertVertexInState(ExecutionState.CANCELING, ev21); + assertVertexInState(ExecutionState.DEPLOYING, ev22); + ev21.getCurrentExecutionAttempt().completeCancelling(); + manualMainThreadExecutor.triggerAll(); + + // verify vertex states + // in eager mode, all affected vertices should be scheduled in failover + assertVertexInState(ExecutionState.DEPLOYING, ev11); + assertVertexInState(ExecutionState.DEPLOYING, ev12); + assertVertexInState(ExecutionState.DEPLOYING, ev21); + assertVertexInState(ExecutionState.DEPLOYING, ev22); + + // verify attempt number + assertEquals(1, ev11.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(0, ev12.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev21.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(0, ev22.getCurrentExecutionAttempt().getAttemptNumber()); + } + + /** + * Tests for scenario where a task fails for its own error, in which case the + * region containing the failed task and its consumer regions should be restarted. + * <pre> + * (v11) -+-> (v21) + * x + * (v12) -+-> (v22) + * + * ^ + * | + * (blocking) + * </pre> + */ + @Test + public void testRegionFailoverForRegionInternalErrorsInLazyMode() throws Exception { + final JobGraph jobGraph = createBatchJobGraph(); + final ExecutionGraph eg = createExecutionGraph(jobGraph); + + final Iterator<ExecutionVertex> vertexIterator = eg.getAllExecutionVertices().iterator(); + final ExecutionVertex ev11 = vertexIterator.next(); + final ExecutionVertex ev12 = vertexIterator.next(); + final ExecutionVertex ev21 = vertexIterator.next(); + final ExecutionVertex ev22 = vertexIterator.next(); + + // trigger task failure of ev11 + // regions {ev11}, {ev21}, {ev22} should be affected + ev11.getCurrentExecutionAttempt().fail(new Exception("Test Exception")); + manualMainThreadExecutor.triggerAll(); + + // verify vertex states + // only vertices with consumable inputs can be scheduled + assertVertexInState(ExecutionState.DEPLOYING, ev11); + assertVertexInState(ExecutionState.DEPLOYING, ev12); + assertVertexInState(ExecutionState.CREATED, ev21); + assertVertexInState(ExecutionState.CREATED, ev22); + + // verify attempt number + assertEquals(1, ev11.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(0, ev12.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev21.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev22.getCurrentExecutionAttempt().getAttemptNumber()); + } + + /** + * Tests that the failure is properly propagated to underlying strategy + * to calculate tasks to restart. + * <pre> + * (v11) -+-> (v21) + * x + * (v12) -+-> (v22) + * + * ^ + * | + * (blocking) + * </pre> + */ + @Test + public void testFailurePropagationToUnderlyingStrategy() throws Exception { + final JobGraph jobGraph = createBatchJobGraph(); + final ExecutionGraph eg = createExecutionGraph(jobGraph); + + final TestAdaptedRestartPipelinedRegionStrategyNG failoverStrategy = + (TestAdaptedRestartPipelinedRegionStrategyNG) eg.getFailoverStrategy(); + + final Iterator<ExecutionVertex> vertexIterator = eg.getAllExecutionVertices().iterator(); + final ExecutionVertex ev11 = vertexIterator.next(); + final ExecutionVertex ev12 = vertexIterator.next(); + final ExecutionVertex ev21 = vertexIterator.next(); + + // trigger downstream regions to schedule + componentMainThreadExecutor.execute(() -> { + // finish upstream regions to trigger scheduling of downstream regions + ev11.getCurrentExecutionAttempt().markFinished(); + ev12.getCurrentExecutionAttempt().markFinished(); + }); + + // trigger task failure of ev21 on consuming data from ev11 + Exception taskFailureCause = new PartitionConnectionException( + new ResultPartitionID( + ev11.getProducedPartitions().keySet().iterator().next(), + ev11.getCurrentExecutionAttempt().getAttemptId()), + new Exception("Test failure")); + ev21.getCurrentExecutionAttempt().fail(taskFailureCause); + manualMainThreadExecutor.triggerAll(); + + assertThat(failoverStrategy.getLastTasksToCancel(), containsInAnyOrder( + new ExecutionVertexID(ev11.getJobvertexId(), 0), + new ExecutionVertexID(ev21.getJobvertexId(), 0), + new ExecutionVertexID(ev21.getJobvertexId(), 1))); + } + + /** + * Tests that the vertex partition state is reset property during failover. + * <pre> + * (v11) -+-> (v21) + * x + * (v12) -+-> (v22) + * + * ^ + * | + * (blocking) + * </pre> + */ + @Test + public void testStatusResettingOnRegionFailover() throws Exception { + // create a batch job graph with LAZY_FROM_SOURCES schedule mode + final JobGraph jobGraph = createBatchJobGraph(); + final ExecutionGraph eg = createExecutionGraph(jobGraph); + final AdaptedRestartPipelinedRegionStrategyNG strategy = + (AdaptedRestartPipelinedRegionStrategyNG) eg.getFailoverStrategy(); + + final Iterator<ExecutionVertex> vertexIterator = eg.getAllExecutionVertices().iterator(); + final ExecutionVertex ev11 = vertexIterator.next(); + final ExecutionVertex ev12 = vertexIterator.next(); + final ExecutionVertex ev21 = vertexIterator.next(); + final ExecutionVertex ev22 = vertexIterator.next(); + + // finish upstream regions so that all the blocking result partitions should be finished and consumable + + // finish upstream regions to trigger scheduling of downstream regions + ev11.getCurrentExecutionAttempt().markFinished(); + ev12.getCurrentExecutionAttempt().markFinished(); + + // force FINISHED ev11 to fail to reset its partition + strategy.onTaskFailure(ev11.getCurrentExecutionAttempt(), new FlinkException("Fail for testing")); + + // complete cancelling + ev21.getCurrentExecutionAttempt().completeCancelling(); + ev22.getCurrentExecutionAttempt().completeCancelling(); + + manualMainThreadExecutor.triggerAll(); + + // the blocking result partitions should be no more consumable after the resetting + IntermediateResult result = ev11.getJobVertex().getProducedDataSets()[0]; + assertFalse(result.areAllPartitionsFinished()); + assertFalse(result.getPartitions()[0].isConsumable()); + assertFalse(result.getPartitions()[1].isConsumable()); + } + + /** + * Tests that when a task fail, and restart strategy doesn't support restarting, the job will go to failed. + */ + @Test + public void testNoRestart() throws Exception { + final JobGraph jobGraph = createBatchJobGraph(); + final NoRestartStrategy restartStrategy = new NoRestartStrategy(); + final ExecutionGraph eg = createExecutionGraph(jobGraph, restartStrategy); + + final ExecutionVertex ev = eg.getAllExecutionVertices().iterator().next(); + + ev.fail(new Exception("Test Exception")); + + for (ExecutionVertex evs : eg.getAllExecutionVertices()) { + evs.getCurrentExecutionAttempt().completeCancelling(); + } + + manualMainThreadExecutor.triggerAll(); + + assertEquals(JobStatus.FAILED, eg.getState()); + } + + @Test + public void testFailGlobalIfErrorOnRestartingTasks() throws Exception { + final JobGraph jobGraph = createStreamingJobGraph(); + final ExecutionGraph eg = createExecutionGraph(jobGraph); + + final Iterator<ExecutionVertex> vertexIterator = eg.getAllExecutionVertices().iterator(); + final ExecutionVertex ev11 = vertexIterator.next(); + final ExecutionVertex ev12 = vertexIterator.next(); + final ExecutionVertex ev21 = vertexIterator.next(); + final ExecutionVertex ev22 = vertexIterator.next(); + + final long globalModVersionBeforeFailure = eg.getGlobalModVersion(); + + slotProvider.setFailSlotAllocation(true); + ev11.fail(new Exception("Test Exception")); + completeCancelling(ev11, ev12, ev21, ev22); + + manualMainThreadExecutor.triggerAll(); + + final long globalModVersionAfterFailure = eg.getGlobalModVersion(); + + assertNotEquals(globalModVersionBeforeFailure, globalModVersionAfterFailure); + } + + // ------------------------------- Test Utils ----------------------------------------- + + /** + * Creating job graph as below (execution view). + * It's a representative of streaming job. + * <pre> + * (v11) -+-> (v21) + * + * (v12) -+-> (v22) + * + * ^ + * | + * (pipelined) + * </pre> + * 2 regions. Each has 2 pipelined connected vertices. + */ + private JobGraph createStreamingJobGraph() { + final JobVertex v1 = new JobVertex("vertex1"); + final JobVertex v2 = new JobVertex("vertex2"); + + v1.setParallelism(2); + v2.setParallelism(2); + + v1.setInvokableClass(AbstractInvokable.class); + v2.setInvokableClass(AbstractInvokable.class); + + v2.connectNewDataSetAsInput(v1, DistributionPattern.POINTWISE, ResultPartitionType.PIPELINED); + + final JobGraph jobGraph = new JobGraph(TEST_JOB_ID, "Testjob" ,v1, v2); + jobGraph.setScheduleMode(ScheduleMode.EAGER); + + return jobGraph; + } + + /** + * Creating job graph as below (execution view). + * It's a representative of batch job. + * <pre> + * (v11) -+-> (v21) + * x + * (v12) -+-> (v22) + * + * ^ + * | + * (blocking) + * </pre> + * 4 regions. Each consists of one individual vertex. + */ + private JobGraph createBatchJobGraph() { + final JobVertex v1 = new JobVertex("vertex1"); + final JobVertex v2 = new JobVertex("vertex2"); + + v1.setParallelism(2); + v2.setParallelism(2); + + v1.setInvokableClass(AbstractInvokable.class); + v2.setInvokableClass(AbstractInvokable.class); + + v2.connectNewDataSetAsInput(v1, DistributionPattern.ALL_TO_ALL, ResultPartitionType.BLOCKING); + + final JobGraph jobGraph = new JobGraph(v1, v2); + jobGraph.setScheduleMode(ScheduleMode.LAZY_FROM_SOURCES); + + return jobGraph; + } + + private ExecutionGraph createExecutionGraph(final JobGraph jobGraph) throws Exception { + return createExecutionGraph(jobGraph, new InfiniteDelayRestartStrategy(10)); + } + + private ExecutionGraph createExecutionGraph( + final JobGraph jobGraph, + final RestartStrategy restartStrategy) throws Exception { + + final ExecutionGraph eg = new ExecutionGraph( + new DummyJobInformation( + jobGraph.getJobID(), + jobGraph.getName()), + TestingUtils.defaultExecutor(), + TestingUtils.defaultExecutor(), + AkkaUtils.getDefaultTimeout(), + restartStrategy, + TestAdaptedRestartPipelinedRegionStrategyNG::new, + slotProvider); + eg.attachJobGraph(jobGraph.getVerticesSortedTopologicallyFromSources()); + + eg.setScheduleMode(jobGraph.getScheduleMode()); + + eg.start(componentMainThreadExecutor); + eg.scheduleForExecution(); + manualMainThreadExecutor.triggerAll(); + + return eg; + } + + private static void assertVertexInState(final ExecutionState state, final ExecutionVertex vertex) { + assertEquals(state, vertex.getExecutionState()); + } + + private static void completeCancelling(ExecutionVertex... executionVertices) { + for (final ExecutionVertex executionVertex : executionVertices) { + executionVertex.getCurrentExecutionAttempt().completeCancelling(); + } + } + + /** + * Test implementation of the {@link AdaptedRestartPipelinedRegionStrategyNG} that makes it possible + * to control when the failover action is performed via {@link CompletableFuture}. + * It also exposes some internal state of {@link AdaptedRestartPipelinedRegionStrategyNG}. + */ + static class TestAdaptedRestartPipelinedRegionStrategyNG extends AdaptedRestartPipelinedRegionStrategyNG { + + private CompletableFuture<?> blockerFuture; + + private Set<ExecutionVertexID> lastTasksToRestart; + + TestAdaptedRestartPipelinedRegionStrategyNG(ExecutionGraph executionGraph) { + super(executionGraph); + this.blockerFuture = CompletableFuture.completedFuture(null); + } + + void setBlockerFuture(CompletableFuture<?> blockerFuture) { + this.blockerFuture = blockerFuture; + } + + @Override + protected void restartTasks(final Set<ExecutionVertexID> verticesToRestart) { + this.lastTasksToRestart = verticesToRestart; + super.restartTasks(verticesToRestart); + } + + @Override + protected CompletableFuture<?> cancelTasks(final Set<ExecutionVertexID> vertices) { + final List<CompletableFuture<?>> terminationAndBlocker = Arrays.asList( + super.cancelTasks(vertices), + blockerFuture); + return FutureUtils.waitForAll(terminationAndBlocker); + } + + CompletableFuture<?> getBlockerFuture() { + return blockerFuture; + } + + Set<ExecutionVertexID> getLastTasksToCancel() { + return lastTasksToRestart; + } + } + + private static class FailingSlotProviderDecorator implements SlotProvider { + + private final SlotProvider delegate; + + Review comment: good catch, fixed ---------------------------------------------------------------- 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 With regards, Apache Git Services
