zhuzhurk 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_r298552634
########## File path: flink-runtime/src/test/java/org/apache/flink/runtime/executiongraph/AdaptedRestartPipelinedRegionStrategyNGConcurrentFailoverTest.java ########## @@ -0,0 +1,282 @@ +/* + * 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.blob.VoidBlobWriter; +import org.apache.flink.runtime.concurrent.ComponentMainThreadExecutor; +import org.apache.flink.runtime.concurrent.ManuallyTriggeredScheduledExecutor; +import org.apache.flink.runtime.execution.ExecutionState; +import org.apache.flink.runtime.executiongraph.AdaptedRestartPipelinedRegionStrategyNGFailoverTest.TestAdaptedRestartPipelinedRegionStrategyNG; +import org.apache.flink.runtime.executiongraph.failover.AdaptedRestartPipelinedRegionStrategyNG; +import org.apache.flink.runtime.executiongraph.failover.FailoverStrategy; +import org.apache.flink.runtime.executiongraph.utils.SimpleSlotProvider; +import org.apache.flink.runtime.io.network.partition.ResultPartitionType; +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.testingUtils.TestingUtils; +import org.apache.flink.runtime.testtasks.NoOpInvokable; +import org.apache.flink.util.TestLogger; + +import org.junit.Before; +import org.junit.Test; + +import java.util.Iterator; +import java.util.concurrent.CompletableFuture; + +import static org.junit.Assert.assertEquals; + +/** + * Tests for {@link AdaptedRestartPipelinedRegionStrategyNG} failover handling when concurrent failovers happen. + * There can be local+local and local+global concurrent failovers. + */ +public class AdaptedRestartPipelinedRegionStrategyNGConcurrentFailoverTest extends TestLogger { + + private static final JobID TEST_JOB_ID = new JobID(); + + private static final int DEFAULT_PARALLELISM = 2; + + private ManuallyTriggeredScheduledExecutor manualMainThreadExecutor; + + private ComponentMainThreadExecutor componentMainThreadExecutor; + + private TestRestartStrategy manuallyTriggeredRestartStrategy; + + @Before + public void setUp() { + manualMainThreadExecutor = new ManuallyTriggeredScheduledExecutor(); + componentMainThreadExecutor = new ScheduledExecutorToComponentMainThreadExecutorAdapter(manualMainThreadExecutor, Thread.currentThread()); + manuallyTriggeredRestartStrategy = TestRestartStrategy.manuallyTriggered(); + } + + /** + * Tests that 2 concurrent region failovers can lead to a properly vertex state. + * <pre> + * (v11) -+-> (v21) + * x + * (v12) -+-> (v22) + * + * ^ + * | + * (blocking) + * </pre> + */ + @Test + public void testConcurrentRegionFailovers() throws Exception { + + // the logic in this test is as follows: + // - start a job + // - cause {ev11} failure and delay the local recovery action via the manual executor + // - cause {ev12} failure and delay the local recovery action via the manual executor + // - resume local recovery actions + // - validate that each task is restarted only once + + final ExecutionGraph eg = createExecutionGraph(); + + final TestAdaptedRestartPipelinedRegionStrategyNG failoverStrategy = + (TestAdaptedRestartPipelinedRegionStrategyNG) eg.getFailoverStrategy(); + failoverStrategy.setBlockerFuture(new CompletableFuture<>()); + + 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(); + + // start job scheduling + eg.scheduleForExecution(); + manualMainThreadExecutor.triggerAll(); + + // fail ev11 to trigger region failover of {ev11}, {ev21}, {ev22} + ev11.getCurrentExecutionAttempt().fail(new Exception("task failure 1")); + manualMainThreadExecutor.triggerAll(); + assertEquals(ExecutionState.FAILED, ev11.getExecutionState()); + assertEquals(ExecutionState.DEPLOYING, ev12.getExecutionState()); + assertEquals(ExecutionState.CANCELED, ev21.getExecutionState()); + assertEquals(ExecutionState.CANCELED, ev22.getExecutionState()); + + // fail ev12 to trigger region failover of {ev12}, {ev21}, {ev22} + ev12.getCurrentExecutionAttempt().fail(new Exception("task failure 2")); + manualMainThreadExecutor.triggerAll(); + assertEquals(ExecutionState.FAILED, ev11.getExecutionState()); + assertEquals(ExecutionState.FAILED, ev12.getExecutionState()); + assertEquals(ExecutionState.CANCELED, ev21.getExecutionState()); + assertEquals(ExecutionState.CANCELED, ev22.getExecutionState()); + + // complete region failover blocker to trigger region failover recovery + failoverStrategy.getBlockerFuture().complete(null); + manualMainThreadExecutor.triggerAll(); + + // verify that all tasks are recovered and no task is restarted more than once + assertEquals(ExecutionState.DEPLOYING, ev11.getExecutionState()); + assertEquals(ExecutionState.DEPLOYING, ev12.getExecutionState()); + assertEquals(ExecutionState.CREATED, ev21.getExecutionState()); + assertEquals(ExecutionState.CREATED, ev22.getExecutionState()); + assertEquals(1, ev11.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev12.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev21.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev22.getCurrentExecutionAttempt().getAttemptNumber()); + } + + /** + * Tests that a global failover will take precedence over local failovers. + * <pre> + * (v11) -+-> (v21) + * x + * (v12) -+-> (v22) + * + * ^ + * | + * (blocking) + * </pre> + */ + @Test + public void testRegionFailoverInterruptedByGlobalFailover() throws Exception { + + // the logic in this test is as follows: + // - start a job + // - cause a task failure and delay the local recovery action via the manual executor + // - cause a global failure + // - resume in local recovery action + // - validate that the local recovery does not restart tasks + + final ExecutionGraph eg = createExecutionGraph(); + + final TestAdaptedRestartPipelinedRegionStrategyNG failoverStrategy = + (TestAdaptedRestartPipelinedRegionStrategyNG) eg.getFailoverStrategy(); + failoverStrategy.setBlockerFuture(new CompletableFuture<>()); + + 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(); + + // start job scheduling + eg.scheduleForExecution(); + manualMainThreadExecutor.triggerAll(); + + // fail ev11 to trigger region failover of {ev11}, {ev21}, {ev22} + ev11.getCurrentExecutionAttempt().fail(new Exception("task failure")); + manualMainThreadExecutor.triggerAll(); + assertEquals(JobStatus.RUNNING, eg.getState()); + assertEquals(ExecutionState.FAILED, ev11.getExecutionState()); + assertEquals(ExecutionState.DEPLOYING, ev12.getExecutionState()); + assertEquals(ExecutionState.CANCELED, ev21.getExecutionState()); + assertEquals(ExecutionState.CANCELED, ev22.getExecutionState()); + + // trigger global failover cancelling and immediately recovery + eg.failGlobal(new Exception("Test global failure")); + ev12.getCurrentExecutionAttempt().completeCancelling(); + manuallyTriggeredRestartStrategy.triggerNextAction(); + manualMainThreadExecutor.triggerAll(); + + // verify the job state and vertex attempt number + assertEquals(2, eg.getGlobalModVersion()); + assertEquals(1, ev11.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev12.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev21.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev22.getCurrentExecutionAttempt().getAttemptNumber()); + + // complete region failover blocker to trigger region failover + failoverStrategy.getBlockerFuture().complete(null); + manualMainThreadExecutor.triggerAll(); + + // verify that no task is restarted by region failover + assertEquals(ExecutionState.DEPLOYING, ev11.getExecutionState()); + assertEquals(ExecutionState.DEPLOYING, ev12.getExecutionState()); + assertEquals(ExecutionState.CREATED, ev21.getExecutionState()); + assertEquals(ExecutionState.CREATED, ev22.getExecutionState()); + assertEquals(1, ev11.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev12.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev21.getCurrentExecutionAttempt().getAttemptNumber()); + assertEquals(1, ev22.getCurrentExecutionAttempt().getAttemptNumber()); + } + + @Test + public void testSkipFailoverIfExecutionStateIsNotRunning() throws Exception { + final ExecutionGraph executionGraph = createExecutionGraph(); + + final Iterator<ExecutionVertex> vertexIterator = executionGraph.getAllExecutionVertices().iterator(); + final ExecutionVertex firstVertex = vertexIterator.next(); + + executionGraph.cancel(); + + final FailoverStrategy failoverStrategy = executionGraph.getFailoverStrategy(); + failoverStrategy.onTaskFailure(firstVertex.getCurrentExecutionAttempt(), new Exception("Test Exception")); + Review comment: The vertex get directly transitioned to `CANCELED ` because it was no scheduled yet and the graph get cancelled. Seems it does not prove whether the local failover is skipped. (Even though a local failover is not skipped, the recovery action will be queued in main thread executor and not able to change the vertex state out from CANCELED). If we want to verify "skip local failover if the job status is not Running", how about trigger task failure directly without graph.cancel()? In this case, the local failover will be skipped since the graph is in CREATED state, and the vertex will remain in `CREATED` state. Without the `triggerAll`, the CANCELED state does not prove that the local failover is skipped, since the local failover may have also triggered the task cancellation and the recovery action is queued in the main thread executor. So the vertex state remains in CANCELLED. ---------------------------------------------------------------- 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
