Examples with ExecutionAttemptID org.apache.flink.runtime.executiongraph.ExecutionAttemptID used on opensource projects

Example 26 with ExecutionAttemptID

use of org.apache.flink.runtime.executiongraph.ExecutionAttemptID in project flink by apache.

the class TaskManagerTest method testRemotePartitionNotFound.

/**
	 * Tests that repeated remote {@link PartitionNotFoundException}s ultimately fail the receiver.
	 */
@Test
public void testRemotePartitionNotFound() throws Exception {
    new JavaTestKit(system) {

        {
            ActorGateway jobManager = null;
            ActorGateway taskManager = null;
            final ActorGateway testActorGateway = new AkkaActorGateway(getTestActor(), leaderSessionID);
            try {
                final IntermediateDataSetID resultId = new IntermediateDataSetID();
                // Create the JM
                ActorRef jm = system.actorOf(Props.create(new SimplePartitionStateLookupJobManagerCreator(leaderSessionID, getTestActor())));
                jobManager = new AkkaActorGateway(jm, leaderSessionID);
                final int dataPort = NetUtils.getAvailablePort();
                Configuration config = new Configuration();
                config.setInteger(ConfigConstants.TASK_MANAGER_DATA_PORT_KEY, dataPort);
                config.setInteger(TaskManagerOptions.NETWORK_REQUEST_BACKOFF_INITIAL, 100);
                config.setInteger(TaskManagerOptions.NETWORK_REQUEST_BACKOFF_MAX, 200);
                taskManager = TestingUtils.createTaskManager(system, jobManager, config, false, true);
                // ---------------------------------------------------------------------------------
                final ActorGateway tm = taskManager;
                final JobID jid = new JobID();
                final JobVertexID vid = new JobVertexID();
                final ExecutionAttemptID eid = new ExecutionAttemptID();
                final ResultPartitionID partitionId = new ResultPartitionID();
                // Remote location (on the same TM though) for the partition
                final ResultPartitionLocation loc = ResultPartitionLocation.createRemote(new ConnectionID(new InetSocketAddress("localhost", dataPort), 0));
                final InputChannelDeploymentDescriptor[] icdd = new InputChannelDeploymentDescriptor[] { new InputChannelDeploymentDescriptor(partitionId, loc) };
                final InputGateDeploymentDescriptor igdd = new InputGateDeploymentDescriptor(resultId, ResultPartitionType.PIPELINED, 0, icdd);
                final TaskDeploymentDescriptor tdd = createTaskDeploymentDescriptor(jid, "TestJob", vid, eid, new SerializedValue<>(new ExecutionConfig()), "Receiver", 1, 0, 1, 0, new Configuration(), new Configuration(), Tasks.AgnosticReceiver.class.getName(), Collections.<ResultPartitionDeploymentDescriptor>emptyList(), Collections.singletonList(igdd), Collections.<BlobKey>emptyList(), Collections.<URL>emptyList(), 0);
                new Within(d) {

                    @Override
                    protected void run() {
                        // Submit the task
                        tm.tell(new SubmitTask(tdd), testActorGateway);
                        expectMsgClass(Acknowledge.get().getClass());
                        // Wait to be notified about the final execution state by the mock JM
                        TaskExecutionState msg = expectMsgClass(TaskExecutionState.class);
                        // The task should fail after repeated requests
                        assertEquals(ExecutionState.FAILED, msg.getExecutionState());
                        Throwable t = msg.getError(ClassLoader.getSystemClassLoader());
                        assertEquals("Thrown exception was not a PartitionNotFoundException: " + t.getMessage(), PartitionNotFoundException.class, t.getClass());
                    }
                };
            } catch (Exception e) {
                e.printStackTrace();
                fail(e.getMessage());
            } finally {
                TestingUtils.stopActor(taskManager);
                TestingUtils.stopActor(jobManager);
            }
        }
    };
}

Example 27 with ExecutionAttemptID

use of org.apache.flink.runtime.executiongraph.ExecutionAttemptID in project flink by apache.

the class TaskManagerTest method testJobSubmissionAndCanceling.

@Test
public void testJobSubmissionAndCanceling() {
    new JavaTestKit(system) {

        {
            ActorGateway jobManager = null;
            ActorGateway taskManager = null;
            final ActorGateway testActorGateway = new AkkaActorGateway(getTestActor(), leaderSessionID);
            try {
                ActorRef jm = system.actorOf(Props.create(SimpleJobManager.class, leaderSessionID));
                jobManager = new AkkaActorGateway(jm, leaderSessionID);
                taskManager = TestingUtils.createTaskManager(system, jobManager, new Configuration(), true, true);
                final JobID jid1 = new JobID();
                final JobID jid2 = new JobID();
                JobVertexID vid1 = new JobVertexID();
                JobVertexID vid2 = new JobVertexID();
                final ExecutionAttemptID eid1 = new ExecutionAttemptID();
                final ExecutionAttemptID eid2 = new ExecutionAttemptID();
                final TaskDeploymentDescriptor tdd1 = createTaskDeploymentDescriptor(jid1, "TestJob1", vid1, eid1, new SerializedValue<>(new ExecutionConfig()), "TestTask1", 5, 1, 5, 0, new Configuration(), new Configuration(), TestInvokableBlockingCancelable.class.getName(), Collections.<ResultPartitionDeploymentDescriptor>emptyList(), Collections.<InputGateDeploymentDescriptor>emptyList(), new ArrayList<BlobKey>(), Collections.<URL>emptyList(), 0);
                final TaskDeploymentDescriptor tdd2 = createTaskDeploymentDescriptor(jid2, "TestJob2", vid2, eid2, new SerializedValue<>(new ExecutionConfig()), "TestTask2", 7, 2, 7, 0, new Configuration(), new Configuration(), TestInvokableBlockingCancelable.class.getName(), Collections.<ResultPartitionDeploymentDescriptor>emptyList(), Collections.<InputGateDeploymentDescriptor>emptyList(), new ArrayList<BlobKey>(), Collections.<URL>emptyList(), 0);
                final ActorGateway tm = taskManager;
                new Within(d) {

                    @Override
                    protected void run() {
                        try {
                            Future<Object> t1Running = tm.ask(new TestingTaskManagerMessages.NotifyWhenTaskIsRunning(eid1), timeout);
                            Future<Object> t2Running = tm.ask(new TestingTaskManagerMessages.NotifyWhenTaskIsRunning(eid2), timeout);
                            tm.tell(new SubmitTask(tdd1), testActorGateway);
                            tm.tell(new SubmitTask(tdd2), testActorGateway);
                            expectMsgEquals(Acknowledge.get());
                            expectMsgEquals(Acknowledge.get());
                            Await.ready(t1Running, d);
                            Await.ready(t2Running, d);
                            tm.tell(TestingTaskManagerMessages.getRequestRunningTasksMessage(), testActorGateway);
                            Map<ExecutionAttemptID, Task> runningTasks = expectMsgClass(TestingTaskManagerMessages.ResponseRunningTasks.class).asJava();
                            assertEquals(2, runningTasks.size());
                            Task t1 = runningTasks.get(eid1);
                            Task t2 = runningTasks.get(eid2);
                            assertNotNull(t1);
                            assertNotNull(t2);
                            assertEquals(ExecutionState.RUNNING, t1.getExecutionState());
                            assertEquals(ExecutionState.RUNNING, t2.getExecutionState());
                            tm.tell(new CancelTask(eid1), testActorGateway);
                            expectMsgEquals(Acknowledge.get());
                            Future<Object> response = tm.ask(new TestingTaskManagerMessages.NotifyWhenTaskRemoved(eid1), timeout);
                            Await.ready(response, d);
                            assertEquals(ExecutionState.CANCELED, t1.getExecutionState());
                            tm.tell(TestingTaskManagerMessages.getRequestRunningTasksMessage(), testActorGateway);
                            runningTasks = expectMsgClass(TestingTaskManagerMessages.ResponseRunningTasks.class).asJava();
                            assertEquals(1, runningTasks.size());
                            tm.tell(new CancelTask(eid1), testActorGateway);
                            expectMsgEquals(Acknowledge.get());
                            tm.tell(new CancelTask(eid2), testActorGateway);
                            expectMsgEquals(Acknowledge.get());
                            response = tm.ask(new TestingTaskManagerMessages.NotifyWhenTaskRemoved(eid2), timeout);
                            Await.ready(response, d);
                            assertEquals(ExecutionState.CANCELED, t2.getExecutionState());
                            tm.tell(TestingTaskManagerMessages.getRequestRunningTasksMessage(), testActorGateway);
                            runningTasks = expectMsgClass(TestingTaskManagerMessages.ResponseRunningTasks.class).asJava();
                            assertEquals(0, runningTasks.size());
                        } catch (Exception e) {
                            e.printStackTrace();
                            fail(e.getMessage());
                        }
                    }
                };
            } catch (Exception e) {
                e.printStackTrace();
                fail(e.getMessage());
            } finally {
                TestingUtils.stopActor(taskManager);
                TestingUtils.stopActor(jobManager);
            }
        }
    };
}

Example 28 with ExecutionAttemptID

use of org.apache.flink.runtime.executiongraph.ExecutionAttemptID in project flink by apache.

the class TaskManagerTest method testFailingScheduleOrUpdateConsumersMessage.

/**
	 * Test that a failing schedule or update consumers call leads to the failing of the respective
	 * task.
	 *
	 * IMPORTANT: We have to make sure that the invokable's cancel method is called, because only
	 * then the future is completed. We do this by not eagerly deploy consumer tasks and requiring
	 * the invokable to fill one memory segment. The completed memory segment will trigger the
	 * scheduling of the downstream operator since it is in pipeline mode. After we've filled the
	 * memory segment, we'll block the invokable and wait for the task failure due to the failed
	 * schedule or update consumers call.
	 */
@Test(timeout = 10000L)
public void testFailingScheduleOrUpdateConsumersMessage() throws Exception {
    new JavaTestKit(system) {

        {
            final Configuration configuration = new Configuration();
            // set the memory segment to the smallest size possible, because we have to fill one
            // memory buffer to trigger the schedule or update consumers message to the downstream
            // operators
            configuration.setInteger(ConfigConstants.TASK_MANAGER_MEMORY_SEGMENT_SIZE_KEY, 4096);
            final JobID jid = new JobID();
            final JobVertexID vid = new JobVertexID();
            final ExecutionAttemptID eid = new ExecutionAttemptID();
            final SerializedValue<ExecutionConfig> executionConfig = new SerializedValue<>(new ExecutionConfig());
            final ResultPartitionDeploymentDescriptor resultPartitionDeploymentDescriptor = new ResultPartitionDeploymentDescriptor(new IntermediateDataSetID(), new IntermediateResultPartitionID(), ResultPartitionType.PIPELINED, 1, 1, true);
            final TaskDeploymentDescriptor tdd = createTaskDeploymentDescriptor(jid, "TestJob", vid, eid, executionConfig, "TestTask", 1, 0, 1, 0, new Configuration(), new Configuration(), TestInvokableRecordCancel.class.getName(), Collections.singletonList(resultPartitionDeploymentDescriptor), Collections.<InputGateDeploymentDescriptor>emptyList(), new ArrayList<BlobKey>(), Collections.<URL>emptyList(), 0);
            ActorRef jmActorRef = system.actorOf(Props.create(FailingScheduleOrUpdateConsumersJobManager.class, leaderSessionID), "jobmanager");
            ActorGateway jobManager = new AkkaActorGateway(jmActorRef, leaderSessionID);
            final ActorGateway taskManager = TestingUtils.createTaskManager(system, jobManager, configuration, true, true);
            try {
                TestInvokableRecordCancel.resetGotCanceledFuture();
                Future<Object> result = taskManager.ask(new SubmitTask(tdd), timeout);
                Await.result(result, timeout);
                org.apache.flink.runtime.concurrent.Future<Boolean> cancelFuture = TestInvokableRecordCancel.gotCanceled();
                assertEquals(true, cancelFuture.get());
            } finally {
                TestingUtils.stopActor(taskManager);
                TestingUtils.stopActor(jobManager);
            }
        }
    };
}

Example 29 with ExecutionAttemptID

use of org.apache.flink.runtime.executiongraph.ExecutionAttemptID in project flink by apache.

the class CheckpointCoordinatorTest method testMaxConcurrentAttemptsWithSubsumption.

@Test
public void testMaxConcurrentAttemptsWithSubsumption() throws Exception {
    final int maxConcurrentAttempts = 2;
    JobVertexID jobVertexID1 = new JobVertexID();
    ExecutionGraph graph = new CheckpointCoordinatorTestingUtils.CheckpointExecutionGraphBuilder().addJobVertex(jobVertexID1).build();
    ExecutionVertex vertex1 = graph.getJobVertex(jobVertexID1).getTaskVertices()[0];
    ExecutionAttemptID attemptID1 = vertex1.getCurrentExecutionAttempt().getAttemptId();
    CheckpointCoordinatorConfiguration chkConfig = new CheckpointCoordinatorConfiguration.CheckpointCoordinatorConfigurationBuilder().setCheckpointInterval(// periodic interval is 10 ms
    10).setCheckpointTimeout(// timeout is very long (200 s)
    200000).setMinPauseBetweenCheckpoints(// no extra delay
    0L).setMaxConcurrentCheckpoints(maxConcurrentAttempts).build();
    CheckpointCoordinator checkpointCoordinator = new CheckpointCoordinatorBuilder().setExecutionGraph(graph).setCheckpointCoordinatorConfiguration(chkConfig).setCompletedCheckpointStore(new StandaloneCompletedCheckpointStore(2)).setTimer(manuallyTriggeredScheduledExecutor).build();
    checkpointCoordinator.startCheckpointScheduler();
    do {
        manuallyTriggeredScheduledExecutor.triggerPeriodicScheduledTasks();
        manuallyTriggeredScheduledExecutor.triggerAll();
    } while (checkpointCoordinator.getNumberOfPendingCheckpoints() < maxConcurrentAttempts);
    // validate that the pending checkpoints are there
    assertEquals(maxConcurrentAttempts, checkpointCoordinator.getNumberOfPendingCheckpoints());
    assertNotNull(checkpointCoordinator.getPendingCheckpoints().get(1L));
    assertNotNull(checkpointCoordinator.getPendingCheckpoints().get(2L));
    // now we acknowledge the second checkpoint, which should subsume the first checkpoint
    // and allow two more checkpoints to be triggered
    // now, once we acknowledge one checkpoint, it should trigger the next one
    checkpointCoordinator.receiveAcknowledgeMessage(new AcknowledgeCheckpoint(graph.getJobID(), attemptID1, 2L), TASK_MANAGER_LOCATION_INFO);
    // after a while, there should be the new checkpoints
    do {
        manuallyTriggeredScheduledExecutor.triggerPeriodicScheduledTasks();
        manuallyTriggeredScheduledExecutor.triggerAll();
    } while (checkpointCoordinator.getNumberOfPendingCheckpoints() < maxConcurrentAttempts);
    // do the final check
    assertEquals(maxConcurrentAttempts, checkpointCoordinator.getNumberOfPendingCheckpoints());
    assertNotNull(checkpointCoordinator.getPendingCheckpoints().get(3L));
    assertNotNull(checkpointCoordinator.getPendingCheckpoints().get(4L));
    checkpointCoordinator.shutdown();
}

Example 30 with ExecutionAttemptID

use of org.apache.flink.runtime.executiongraph.ExecutionAttemptID in project flink by apache.

the class CheckpointCoordinatorTest method testTasksFinishDuringTriggering.

@Test
public void testTasksFinishDuringTriggering() throws Exception {
    JobVertexID jobVertexID1 = new JobVertexID();
    JobVertexID jobVertexID2 = new JobVertexID();
    ExecutionGraph graph = new CheckpointCoordinatorTestingUtils.CheckpointExecutionGraphBuilder().setTransitToRunning(false).addJobVertex(jobVertexID1, 1, 256).addJobVertex(jobVertexID2, 1, 256).build();
    ExecutionJobVertex jobVertex1 = graph.getJobVertex(jobVertexID1);
    ExecutionVertex taskVertex = jobVertex1.getTaskVertices()[0];
    ExecutionJobVertex jobVertex2 = graph.getJobVertex(jobVertexID2);
    ExecutionVertex taskVertex2 = jobVertex2.getTaskVertices()[0];
    AtomicBoolean checkpointAborted = new AtomicBoolean(false);
    LogicalSlot slot1 = new TestingLogicalSlotBuilder().setTaskManagerGateway(new SimpleAckingTaskManagerGateway() {

        @Override
        public CompletableFuture<Acknowledge> triggerCheckpoint(ExecutionAttemptID executionAttemptID, JobID jobId, long checkpointId, long timestamp, CheckpointOptions checkpointOptions) {
            taskVertex.getCurrentExecutionAttempt().markFinished();
            return FutureUtils.completedExceptionally(new RpcException(""));
        }
    }).createTestingLogicalSlot();
    LogicalSlot slot2 = new TestingLogicalSlotBuilder().setTaskManagerGateway(new SimpleAckingTaskManagerGateway() {

        @Override
        public void notifyCheckpointAborted(ExecutionAttemptID executionAttemptID, JobID jobId, long checkpointId, long latestCompletedCheckpointId, long timestamp) {
            checkpointAborted.set(true);
        }
    }).createTestingLogicalSlot();
    ExecutionGraphTestUtils.setVertexResource(taskVertex, slot1);
    taskVertex.getCurrentExecutionAttempt().transitionState(ExecutionState.RUNNING);
    ExecutionGraphTestUtils.setVertexResource(taskVertex2, slot2);
    taskVertex2.getCurrentExecutionAttempt().transitionState(ExecutionState.RUNNING);
    CheckpointCoordinator checkpointCoordinator = new CheckpointCoordinatorBuilder().setExecutionGraph(graph).setTimer(manuallyTriggeredScheduledExecutor).setAllowCheckpointsAfterTasksFinished(true).build();
    // nothing should be happening
    assertEquals(0, checkpointCoordinator.getNumberOfPendingCheckpoints());
    assertEquals(0, checkpointCoordinator.getNumberOfRetainedSuccessfulCheckpoints());
    // trigger the first checkpoint. this will not fail because we allow checkpointing even with
    // finished tasks
    final CompletableFuture<CompletedCheckpoint> checkpointFuture = checkpointCoordinator.triggerCheckpoint(false);
    manuallyTriggeredScheduledExecutor.triggerAll();
    assertTrue(checkpointFuture.isCompletedExceptionally());
    assertTrue(checkpointAborted.get());
}