Examples with ActorGateway org.apache.flink.runtime.instance.ActorGateway used on opensource projects

Example 1 with ActorGateway

use of org.apache.flink.runtime.instance.ActorGateway in project flink by apache.

the class ExecutionVertexStopTest method testStopRpc.

@Test
public void testStopRpc() throws Exception {
    final JobVertexID jid = new JobVertexID();
    final ExecutionJobVertex ejv = getExecutionVertex(jid);
    final ExecutionVertex vertex = new ExecutionVertex(ejv, 0, new IntermediateResult[0], AkkaUtils.getDefaultTimeout());
    final ExecutionAttemptID execId = vertex.getCurrentExecutionAttempt().getAttemptId();
    setVertexState(vertex, ExecutionState.SCHEDULED);
    assertEquals(ExecutionState.SCHEDULED, vertex.getExecutionState());
    final ActorGateway gateway = new StopSequenceInstanceGateway(TestingUtils.defaultExecutionContext());
    Instance instance = getInstance(new ActorTaskManagerGateway(gateway));
    SimpleSlot slot = instance.allocateSimpleSlot(new JobID());
    vertex.deployToSlot(slot);
    receivedStopSignal = false;
    vertex.stop();
    assertTrue(receivedStopSignal);
}

Example 2 with ActorGateway

use of org.apache.flink.runtime.instance.ActorGateway in project flink by apache.

the class ExecutionVertexCancelTest method testCancelConcurrentlyToDeploying_CallsNotOvertaking.

@Test
public void testCancelConcurrentlyToDeploying_CallsNotOvertaking() {
    try {
        final JobVertexID jid = new JobVertexID();
        final TestingUtils.QueuedActionExecutionContext executionContext = TestingUtils.queuedActionExecutionContext();
        final TestingUtils.ActionQueue actions = executionContext.actionQueue();
        final ExecutionJobVertex ejv = getExecutionVertex(jid, executionContext);
        final ExecutionVertex vertex = new ExecutionVertex(ejv, 0, new IntermediateResult[0], AkkaUtils.getDefaultTimeout());
        setVertexState(vertex, ExecutionState.SCHEDULED);
        assertEquals(ExecutionState.SCHEDULED, vertex.getExecutionState());
        ActorGateway actorGateway = new CancelSequenceActorGateway(executionContext, 2);
        Instance instance = getInstance(new ActorTaskManagerGateway(actorGateway));
        SimpleSlot slot = instance.allocateSimpleSlot(new JobID());
        vertex.deployToSlot(slot);
        assertEquals(ExecutionState.DEPLOYING, vertex.getExecutionState());
        vertex.cancel();
        assertEquals(ExecutionState.CANCELING, vertex.getExecutionState());
        // first action happens (deploy)
        actions.triggerNextAction();
        assertEquals(ExecutionState.CANCELING, vertex.getExecutionState());
        // the deploy call found itself in canceling after it returned and needs to send a cancel call
        // the call did not yet execute, so it is still in canceling
        assertEquals(ExecutionState.CANCELING, vertex.getExecutionState());
        // second action happens (cancel call from cancel function)
        actions.triggerNextAction();
        // TaskManager reports back (canceling done)
        vertex.getCurrentExecutionAttempt().cancelingComplete();
        // should properly set state to cancelled
        assertEquals(ExecutionState.CANCELED, vertex.getExecutionState());
        // trigger the correction canceling call
        actions.triggerNextAction();
        assertEquals(ExecutionState.CANCELED, vertex.getExecutionState());
        assertTrue(slot.isReleased());
        assertNull(vertex.getFailureCause());
        assertTrue(vertex.getStateTimestamp(ExecutionState.CREATED) > 0);
        assertTrue(vertex.getStateTimestamp(ExecutionState.CANCELING) > 0);
        assertTrue(vertex.getStateTimestamp(ExecutionState.CANCELED) > 0);
    } catch (Exception e) {
        e.printStackTrace();
        fail(e.getMessage());
    }
}

Example 3 with ActorGateway

use of org.apache.flink.runtime.instance.ActorGateway in project flink by apache.

the class CoordinatorShutdownTest method testCoordinatorShutsDownOnSuccess.

@Test
public void testCoordinatorShutsDownOnSuccess() {
    LocalFlinkMiniCluster cluster = null;
    try {
        Configuration config = new Configuration();
        config.setInteger(ConfigConstants.LOCAL_NUMBER_TASK_MANAGER, 1);
        config.setInteger(ConfigConstants.TASK_MANAGER_NUM_TASK_SLOTS, 1);
        cluster = new LocalFlinkMiniCluster(config, true);
        cluster.start();
        // build a test graph with snapshotting enabled
        JobVertex vertex = new JobVertex("Test Vertex");
        vertex.setInvokableClass(BlockingInvokable.class);
        List<JobVertexID> vertexIdList = Collections.singletonList(vertex.getID());
        JobGraph testGraph = new JobGraph("test job", vertex);
        testGraph.setSnapshotSettings(new JobSnapshottingSettings(vertexIdList, vertexIdList, vertexIdList, 5000, 60000, 0L, Integer.MAX_VALUE, ExternalizedCheckpointSettings.none(), null, true));
        ActorGateway jmGateway = cluster.getLeaderGateway(TestingUtils.TESTING_DURATION());
        FiniteDuration timeout = new FiniteDuration(60, TimeUnit.SECONDS);
        JobManagerMessages.SubmitJob submitMessage = new JobManagerMessages.SubmitJob(testGraph, ListeningBehaviour.EXECUTION_RESULT);
        // submit is successful, but then the job blocks due to the invokable
        Future<Object> submitFuture = jmGateway.ask(submitMessage, timeout);
        Await.result(submitFuture, timeout);
        // get the execution graph and store the ExecutionGraph reference
        Future<Object> jobRequestFuture = jmGateway.ask(new JobManagerMessages.RequestJob(testGraph.getJobID()), timeout);
        ExecutionGraph graph = (ExecutionGraph) ((JobManagerMessages.JobFound) Await.result(jobRequestFuture, timeout)).executionGraph();
        assertNotNull(graph);
        BlockingInvokable.unblock();
        graph.waitUntilFinished();
        // verify that the coordinator was shut down
        CheckpointCoordinator coord = graph.getCheckpointCoordinator();
        assertTrue(coord == null || coord.isShutdown());
    } catch (Exception e) {
        e.printStackTrace();
        fail(e.getMessage());
    } finally {
        if (cluster != null) {
            cluster.shutdown();
            cluster.awaitTermination();
        }
    }
}

Example 4 with ActorGateway

use of org.apache.flink.runtime.instance.ActorGateway in project flink by apache.

the class JobClientActorRecoveryITCase method testJobClientRecovery.

/**
	 * Tests wether the JobClientActor can connect to a newly elected leading job manager to obtain
	 * the JobExecutionResult. The submitted job blocks for the first execution attempt. The
	 * leading job manager will be killed so that the second job manager will be elected as the
	 * leader. The newly elected leader has to retrieve the checkpointed job from ZooKeeper
	 * and continue its execution. This time, the job does not block and, thus, can be finished.
	 * The execution result should be sent to the JobClientActor which originally submitted the
	 * job.
	 *
	 * @throws Exception
	 */
@Test
public void testJobClientRecovery() throws Exception {
    File rootFolder = tempFolder.getRoot();
    Configuration config = ZooKeeperTestUtils.createZooKeeperHAConfig(zkServer.getConnectString(), rootFolder.getPath());
    config.setInteger(ConfigConstants.LOCAL_NUMBER_JOB_MANAGER, 2);
    config.setInteger(ConfigConstants.LOCAL_NUMBER_TASK_MANAGER, 1);
    final TestingCluster cluster = new TestingCluster(config);
    cluster.start();
    JobVertex blockingVertex = new JobVertex("Blocking Vertex");
    blockingVertex.setInvokableClass(BlockingTask.class);
    blockingVertex.setParallelism(1);
    final JobGraph jobGraph = new JobGraph("Blocking Test Job", blockingVertex);
    final Promise<JobExecutionResult> promise = new scala.concurrent.impl.Promise.DefaultPromise<>();
    Deadline deadline = new FiniteDuration(2, TimeUnit.MINUTES).fromNow();
    try {
        Thread submitter = new Thread(new Runnable() {

            @Override
            public void run() {
                try {
                    JobExecutionResult result = cluster.submitJobAndWait(jobGraph, false);
                    promise.success(result);
                } catch (Exception e) {
                    promise.failure(e);
                }
            }
        });
        submitter.start();
        synchronized (BlockingTask.waitLock) {
            while (BlockingTask.HasBlockedExecution < 1 && deadline.hasTimeLeft()) {
                BlockingTask.waitLock.wait(deadline.timeLeft().toMillis());
            }
        }
        if (deadline.isOverdue()) {
            Assert.fail("The job has not blocked within the given deadline.");
        }
        ActorGateway gateway = cluster.getLeaderGateway(deadline.timeLeft());
        gateway.tell(TestingJobManagerMessages.getDisablePostStop());
        gateway.tell(PoisonPill.getInstance());
        // if the job fails then an exception is thrown here
        Await.result(promise.future(), deadline.timeLeft());
    } finally {
        cluster.shutdown();
    }
}

Example 5 with ActorGateway

use of org.apache.flink.runtime.instance.ActorGateway in project flink by apache.

the class ClusterShutdownITCase method testClusterShutdownWithoutResourceManager.

/**
	 * Tests a faked cluster shutdown procedure without the ResourceManager.
	 */
@Test
public void testClusterShutdownWithoutResourceManager() {
    new JavaTestKit(system) {

        {
            new Within(duration("30 seconds")) {

                @Override
                protected void run() {
                    ActorGateway me = TestingUtils.createForwardingActor(system, getTestActor(), Option.<String>empty());
                    // start job manager which doesn't shutdown the actor system
                    ActorGateway jobManager = TestingUtils.createJobManager(system, TestingUtils.defaultExecutor(), TestingUtils.defaultExecutor(), config, "jobmanager1");
                    // Tell the JobManager to inform us of shutdown actions
                    jobManager.tell(TestingMessages.getNotifyOfComponentShutdown(), me);
                    // Register a TaskManager
                    ActorGateway taskManager = TestingUtils.createTaskManager(system, jobManager, config, true, true);
                    // Tell the TaskManager to inform us of TaskManager shutdowns
                    taskManager.tell(TestingMessages.getNotifyOfComponentShutdown(), me);
                    // No resource manager connected
                    jobManager.tell(new StopCluster(ApplicationStatus.SUCCEEDED, "Shutting down."), me);
                    expectMsgAllOf(new TestingMessages.ComponentShutdown(taskManager.actor()), new TestingMessages.ComponentShutdown(jobManager.actor()), StopClusterSuccessful.getInstance());
                }
            };
        }
    };
}