Examples with JobMasterGateway org.apache.flink.runtime.jobmaster.JobMasterGateway used on opensource projects

Example 1 with JobMasterGateway

use of org.apache.flink.runtime.jobmaster.JobMasterGateway in project flink by apache.

the class TaskExecutor method offerSlotsToJobManager.

// ------------------------------------------------------------------------
//  Internal job manager connection methods
// ------------------------------------------------------------------------
private void offerSlotsToJobManager(final JobID jobId) {
    final JobManagerConnection jobManagerConnection = jobManagerTable.get(jobId);
    if (jobManagerConnection == null) {
        log.debug("There is no job manager connection to the leader of job {}.", jobId);
    } else {
        if (taskSlotTable.hasAllocatedSlots(jobId)) {
            log.info("Offer reserved slots to the leader of job {}.", jobId);
            final JobMasterGateway jobMasterGateway = jobManagerConnection.getJobManagerGateway();
            final Iterator<TaskSlot> reservedSlotsIterator = taskSlotTable.getAllocatedSlots(jobId);
            final UUID leaderId = jobManagerConnection.getLeaderId();
            final Collection<SlotOffer> reservedSlots = new HashSet<>(2);
            while (reservedSlotsIterator.hasNext()) {
                SlotOffer offer = reservedSlotsIterator.next().generateSlotOffer();
                try {
                    if (!taskSlotTable.markSlotActive(offer.getAllocationId())) {
                        // the slot is either free or releasing at the moment
                        final String message = "Could not mark slot " + jobId + " active.";
                        log.debug(message);
                        jobMasterGateway.failSlot(getResourceID(), offer.getAllocationId(), leaderId, new Exception(message));
                    }
                } catch (SlotNotFoundException e) {
                    final String message = "Could not mark slot " + jobId + " active.";
                    jobMasterGateway.failSlot(getResourceID(), offer.getAllocationId(), leaderId, new Exception(message));
                    continue;
                }
                reservedSlots.add(offer);
            }
            Future<Iterable<SlotOffer>> acceptedSlotsFuture = jobMasterGateway.offerSlots(getResourceID(), reservedSlots, leaderId, taskManagerConfiguration.getTimeout());
            acceptedSlotsFuture.thenAcceptAsync(new AcceptFunction<Iterable<SlotOffer>>() {

                @Override
                public void accept(Iterable<SlotOffer> acceptedSlots) {
                    // check if the response is still valid
                    if (isJobManagerConnectionValid(jobId, leaderId)) {
                        // mark accepted slots active
                        for (SlotOffer acceptedSlot : acceptedSlots) {
                            reservedSlots.remove(acceptedSlot);
                        }
                        final Exception e = new Exception("The slot was rejected by the JobManager.");
                        for (SlotOffer rejectedSlot : reservedSlots) {
                            freeSlot(rejectedSlot.getAllocationId(), e);
                        }
                    } else {
                        // discard the response since there is a new leader for the job
                        log.debug("Discard offer slot response since there is a new leader " + "for the job {}.", jobId);
                    }
                }
            }, getMainThreadExecutor());
            acceptedSlotsFuture.exceptionallyAsync(new ApplyFunction<Throwable, Void>() {

                @Override
                public Void apply(Throwable throwable) {
                    if (throwable instanceof TimeoutException) {
                        // We ran into a timeout. Try again.
                        offerSlotsToJobManager(jobId);
                    } else {
                        // We encountered an exception. Free the slots and return them to the RM.
                        for (SlotOffer reservedSlot : reservedSlots) {
                            freeSlot(reservedSlot.getAllocationId(), throwable);
                        }
                    }
                    return null;
                }
            }, getMainThreadExecutor());
        } else {
            log.debug("There are no unassigned slots for the job {}.", jobId);
        }
    }
}

Example 2 with JobMasterGateway

use of org.apache.flink.runtime.jobmaster.JobMasterGateway in project flink by apache.

the class ResourceManager method registerJobManager.

// ------------------------------------------------------------------------
//  RPC methods
// ------------------------------------------------------------------------
@RpcMethod
public Future<RegistrationResponse> registerJobManager(final UUID resourceManagerLeaderId, final UUID jobManagerLeaderId, final String jobManagerAddress, final JobID jobId) {
    checkNotNull(resourceManagerLeaderId);
    checkNotNull(jobManagerLeaderId);
    checkNotNull(jobManagerAddress);
    checkNotNull(jobId);
    if (isValid(resourceManagerLeaderId)) {
        if (!jobLeaderIdService.containsJob(jobId)) {
            try {
                jobLeaderIdService.addJob(jobId);
            } catch (Exception e) {
                ResourceManagerException exception = new ResourceManagerException("Could not add the job " + jobId + " to the job id leader service.", e);
                onFatalErrorAsync(exception);
                log.error("Could not add job {} to job leader id service.", jobId, e);
                return FlinkCompletableFuture.completedExceptionally(exception);
            }
        }
        log.info("Registering job manager {}@{} for job {}.", jobManagerLeaderId, jobManagerAddress, jobId);
        Future<UUID> jobLeaderIdFuture;
        try {
            jobLeaderIdFuture = jobLeaderIdService.getLeaderId(jobId);
        } catch (Exception e) {
            // we cannot check the job leader id so let's fail
            // TODO: Maybe it's also ok to skip this check in case that we cannot check the leader id
            ResourceManagerException exception = new ResourceManagerException("Cannot obtain the " + "job leader id future to verify the correct job leader.", e);
            onFatalErrorAsync(exception);
            log.debug("Could not obtain the job leader id future to verify the correct job leader.");
            return FlinkCompletableFuture.completedExceptionally(exception);
        }
        Future<JobMasterGateway> jobMasterGatewayFuture = getRpcService().connect(jobManagerAddress, JobMasterGateway.class);
        Future<RegistrationResponse> registrationResponseFuture = jobMasterGatewayFuture.thenCombineAsync(jobLeaderIdFuture, new BiFunction<JobMasterGateway, UUID, RegistrationResponse>() {

            @Override
            public RegistrationResponse apply(JobMasterGateway jobMasterGateway, UUID jobLeaderId) {
                if (isValid(resourceManagerLeaderId)) {
                    if (jobLeaderId.equals(jobManagerLeaderId)) {
                        if (jobManagerRegistrations.containsKey(jobId)) {
                            JobManagerRegistration oldJobManagerRegistration = jobManagerRegistrations.get(jobId);
                            if (oldJobManagerRegistration.getLeaderID().equals(jobLeaderId)) {
                                // same registration
                                log.debug("Job manager {}@{} was already registered.", jobManagerLeaderId, jobManagerAddress);
                            } else {
                                // tell old job manager that he is no longer the job leader
                                disconnectJobManager(oldJobManagerRegistration.getJobID(), new Exception("New job leader for job " + jobId + " found."));
                                JobManagerRegistration jobManagerRegistration = new JobManagerRegistration(jobId, jobLeaderId, jobMasterGateway);
                                jobManagerRegistrations.put(jobId, jobManagerRegistration);
                            }
                        } else {
                            // new registration for the job
                            JobManagerRegistration jobManagerRegistration = new JobManagerRegistration(jobId, jobLeaderId, jobMasterGateway);
                            jobManagerRegistrations.put(jobId, jobManagerRegistration);
                        }
                        log.info("Registered job manager {}@{} for job {}.", jobManagerLeaderId, jobManagerAddress, jobId);
                        return new JobMasterRegistrationSuccess(resourceManagerConfiguration.getHeartbeatInterval().toMilliseconds(), getLeaderSessionId());
                    } else {
                        log.debug("The job manager leader id {} did not match the job " + "leader id {}.", jobManagerLeaderId, jobLeaderId);
                        return new RegistrationResponse.Decline("Job manager leader id did not match.");
                    }
                } else {
                    log.debug("The resource manager leader id changed {}. Discarding job " + "manager registration from {}.", getLeaderSessionId(), jobManagerAddress);
                    return new RegistrationResponse.Decline("Resource manager leader id changed.");
                }
            }
        }, getMainThreadExecutor());
        // handle exceptions which might have occurred in one of the futures inputs of combine
        return registrationResponseFuture.handleAsync(new BiFunction<RegistrationResponse, Throwable, RegistrationResponse>() {

            @Override
            public RegistrationResponse apply(RegistrationResponse registrationResponse, Throwable throwable) {
                if (throwable != null) {
                    if (log.isDebugEnabled()) {
                        log.debug("Registration of job manager {}@{} failed.", jobManagerLeaderId, jobManagerAddress, throwable);
                    } else {
                        log.info("Registration of job manager {}@{} failed.", jobManagerLeaderId, jobManagerAddress);
                    }
                    return new RegistrationResponse.Decline(throwable.getMessage());
                } else {
                    return registrationResponse;
                }
            }
        }, getRpcService().getExecutor());
    } else {
        log.debug("Discard register job manager message from {}, because the leader id " + "{} did not match the expected leader id {}.", jobManagerAddress, resourceManagerLeaderId, leaderSessionId);
        return FlinkCompletableFuture.<RegistrationResponse>completed(new RegistrationResponse.Decline("Resource manager leader id did not match."));
    }
}

Example 3 with JobMasterGateway

use of org.apache.flink.runtime.jobmaster.JobMasterGateway in project flink by apache.

the class TaskExecutorTest method testJobLeaderDetection.

/**
	 * Tests that a TaskManager detects a job leader for which has reserved slots. Upon detecting
	 * the job leader, it will offer all reserved slots to the JobManager.
	 */
@Test
public void testJobLeaderDetection() throws Exception {
    final JobID jobId = new JobID();
    final TestingSerialRpcService rpc = new TestingSerialRpcService();
    final Configuration configuration = new Configuration();
    final TaskManagerConfiguration taskManagerConfiguration = TaskManagerConfiguration.fromConfiguration(configuration);
    final ResourceID resourceId = new ResourceID("foobar");
    final TaskManagerLocation taskManagerLocation = new TaskManagerLocation(resourceId, InetAddress.getLoopbackAddress(), 1234);
    final TestingHighAvailabilityServices haServices = new TestingHighAvailabilityServices();
    final TimerService<AllocationID> timerService = mock(TimerService.class);
    final TaskSlotTable taskSlotTable = new TaskSlotTable(Arrays.asList(mock(ResourceProfile.class)), timerService);
    final JobManagerTable jobManagerTable = new JobManagerTable();
    final JobLeaderService jobLeaderService = new JobLeaderService(taskManagerLocation);
    final TestingFatalErrorHandler testingFatalErrorHandler = new TestingFatalErrorHandler();
    final TestingLeaderRetrievalService resourceManagerLeaderRetrievalService = new TestingLeaderRetrievalService();
    final TestingLeaderRetrievalService jobManagerLeaderRetrievalService = new TestingLeaderRetrievalService();
    haServices.setResourceManagerLeaderRetriever(resourceManagerLeaderRetrievalService);
    haServices.setJobMasterLeaderRetriever(jobId, jobManagerLeaderRetrievalService);
    final String resourceManagerAddress = "rm";
    final UUID resourceManagerLeaderId = UUID.randomUUID();
    final ResourceManagerGateway resourceManagerGateway = mock(ResourceManagerGateway.class);
    final InstanceID registrationId = new InstanceID();
    when(resourceManagerGateway.registerTaskExecutor(eq(resourceManagerLeaderId), any(String.class), eq(resourceId), any(SlotReport.class), any(Time.class))).thenReturn(FlinkCompletableFuture.<RegistrationResponse>completed(new TaskExecutorRegistrationSuccess(registrationId, 1000L)));
    final String jobManagerAddress = "jm";
    final UUID jobManagerLeaderId = UUID.randomUUID();
    final ResourceID jmResourceId = new ResourceID(jobManagerAddress);
    final int blobPort = 42;
    final JobMasterGateway jobMasterGateway = mock(JobMasterGateway.class);
    when(jobMasterGateway.registerTaskManager(any(String.class), eq(taskManagerLocation), eq(jobManagerLeaderId), any(Time.class))).thenReturn(FlinkCompletableFuture.<RegistrationResponse>completed(new JMTMRegistrationSuccess(jmResourceId, blobPort)));
    when(jobMasterGateway.getHostname()).thenReturn(jobManagerAddress);
    rpc.registerGateway(resourceManagerAddress, resourceManagerGateway);
    rpc.registerGateway(jobManagerAddress, jobMasterGateway);
    final AllocationID allocationId = new AllocationID();
    final SlotID slotId = new SlotID(resourceId, 0);
    final SlotOffer slotOffer = new SlotOffer(allocationId, 0, ResourceProfile.UNKNOWN);
    try {
        TaskExecutor taskManager = new TaskExecutor(taskManagerConfiguration, taskManagerLocation, rpc, mock(MemoryManager.class), mock(IOManager.class), mock(NetworkEnvironment.class), haServices, mock(HeartbeatServices.class, RETURNS_MOCKS), mock(MetricRegistry.class), mock(TaskManagerMetricGroup.class), mock(BroadcastVariableManager.class), mock(FileCache.class), taskSlotTable, jobManagerTable, jobLeaderService, testingFatalErrorHandler);
        taskManager.start();
        // tell the task manager about the rm leader
        resourceManagerLeaderRetrievalService.notifyListener(resourceManagerAddress, resourceManagerLeaderId);
        // request slots from the task manager under the given allocation id
        TMSlotRequestReply reply = taskManager.requestSlot(slotId, jobId, allocationId, jobManagerAddress, resourceManagerLeaderId);
        // this is hopefully successful :-)
        assertTrue(reply instanceof TMSlotRequestRegistered);
        // now inform the task manager about the new job leader
        jobManagerLeaderRetrievalService.notifyListener(jobManagerAddress, jobManagerLeaderId);
        // the job leader should get the allocation id offered
        verify(jobMasterGateway).offerSlots(any(ResourceID.class), (Iterable<SlotOffer>) Matchers.argThat(contains(slotOffer)), eq(jobManagerLeaderId), any(Time.class));
        // check if a concurrent error occurred
        testingFatalErrorHandler.rethrowError();
    } finally {
        rpc.stopService();
    }
}

Example 4 with JobMasterGateway

use of org.apache.flink.runtime.jobmaster.JobMasterGateway in project flink by apache.

the class TaskExecutorTest method testSubmitTaskBeforeAcceptSlot.

/**
	 * This tests task executor receive SubmitTask before OfferSlot response.
	 */
@Test
public void testSubmitTaskBeforeAcceptSlot() throws Exception {
    final JobID jobId = new JobID();
    final TestingSerialRpcService rpc = new TestingSerialRpcService();
    final Configuration configuration = new Configuration();
    final TaskManagerConfiguration taskManagerConfiguration = TaskManagerConfiguration.fromConfiguration(configuration);
    final ResourceID resourceId = new ResourceID("foobar");
    final TaskManagerLocation taskManagerLocation = new TaskManagerLocation(resourceId, InetAddress.getLoopbackAddress(), 1234);
    final TestingHighAvailabilityServices haServices = new TestingHighAvailabilityServices();
    final TimerService<AllocationID> timerService = mock(TimerService.class);
    final TaskSlotTable taskSlotTable = new TaskSlotTable(Arrays.asList(mock(ResourceProfile.class), mock(ResourceProfile.class)), timerService);
    final JobManagerTable jobManagerTable = new JobManagerTable();
    final JobLeaderService jobLeaderService = new JobLeaderService(taskManagerLocation);
    final TestingFatalErrorHandler testingFatalErrorHandler = new TestingFatalErrorHandler();
    final String resourceManagerAddress = "rm";
    final UUID resourceManagerLeaderId = UUID.randomUUID();
    final String jobManagerAddress = "jm";
    final UUID jobManagerLeaderId = UUID.randomUUID();
    final LeaderRetrievalService resourceManagerLeaderRetrievalService = new TestingLeaderRetrievalService(resourceManagerAddress, resourceManagerLeaderId);
    final LeaderRetrievalService jobManagerLeaderRetrievalService = new TestingLeaderRetrievalService(jobManagerAddress, jobManagerLeaderId);
    haServices.setResourceManagerLeaderRetriever(resourceManagerLeaderRetrievalService);
    haServices.setJobMasterLeaderRetriever(jobId, jobManagerLeaderRetrievalService);
    final ResourceManagerGateway resourceManagerGateway = mock(ResourceManagerGateway.class);
    final InstanceID registrationId = new InstanceID();
    when(resourceManagerGateway.registerTaskExecutor(eq(resourceManagerLeaderId), any(String.class), eq(resourceId), any(SlotReport.class), any(Time.class))).thenReturn(FlinkCompletableFuture.<RegistrationResponse>completed(new TaskExecutorRegistrationSuccess(registrationId, 1000L)));
    final ResourceID jmResourceId = new ResourceID(jobManagerAddress);
    final int blobPort = 42;
    final AllocationID allocationId1 = new AllocationID();
    final AllocationID allocationId2 = new AllocationID();
    final SlotOffer offer1 = new SlotOffer(allocationId1, 0, ResourceProfile.UNKNOWN);
    final JobMasterGateway jobMasterGateway = mock(JobMasterGateway.class);
    when(jobMasterGateway.registerTaskManager(any(String.class), eq(taskManagerLocation), eq(jobManagerLeaderId), any(Time.class))).thenReturn(FlinkCompletableFuture.<RegistrationResponse>completed(new JMTMRegistrationSuccess(jmResourceId, blobPort)));
    when(jobMasterGateway.getHostname()).thenReturn(jobManagerAddress);
    rpc.registerGateway(resourceManagerAddress, resourceManagerGateway);
    rpc.registerGateway(jobManagerAddress, jobMasterGateway);
    final LibraryCacheManager libraryCacheManager = mock(LibraryCacheManager.class);
    when(libraryCacheManager.getClassLoader(eq(jobId))).thenReturn(getClass().getClassLoader());
    final JobManagerConnection jobManagerConnection = new JobManagerConnection(jobId, jmResourceId, jobMasterGateway, jobManagerLeaderId, mock(TaskManagerActions.class), mock(CheckpointResponder.class), libraryCacheManager, mock(ResultPartitionConsumableNotifier.class), mock(PartitionProducerStateChecker.class));
    jobManagerTable.put(jobId, jobManagerConnection);
    try {
        final TaskExecutor taskManager = new TaskExecutor(taskManagerConfiguration, taskManagerLocation, rpc, mock(MemoryManager.class), mock(IOManager.class), mock(NetworkEnvironment.class), haServices, mock(HeartbeatServices.class, RETURNS_MOCKS), mock(MetricRegistry.class), mock(TaskManagerMetricGroup.class), mock(BroadcastVariableManager.class), mock(FileCache.class), taskSlotTable, jobManagerTable, jobLeaderService, testingFatalErrorHandler);
        taskManager.start();
        taskSlotTable.allocateSlot(0, jobId, allocationId1, Time.milliseconds(10000L));
        taskSlotTable.allocateSlot(1, jobId, allocationId2, Time.milliseconds(10000L));
        final JobVertexID jobVertexId = new JobVertexID();
        JobInformation jobInformation = new JobInformation(jobId, name.getMethodName(), new SerializedValue<>(new ExecutionConfig()), new Configuration(), Collections.<BlobKey>emptyList(), Collections.<URL>emptyList());
        TaskInformation taskInformation = new TaskInformation(jobVertexId, "test task", 1, 1, TestInvokable.class.getName(), new Configuration());
        SerializedValue<JobInformation> serializedJobInformation = new SerializedValue<>(jobInformation);
        SerializedValue<TaskInformation> serializedJobVertexInformation = new SerializedValue<>(taskInformation);
        final TaskDeploymentDescriptor tdd = new TaskDeploymentDescriptor(serializedJobInformation, serializedJobVertexInformation, new ExecutionAttemptID(), allocationId1, 0, 0, 0, null, Collections.<ResultPartitionDeploymentDescriptor>emptyList(), Collections.<InputGateDeploymentDescriptor>emptyList());
        CompletableFuture<Iterable<SlotOffer>> offerResultFuture = new FlinkCompletableFuture<>();
        // submit task first and then return acceptance response
        when(jobMasterGateway.offerSlots(any(ResourceID.class), any(Iterable.class), eq(jobManagerLeaderId), any(Time.class))).thenReturn(offerResultFuture);
        // we have to add the job after the TaskExecutor, because otherwise the service has not
        // been properly started. This will also offer the slots to the job master
        jobLeaderService.addJob(jobId, jobManagerAddress);
        verify(jobMasterGateway).offerSlots(any(ResourceID.class), any(Iterable.class), eq(jobManagerLeaderId), any(Time.class));
        // submit the task without having acknowledge the offered slots
        taskManager.submitTask(tdd, jobManagerLeaderId);
        // acknowledge the offered slots
        offerResultFuture.complete(Collections.singleton(offer1));
        verify(resourceManagerGateway).notifySlotAvailable(eq(resourceManagerLeaderId), eq(registrationId), eq(new SlotID(resourceId, 1)));
        assertTrue(taskSlotTable.existsActiveSlot(jobId, allocationId1));
        assertFalse(taskSlotTable.existsActiveSlot(jobId, allocationId2));
        assertTrue(taskSlotTable.isSlotFree(1));
        // check if a concurrent error occurred
        testingFatalErrorHandler.rethrowError();
    } finally {
        rpc.stopService();
    }
}

Example 5 with JobMasterGateway

use of org.apache.flink.runtime.jobmaster.JobMasterGateway in project flink by apache.

the class ResourceManager method disconnectJobManager.

/**
	 * Disconnects the job manager which is connected for the given job from the resource manager.
	 *
	 * @param jobId identifying the job whose leader shall be disconnected
	 */
protected void disconnectJobManager(JobID jobId, Exception cause) {
    JobManagerRegistration jobManagerRegistration = jobManagerRegistrations.remove(jobId);
    if (jobManagerRegistration != null) {
        log.info("Disconnect job manager {}@{} for job {} from the resource manager.", jobManagerRegistration.getLeaderID(), jobManagerRegistration.getJobManagerGateway().getAddress(), jobId);
        JobMasterGateway jobMasterGateway = jobManagerRegistration.getJobManagerGateway();
        // tell the job manager about the disconnect
        jobMasterGateway.disconnectResourceManager(jobManagerRegistration.getLeaderID(), getLeaderSessionId(), cause);
    } else {
        log.debug("There was no registered job manager for job {}.", jobId);
    }
}