Search in sources :

Example 1 with AllocatedSlotInfo

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

the class DeclarativeSlotPoolService method createAllocatedSlotReport.

@Override
public AllocatedSlotReport createAllocatedSlotReport(ResourceID taskManagerId) {
    assertHasBeenStarted();
    final Collection<AllocatedSlotInfo> allocatedSlotInfos = new ArrayList<>();
    for (SlotInfo slotInfo : declarativeSlotPool.getAllSlotsInformation()) {
        if (slotInfo.getTaskManagerLocation().getResourceID().equals(taskManagerId)) {
            allocatedSlotInfos.add(new AllocatedSlotInfo(slotInfo.getPhysicalSlotNumber(), slotInfo.getAllocationId()));
        }
    }
    return new AllocatedSlotReport(jobId, allocatedSlotInfos);
}
Also used : AllocatedSlotReport(org.apache.flink.runtime.jobmaster.AllocatedSlotReport) ArrayList(java.util.ArrayList) AllocatedSlotInfo(org.apache.flink.runtime.jobmaster.AllocatedSlotInfo) SlotInfo(org.apache.flink.runtime.jobmaster.SlotInfo) AllocatedSlotInfo(org.apache.flink.runtime.jobmaster.AllocatedSlotInfo)

Example 2 with AllocatedSlotInfo

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

the class TaskExecutorTest method testJobManagerBecomesUnreachableTriggersDisconnect.

@Test
public void testJobManagerBecomesUnreachableTriggersDisconnect() throws Exception {
    final ResourceID jmResourceId = ResourceID.generate();
    runJobManagerHeartbeatTest(jmResourceId, failedRpcEnabledHeartbeatServices, jobMasterGatewayBuilder -> jobMasterGatewayBuilder.setTaskManagerHeartbeatFunction((resourceID, taskExecutorToJobManagerHeartbeatPayload) -> FutureUtils.completedExceptionally(new RecipientUnreachableException("sender", "recipient", "job manager is unreachable."))), (ignoredTaskExecutorResourceId, taskExecutorGateway, allocationId) -> taskExecutorGateway.heartbeatFromJobManager(jmResourceId, new AllocatedSlotReport(jobId, Collections.singleton(new AllocatedSlotInfo(0, allocationId)))));
}
Also used : Arrays(java.util.Arrays) Tuple3(org.apache.flink.api.java.tuple.Tuple3) TaskSlotUtils.createDefaultTimerService(org.apache.flink.runtime.taskexecutor.slot.TaskSlotUtils.createDefaultTimerService) MemorySize(org.apache.flink.configuration.MemorySize) NetUtils(org.apache.flink.util.NetUtils) InetAddress(java.net.InetAddress) IOManagerAsync(org.apache.flink.runtime.io.disk.iomanager.IOManagerAsync) ResultPartitionID(org.apache.flink.runtime.io.network.partition.ResultPartitionID) SettableLeaderRetrievalService(org.apache.flink.runtime.leaderretrieval.SettableLeaderRetrievalService) TestingFatalErrorHandler(org.apache.flink.runtime.util.TestingFatalErrorHandler) TaskExecutorPartitionTracker(org.apache.flink.runtime.io.network.partition.TaskExecutorPartitionTracker) FunctionUtils(org.apache.flink.util.function.FunctionUtils) Matchers.nullValue(org.hamcrest.Matchers.nullValue) SlotID(org.apache.flink.runtime.clusterframework.types.SlotID) TestingJobMasterGatewayBuilder(org.apache.flink.runtime.jobmaster.utils.TestingJobMasterGatewayBuilder) Builder(org.apache.flink.runtime.taskexecutor.TaskSubmissionTestEnvironment.Builder) Matchers.notNullValue(org.hamcrest.Matchers.notNullValue) Failure(org.apache.flink.runtime.registration.RegistrationResponse.Failure) NoOpTaskExecutorBlobService(org.apache.flink.runtime.blob.NoOpTaskExecutorBlobService) TestingTaskExecutorPartitionTracker(org.apache.flink.runtime.io.network.partition.TestingTaskExecutorPartitionTracker) NoOpMetricRegistry(org.apache.flink.runtime.metrics.NoOpMetricRegistry) BlockingQueue(java.util.concurrent.BlockingQueue) Matchers.startsWith(org.hamcrest.Matchers.startsWith) HeartbeatServices(org.apache.flink.runtime.heartbeat.HeartbeatServices) TaskExecutorStateChangelogStoragesManager(org.apache.flink.runtime.state.TaskExecutorStateChangelogStoragesManager) SlotOffer(org.apache.flink.runtime.taskexecutor.slot.SlotOffer) Matchers.instanceOf(org.hamcrest.Matchers.instanceOf) ArrayBlockingQueue(java.util.concurrent.ArrayBlockingQueue) CountDownLatch(java.util.concurrent.CountDownLatch) TimeUtils(org.apache.flink.util.TimeUtils) Matchers.contains(org.hamcrest.Matchers.contains) Matchers.is(org.hamcrest.Matchers.is) Matchers.containsString(org.hamcrest.Matchers.containsString) Time(org.apache.flink.api.common.time.Time) TaskExecutorRegistration(org.apache.flink.runtime.resourcemanager.TaskExecutorRegistration) TaskExecutorPartitionTrackerImpl(org.apache.flink.runtime.io.network.partition.TaskExecutorPartitionTrackerImpl) FlinkException(org.apache.flink.util.FlinkException) TaskSlotTableImpl(org.apache.flink.runtime.taskexecutor.slot.TaskSlotTableImpl) MemoryManager(org.apache.flink.runtime.memory.MemoryManager) TaskSubmissionException(org.apache.flink.runtime.taskexecutor.exceptions.TaskSubmissionException) Callable(java.util.concurrent.Callable) DEFAULT_RESOURCE_PROFILE(org.apache.flink.runtime.taskexecutor.slot.TaskSlotUtils.DEFAULT_RESOURCE_PROFILE) TestingJobMasterGateway(org.apache.flink.runtime.jobmaster.utils.TestingJobMasterGateway) TestingTaskSlotTable(org.apache.flink.runtime.taskexecutor.slot.TestingTaskSlotTable) ArrayList(java.util.ArrayList) TaskManagerOptions(org.apache.flink.configuration.TaskManagerOptions) TestingClassLoaderLease(org.apache.flink.runtime.execution.librarycache.TestingClassLoaderLease) FutureUtils(org.apache.flink.util.concurrent.FutureUtils) TestName(org.junit.rules.TestName) ScheduledExecutorService(java.util.concurrent.ScheduledExecutorService) Matchers.hasSize(org.hamcrest.Matchers.hasSize) TriConsumer(org.apache.flink.util.function.TriConsumer) TestFileUtils(org.apache.flink.testutils.TestFileUtils) Before(org.junit.Before) RetryingRegistrationConfiguration(org.apache.flink.runtime.registration.RetryingRegistrationConfiguration) LocalUnresolvedTaskManagerLocation(org.apache.flink.runtime.taskmanager.LocalUnresolvedTaskManagerLocation) SlotNotFoundException(org.apache.flink.runtime.taskexecutor.slot.SlotNotFoundException) TriConsumerWithException(org.apache.flink.util.function.TriConsumerWithException) Assert.assertTrue(org.junit.Assert.assertTrue) Test(org.junit.Test) IOException(java.io.IOException) InstanceID(org.apache.flink.runtime.instance.InstanceID) File(java.io.File) ExecutionException(java.util.concurrent.ExecutionException) Executors(org.apache.flink.util.concurrent.Executors) ThreadSafeTaskSlotTable(org.apache.flink.runtime.taskexecutor.slot.ThreadSafeTaskSlotTable) JobID(org.apache.flink.api.common.JobID) UnregisteredMetricGroups(org.apache.flink.runtime.metrics.groups.UnregisteredMetricGroups) Task(org.apache.flink.runtime.taskmanager.Task) Assert.assertNull(org.junit.Assert.assertNull) UnresolvedTaskManagerLocation(org.apache.flink.runtime.taskmanager.UnresolvedTaskManagerLocation) ArrayDeque(java.util.ArrayDeque) TestingHighAvailabilityServices(org.apache.flink.runtime.highavailability.TestingHighAvailabilityServices) Assert.assertEquals(org.junit.Assert.assertEquals) CPUResource(org.apache.flink.api.common.resources.CPUResource) MultiShotLatch(org.apache.flink.core.testutils.MultiShotLatch) Deadline(org.apache.flink.api.common.time.Deadline) ClusterPartitionReport(org.apache.flink.runtime.taskexecutor.partition.ClusterPartitionReport) IntStream.range(java.util.stream.IntStream.range) RegistrationResponse(org.apache.flink.runtime.registration.RegistrationResponse) TestingRpcService(org.apache.flink.runtime.rpc.TestingRpcService) ShuffleEnvironment(org.apache.flink.runtime.shuffle.ShuffleEnvironment) IOManager(org.apache.flink.runtime.io.disk.iomanager.IOManager) BlockingNoOpInvokable(org.apache.flink.runtime.testtasks.BlockingNoOpInvokable) TimeoutException(java.util.concurrent.TimeoutException) ExceptionUtils(org.apache.flink.util.ExceptionUtils) Lists(org.apache.flink.shaded.guava30.com.google.common.collect.Lists) Assert.assertThat(org.junit.Assert.assertThat) After(org.junit.After) TestLogger(org.apache.flink.util.TestLogger) Assert.fail(org.junit.Assert.fail) TransientBlobKey(org.apache.flink.runtime.blob.TransientBlobKey) RegistrationTimeoutException(org.apache.flink.runtime.taskexecutor.exceptions.RegistrationTimeoutException) Collection(java.util.Collection) KvStateRegistry(org.apache.flink.runtime.query.KvStateRegistry) ResourceManagerId(org.apache.flink.runtime.resourcemanager.ResourceManagerId) CompletionException(java.util.concurrent.CompletionException) UUID(java.util.UUID) NettyShuffleEnvironment(org.apache.flink.runtime.io.network.NettyShuffleEnvironment) IntermediateDataSetID(org.apache.flink.runtime.jobgraph.IntermediateDataSetID) TaskManagerException(org.apache.flink.runtime.taskexecutor.exceptions.TaskManagerException) Collectors(java.util.stream.Collectors) Acknowledge(org.apache.flink.runtime.messages.Acknowledge) ResourceProfile(org.apache.flink.runtime.clusterframework.types.ResourceProfile) ExecutorUtils(org.apache.flink.util.ExecutorUtils) TaskSlotUtils.createTotalResourceProfile(org.apache.flink.runtime.taskexecutor.slot.TaskSlotUtils.createTotalResourceProfile) List(java.util.List) Matchers.containsInAnyOrder(org.hamcrest.Matchers.containsInAnyOrder) Matchers.equalTo(org.hamcrest.Matchers.equalTo) Queue(java.util.Queue) AllocationID(org.apache.flink.runtime.clusterframework.types.AllocationID) AllocatedSlotInfo(org.apache.flink.runtime.jobmaster.AllocatedSlotInfo) OneShotLatch(org.apache.flink.core.testutils.OneShotLatch) LeaderRetrievalListener(org.apache.flink.runtime.leaderretrieval.LeaderRetrievalListener) TaskInvokable(org.apache.flink.runtime.jobgraph.tasks.TaskInvokable) TaskExecutorLocalStateStoresManager(org.apache.flink.runtime.state.TaskExecutorLocalStateStoresManager) JMTMRegistrationSuccess(org.apache.flink.runtime.jobmaster.JMTMRegistrationSuccess) CompletableFuture(java.util.concurrent.CompletableFuture) TaskDeploymentDescriptorBuilder(org.apache.flink.runtime.deployment.TaskDeploymentDescriptorBuilder) TaskDeploymentDescriptor(org.apache.flink.runtime.deployment.TaskDeploymentDescriptor) NettyShuffleEnvironmentOptions(org.apache.flink.configuration.NettyShuffleEnvironmentOptions) TaskSlotTable(org.apache.flink.runtime.taskexecutor.slot.TaskSlotTable) ExternalResourceInfoProvider(org.apache.flink.runtime.externalresource.ExternalResourceInfoProvider) ConfigConstants(org.apache.flink.configuration.ConfigConstants) ClusterInformation(org.apache.flink.runtime.entrypoint.ClusterInformation) ResourceID(org.apache.flink.runtime.clusterframework.types.ResourceID) Nonnull(javax.annotation.Nonnull) TaskSlotUtils(org.apache.flink.runtime.taskexecutor.slot.TaskSlotUtils) JMTMRegistrationRejection(org.apache.flink.runtime.jobmaster.JMTMRegistrationRejection) Matchers.empty(org.hamcrest.Matchers.empty) NettyShuffleEnvironmentBuilder(org.apache.flink.runtime.io.network.NettyShuffleEnvironmentBuilder) Assert.assertNotNull(org.junit.Assert.assertNotNull) Configuration(org.apache.flink.configuration.Configuration) TestingHeartbeatServices(org.apache.flink.runtime.heartbeat.TestingHeartbeatServices) TaskManagerMetricGroup(org.apache.flink.runtime.metrics.groups.TaskManagerMetricGroup) Reference(org.apache.flink.util.Reference) RpcUtils(org.apache.flink.runtime.rpc.RpcUtils) AllocatedSlotReport(org.apache.flink.runtime.jobmaster.AllocatedSlotReport) TimeUnit(java.util.concurrent.TimeUnit) Consumer(java.util.function.Consumer) ExecutionAttemptID(org.apache.flink.runtime.executiongraph.ExecutionAttemptID) TestingResourceManagerGateway(org.apache.flink.runtime.resourcemanager.utils.TestingResourceManagerGateway) Rule(org.junit.Rule) UnregisteredMetricGroups.createUnregisteredTaskManagerMetricGroup(org.apache.flink.runtime.metrics.groups.UnregisteredMetricGroups.createUnregisteredTaskManagerMetricGroup) CommonTestUtils(org.apache.flink.runtime.testutils.CommonTestUtils) Collections(java.util.Collections) TemporaryFolder(org.junit.rules.TemporaryFolder) RecipientUnreachableException(org.apache.flink.runtime.rpc.exceptions.RecipientUnreachableException) NoOpInvokable(org.apache.flink.runtime.testtasks.NoOpInvokable) AllocatedSlotReport(org.apache.flink.runtime.jobmaster.AllocatedSlotReport) ResourceID(org.apache.flink.runtime.clusterframework.types.ResourceID) RecipientUnreachableException(org.apache.flink.runtime.rpc.exceptions.RecipientUnreachableException) AllocatedSlotInfo(org.apache.flink.runtime.jobmaster.AllocatedSlotInfo) Test(org.junit.Test)

Example 3 with AllocatedSlotInfo

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

the class TaskExecutorTest method testSyncSlotsWithJobMasterByHeartbeat.

/**
 * Tests that the TaskExecutor syncs its slots view with the JobMaster's view via the
 * AllocatedSlotReport reported by the heartbeat (See FLINK-11059).
 */
@Test
public void testSyncSlotsWithJobMasterByHeartbeat() throws Exception {
    final CountDownLatch activeSlots = new CountDownLatch(2);
    final TaskSlotTable<Task> taskSlotTable = new ActivateSlotNotifyingTaskSlotTable(2, activeSlots);
    final TaskManagerServices taskManagerServices = new TaskManagerServicesBuilder().setTaskSlotTable(taskSlotTable).build();
    final TaskExecutor taskExecutor = createTaskExecutor(taskManagerServices);
    final TestingResourceManagerGateway testingResourceManagerGateway = new TestingResourceManagerGateway();
    final BlockingQueue<AllocationID> allocationsNotifiedFree = new ArrayBlockingQueue<>(2);
    OneShotLatch initialSlotReporting = new OneShotLatch();
    testingResourceManagerGateway.setSendSlotReportFunction(resourceIDInstanceIDSlotReportTuple3 -> {
        initialSlotReporting.trigger();
        return CompletableFuture.completedFuture(Acknowledge.get());
    });
    testingResourceManagerGateway.setNotifySlotAvailableConsumer(instanceIDSlotIDAllocationIDTuple3 -> allocationsNotifiedFree.offer(instanceIDSlotIDAllocationIDTuple3.f2));
    rpc.registerGateway(testingResourceManagerGateway.getAddress(), testingResourceManagerGateway);
    resourceManagerLeaderRetriever.notifyListener(testingResourceManagerGateway.getAddress(), testingResourceManagerGateway.getFencingToken().toUUID());
    final BlockingQueue<AllocationID> failedSlotFutures = new ArrayBlockingQueue<>(2);
    final ResourceID jobManagerResourceId = ResourceID.generate();
    final TestingJobMasterGateway jobMasterGateway = new TestingJobMasterGatewayBuilder().setFailSlotConsumer((resourceID, allocationID, throwable) -> failedSlotFutures.offer(allocationID)).setOfferSlotsFunction((resourceID, slotOffers) -> CompletableFuture.completedFuture(new ArrayList<>(slotOffers))).setRegisterTaskManagerFunction((ignoredJobId, ignoredTaskManagerRegistrationInformation) -> CompletableFuture.completedFuture(new JMTMRegistrationSuccess(jobManagerResourceId))).build();
    final String jobManagerAddress = jobMasterGateway.getAddress();
    rpc.registerGateway(jobManagerAddress, jobMasterGateway);
    jobManagerLeaderRetriever.notifyListener(jobManagerAddress, jobMasterGateway.getFencingToken().toUUID());
    taskExecutor.start();
    try {
        final TaskExecutorGateway taskExecutorGateway = taskExecutor.getSelfGateway(TaskExecutorGateway.class);
        initialSlotReporting.await();
        final AllocationID allocationIdInBoth = new AllocationID();
        final AllocationID allocationIdOnlyInJM = new AllocationID();
        final AllocationID allocationIdOnlyInTM = new AllocationID();
        taskExecutorGateway.requestSlot(new SlotID(taskExecutor.getResourceID(), 0), jobId, allocationIdInBoth, ResourceProfile.ZERO, "foobar", testingResourceManagerGateway.getFencingToken(), timeout);
        taskExecutorGateway.requestSlot(new SlotID(taskExecutor.getResourceID(), 1), jobId, allocationIdOnlyInTM, ResourceProfile.ZERO, "foobar", testingResourceManagerGateway.getFencingToken(), timeout);
        activeSlots.await();
        List<AllocatedSlotInfo> allocatedSlotInfos = Arrays.asList(new AllocatedSlotInfo(0, allocationIdInBoth), new AllocatedSlotInfo(1, allocationIdOnlyInJM));
        AllocatedSlotReport allocatedSlotReport = new AllocatedSlotReport(jobId, allocatedSlotInfos);
        taskExecutorGateway.heartbeatFromJobManager(jobManagerResourceId, allocatedSlotReport);
        assertThat(failedSlotFutures.take(), is(allocationIdOnlyInJM));
        assertThat(allocationsNotifiedFree.take(), is(allocationIdOnlyInTM));
        assertThat(failedSlotFutures.poll(5L, TimeUnit.MILLISECONDS), nullValue());
        assertThat(allocationsNotifiedFree.poll(5L, TimeUnit.MILLISECONDS), nullValue());
    } finally {
        RpcUtils.terminateRpcEndpoint(taskExecutor, timeout);
    }
}
Also used : Arrays(java.util.Arrays) Tuple3(org.apache.flink.api.java.tuple.Tuple3) TaskSlotUtils.createDefaultTimerService(org.apache.flink.runtime.taskexecutor.slot.TaskSlotUtils.createDefaultTimerService) MemorySize(org.apache.flink.configuration.MemorySize) NetUtils(org.apache.flink.util.NetUtils) InetAddress(java.net.InetAddress) IOManagerAsync(org.apache.flink.runtime.io.disk.iomanager.IOManagerAsync) ResultPartitionID(org.apache.flink.runtime.io.network.partition.ResultPartitionID) SettableLeaderRetrievalService(org.apache.flink.runtime.leaderretrieval.SettableLeaderRetrievalService) TestingFatalErrorHandler(org.apache.flink.runtime.util.TestingFatalErrorHandler) TaskExecutorPartitionTracker(org.apache.flink.runtime.io.network.partition.TaskExecutorPartitionTracker) FunctionUtils(org.apache.flink.util.function.FunctionUtils) Matchers.nullValue(org.hamcrest.Matchers.nullValue) SlotID(org.apache.flink.runtime.clusterframework.types.SlotID) TestingJobMasterGatewayBuilder(org.apache.flink.runtime.jobmaster.utils.TestingJobMasterGatewayBuilder) Builder(org.apache.flink.runtime.taskexecutor.TaskSubmissionTestEnvironment.Builder) Matchers.notNullValue(org.hamcrest.Matchers.notNullValue) Failure(org.apache.flink.runtime.registration.RegistrationResponse.Failure) NoOpTaskExecutorBlobService(org.apache.flink.runtime.blob.NoOpTaskExecutorBlobService) TestingTaskExecutorPartitionTracker(org.apache.flink.runtime.io.network.partition.TestingTaskExecutorPartitionTracker) NoOpMetricRegistry(org.apache.flink.runtime.metrics.NoOpMetricRegistry) BlockingQueue(java.util.concurrent.BlockingQueue) Matchers.startsWith(org.hamcrest.Matchers.startsWith) HeartbeatServices(org.apache.flink.runtime.heartbeat.HeartbeatServices) TaskExecutorStateChangelogStoragesManager(org.apache.flink.runtime.state.TaskExecutorStateChangelogStoragesManager) SlotOffer(org.apache.flink.runtime.taskexecutor.slot.SlotOffer) Matchers.instanceOf(org.hamcrest.Matchers.instanceOf) ArrayBlockingQueue(java.util.concurrent.ArrayBlockingQueue) CountDownLatch(java.util.concurrent.CountDownLatch) TimeUtils(org.apache.flink.util.TimeUtils) Matchers.contains(org.hamcrest.Matchers.contains) Matchers.is(org.hamcrest.Matchers.is) Matchers.containsString(org.hamcrest.Matchers.containsString) Time(org.apache.flink.api.common.time.Time) TaskExecutorRegistration(org.apache.flink.runtime.resourcemanager.TaskExecutorRegistration) TaskExecutorPartitionTrackerImpl(org.apache.flink.runtime.io.network.partition.TaskExecutorPartitionTrackerImpl) FlinkException(org.apache.flink.util.FlinkException) TaskSlotTableImpl(org.apache.flink.runtime.taskexecutor.slot.TaskSlotTableImpl) MemoryManager(org.apache.flink.runtime.memory.MemoryManager) TaskSubmissionException(org.apache.flink.runtime.taskexecutor.exceptions.TaskSubmissionException) Callable(java.util.concurrent.Callable) DEFAULT_RESOURCE_PROFILE(org.apache.flink.runtime.taskexecutor.slot.TaskSlotUtils.DEFAULT_RESOURCE_PROFILE) TestingJobMasterGateway(org.apache.flink.runtime.jobmaster.utils.TestingJobMasterGateway) TestingTaskSlotTable(org.apache.flink.runtime.taskexecutor.slot.TestingTaskSlotTable) ArrayList(java.util.ArrayList) TaskManagerOptions(org.apache.flink.configuration.TaskManagerOptions) TestingClassLoaderLease(org.apache.flink.runtime.execution.librarycache.TestingClassLoaderLease) FutureUtils(org.apache.flink.util.concurrent.FutureUtils) TestName(org.junit.rules.TestName) ScheduledExecutorService(java.util.concurrent.ScheduledExecutorService) Matchers.hasSize(org.hamcrest.Matchers.hasSize) TriConsumer(org.apache.flink.util.function.TriConsumer) TestFileUtils(org.apache.flink.testutils.TestFileUtils) Before(org.junit.Before) RetryingRegistrationConfiguration(org.apache.flink.runtime.registration.RetryingRegistrationConfiguration) LocalUnresolvedTaskManagerLocation(org.apache.flink.runtime.taskmanager.LocalUnresolvedTaskManagerLocation) SlotNotFoundException(org.apache.flink.runtime.taskexecutor.slot.SlotNotFoundException) TriConsumerWithException(org.apache.flink.util.function.TriConsumerWithException) Assert.assertTrue(org.junit.Assert.assertTrue) Test(org.junit.Test) IOException(java.io.IOException) InstanceID(org.apache.flink.runtime.instance.InstanceID) File(java.io.File) ExecutionException(java.util.concurrent.ExecutionException) Executors(org.apache.flink.util.concurrent.Executors) ThreadSafeTaskSlotTable(org.apache.flink.runtime.taskexecutor.slot.ThreadSafeTaskSlotTable) JobID(org.apache.flink.api.common.JobID) UnregisteredMetricGroups(org.apache.flink.runtime.metrics.groups.UnregisteredMetricGroups) Task(org.apache.flink.runtime.taskmanager.Task) Assert.assertNull(org.junit.Assert.assertNull) UnresolvedTaskManagerLocation(org.apache.flink.runtime.taskmanager.UnresolvedTaskManagerLocation) ArrayDeque(java.util.ArrayDeque) TestingHighAvailabilityServices(org.apache.flink.runtime.highavailability.TestingHighAvailabilityServices) Assert.assertEquals(org.junit.Assert.assertEquals) CPUResource(org.apache.flink.api.common.resources.CPUResource) MultiShotLatch(org.apache.flink.core.testutils.MultiShotLatch) Deadline(org.apache.flink.api.common.time.Deadline) ClusterPartitionReport(org.apache.flink.runtime.taskexecutor.partition.ClusterPartitionReport) IntStream.range(java.util.stream.IntStream.range) RegistrationResponse(org.apache.flink.runtime.registration.RegistrationResponse) TestingRpcService(org.apache.flink.runtime.rpc.TestingRpcService) ShuffleEnvironment(org.apache.flink.runtime.shuffle.ShuffleEnvironment) IOManager(org.apache.flink.runtime.io.disk.iomanager.IOManager) BlockingNoOpInvokable(org.apache.flink.runtime.testtasks.BlockingNoOpInvokable) TimeoutException(java.util.concurrent.TimeoutException) ExceptionUtils(org.apache.flink.util.ExceptionUtils) Lists(org.apache.flink.shaded.guava30.com.google.common.collect.Lists) Assert.assertThat(org.junit.Assert.assertThat) After(org.junit.After) TestLogger(org.apache.flink.util.TestLogger) Assert.fail(org.junit.Assert.fail) TransientBlobKey(org.apache.flink.runtime.blob.TransientBlobKey) RegistrationTimeoutException(org.apache.flink.runtime.taskexecutor.exceptions.RegistrationTimeoutException) Collection(java.util.Collection) KvStateRegistry(org.apache.flink.runtime.query.KvStateRegistry) ResourceManagerId(org.apache.flink.runtime.resourcemanager.ResourceManagerId) CompletionException(java.util.concurrent.CompletionException) UUID(java.util.UUID) NettyShuffleEnvironment(org.apache.flink.runtime.io.network.NettyShuffleEnvironment) IntermediateDataSetID(org.apache.flink.runtime.jobgraph.IntermediateDataSetID) TaskManagerException(org.apache.flink.runtime.taskexecutor.exceptions.TaskManagerException) Collectors(java.util.stream.Collectors) Acknowledge(org.apache.flink.runtime.messages.Acknowledge) ResourceProfile(org.apache.flink.runtime.clusterframework.types.ResourceProfile) ExecutorUtils(org.apache.flink.util.ExecutorUtils) TaskSlotUtils.createTotalResourceProfile(org.apache.flink.runtime.taskexecutor.slot.TaskSlotUtils.createTotalResourceProfile) List(java.util.List) Matchers.containsInAnyOrder(org.hamcrest.Matchers.containsInAnyOrder) Matchers.equalTo(org.hamcrest.Matchers.equalTo) Queue(java.util.Queue) AllocationID(org.apache.flink.runtime.clusterframework.types.AllocationID) AllocatedSlotInfo(org.apache.flink.runtime.jobmaster.AllocatedSlotInfo) OneShotLatch(org.apache.flink.core.testutils.OneShotLatch) LeaderRetrievalListener(org.apache.flink.runtime.leaderretrieval.LeaderRetrievalListener) TaskInvokable(org.apache.flink.runtime.jobgraph.tasks.TaskInvokable) TaskExecutorLocalStateStoresManager(org.apache.flink.runtime.state.TaskExecutorLocalStateStoresManager) JMTMRegistrationSuccess(org.apache.flink.runtime.jobmaster.JMTMRegistrationSuccess) CompletableFuture(java.util.concurrent.CompletableFuture) TaskDeploymentDescriptorBuilder(org.apache.flink.runtime.deployment.TaskDeploymentDescriptorBuilder) TaskDeploymentDescriptor(org.apache.flink.runtime.deployment.TaskDeploymentDescriptor) NettyShuffleEnvironmentOptions(org.apache.flink.configuration.NettyShuffleEnvironmentOptions) TaskSlotTable(org.apache.flink.runtime.taskexecutor.slot.TaskSlotTable) ExternalResourceInfoProvider(org.apache.flink.runtime.externalresource.ExternalResourceInfoProvider) ConfigConstants(org.apache.flink.configuration.ConfigConstants) ClusterInformation(org.apache.flink.runtime.entrypoint.ClusterInformation) ResourceID(org.apache.flink.runtime.clusterframework.types.ResourceID) Nonnull(javax.annotation.Nonnull) TaskSlotUtils(org.apache.flink.runtime.taskexecutor.slot.TaskSlotUtils) JMTMRegistrationRejection(org.apache.flink.runtime.jobmaster.JMTMRegistrationRejection) Matchers.empty(org.hamcrest.Matchers.empty) NettyShuffleEnvironmentBuilder(org.apache.flink.runtime.io.network.NettyShuffleEnvironmentBuilder) Assert.assertNotNull(org.junit.Assert.assertNotNull) Configuration(org.apache.flink.configuration.Configuration) TestingHeartbeatServices(org.apache.flink.runtime.heartbeat.TestingHeartbeatServices) TaskManagerMetricGroup(org.apache.flink.runtime.metrics.groups.TaskManagerMetricGroup) Reference(org.apache.flink.util.Reference) RpcUtils(org.apache.flink.runtime.rpc.RpcUtils) AllocatedSlotReport(org.apache.flink.runtime.jobmaster.AllocatedSlotReport) TimeUnit(java.util.concurrent.TimeUnit) Consumer(java.util.function.Consumer) ExecutionAttemptID(org.apache.flink.runtime.executiongraph.ExecutionAttemptID) TestingResourceManagerGateway(org.apache.flink.runtime.resourcemanager.utils.TestingResourceManagerGateway) Rule(org.junit.Rule) UnregisteredMetricGroups.createUnregisteredTaskManagerMetricGroup(org.apache.flink.runtime.metrics.groups.UnregisteredMetricGroups.createUnregisteredTaskManagerMetricGroup) CommonTestUtils(org.apache.flink.runtime.testutils.CommonTestUtils) Collections(java.util.Collections) TemporaryFolder(org.junit.rules.TemporaryFolder) RecipientUnreachableException(org.apache.flink.runtime.rpc.exceptions.RecipientUnreachableException) NoOpInvokable(org.apache.flink.runtime.testtasks.NoOpInvokable) JMTMRegistrationSuccess(org.apache.flink.runtime.jobmaster.JMTMRegistrationSuccess) Task(org.apache.flink.runtime.taskmanager.Task) AllocatedSlotReport(org.apache.flink.runtime.jobmaster.AllocatedSlotReport) AllocationID(org.apache.flink.runtime.clusterframework.types.AllocationID) ArrayList(java.util.ArrayList) Matchers.containsString(org.hamcrest.Matchers.containsString) CountDownLatch(java.util.concurrent.CountDownLatch) SlotID(org.apache.flink.runtime.clusterframework.types.SlotID) TestingJobMasterGateway(org.apache.flink.runtime.jobmaster.utils.TestingJobMasterGateway) ArrayBlockingQueue(java.util.concurrent.ArrayBlockingQueue) ResourceID(org.apache.flink.runtime.clusterframework.types.ResourceID) TestingJobMasterGatewayBuilder(org.apache.flink.runtime.jobmaster.utils.TestingJobMasterGatewayBuilder) TestingResourceManagerGateway(org.apache.flink.runtime.resourcemanager.utils.TestingResourceManagerGateway) OneShotLatch(org.apache.flink.core.testutils.OneShotLatch) AllocatedSlotInfo(org.apache.flink.runtime.jobmaster.AllocatedSlotInfo) Test(org.junit.Test)

Example 4 with AllocatedSlotInfo

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

the class TaskExecutorExecutionDeploymentReconciliationTest method testDeployedExecutionReporting.

@Test
public void testDeployedExecutionReporting() throws Exception {
    final OneShotLatch slotOfferLatch = new OneShotLatch();
    final BlockingQueue<Set<ExecutionAttemptID>> deployedExecutionsQueue = new ArrayBlockingQueue<>(3);
    final CompletableFuture<Void> taskFinishedFuture = new CompletableFuture<>();
    final ResourceID jobManagerResourceId = ResourceID.generate();
    final TestingJobMasterGateway jobMasterGateway = setupJobManagerGateway(slotOfferLatch, deployedExecutionsQueue, taskFinishedFuture, jobManagerResourceId);
    final CompletableFuture<SlotReport> initialSlotReportFuture = new CompletableFuture<>();
    final TestingResourceManagerGateway testingResourceManagerGateway = setupResourceManagerGateway(initialSlotReportFuture);
    final TaskManagerServices taskManagerServices = new TaskManagerServicesBuilder().setTaskSlotTable(TaskSlotUtils.createTaskSlotTable(1, timeout)).setShuffleEnvironment(new NettyShuffleEnvironmentBuilder().build()).build();
    final TestingTaskExecutor taskExecutor = createTestingTaskExecutor(taskManagerServices);
    try {
        taskExecutor.start();
        taskExecutor.waitUntilStarted();
        final TaskExecutorGateway taskExecutorGateway = taskExecutor.getSelfGateway(TaskExecutorGateway.class);
        final TaskDeploymentDescriptor taskDeploymentDescriptor = createTaskDeploymentDescriptor(jobId);
        connectComponentsAndRequestSlot(jobMasterGateway, testingResourceManagerGateway, taskExecutorGateway, taskManagerServices.getJobLeaderService(), initialSlotReportFuture, taskDeploymentDescriptor.getAllocationId());
        TestingInvokable.sync = new BlockerSync();
        // This ensures TM has been successfully registered to JM.
        slotOfferLatch.await();
        AllocatedSlotReport slotAllocationReport = new AllocatedSlotReport(jobId, Collections.singleton(new AllocatedSlotInfo(0, taskDeploymentDescriptor.getAllocationId())));
        // nothing as deployed, so the deployment report should be empty
        taskExecutorGateway.heartbeatFromJobManager(jobManagerResourceId, slotAllocationReport);
        assertThat(deployedExecutionsQueue.take(), hasSize(0));
        taskExecutorGateway.submitTask(taskDeploymentDescriptor, jobMasterGateway.getFencingToken(), timeout).get();
        TestingInvokable.sync.awaitBlocker();
        // task is deployed, so the deployment report should contain it
        taskExecutorGateway.heartbeatFromJobManager(jobManagerResourceId, slotAllocationReport);
        assertThat(deployedExecutionsQueue.take(), hasItem(taskDeploymentDescriptor.getExecutionAttemptId()));
        TestingInvokable.sync.releaseBlocker();
        // task is finished ans was cleaned up, so the deployment report should be empty
        taskFinishedFuture.get();
        taskExecutorGateway.heartbeatFromJobManager(jobManagerResourceId, slotAllocationReport);
        assertThat(deployedExecutionsQueue.take(), hasSize(0));
    } finally {
        RpcUtils.terminateRpcEndpoint(taskExecutor, timeout);
    }
}
Also used : BlockerSync(org.apache.flink.core.testutils.BlockerSync) Set(java.util.Set) AllocatedSlotReport(org.apache.flink.runtime.jobmaster.AllocatedSlotReport) AllocatedSlotReport(org.apache.flink.runtime.jobmaster.AllocatedSlotReport) NettyShuffleEnvironmentBuilder(org.apache.flink.runtime.io.network.NettyShuffleEnvironmentBuilder) CompletableFuture(java.util.concurrent.CompletableFuture) TestingJobMasterGateway(org.apache.flink.runtime.jobmaster.utils.TestingJobMasterGateway) ArrayBlockingQueue(java.util.concurrent.ArrayBlockingQueue) ResourceID(org.apache.flink.runtime.clusterframework.types.ResourceID) TestingResourceManagerGateway(org.apache.flink.runtime.resourcemanager.utils.TestingResourceManagerGateway) OneShotLatch(org.apache.flink.core.testutils.OneShotLatch) TaskDeploymentDescriptor(org.apache.flink.runtime.deployment.TaskDeploymentDescriptor) AllocatedSlotInfo(org.apache.flink.runtime.jobmaster.AllocatedSlotInfo) Test(org.junit.Test)

Aggregations

AllocatedSlotInfo (org.apache.flink.runtime.jobmaster.AllocatedSlotInfo)4 AllocatedSlotReport (org.apache.flink.runtime.jobmaster.AllocatedSlotReport)4 ArrayList (java.util.ArrayList)3 ArrayBlockingQueue (java.util.concurrent.ArrayBlockingQueue)3 CompletableFuture (java.util.concurrent.CompletableFuture)3 OneShotLatch (org.apache.flink.core.testutils.OneShotLatch)3 ResourceID (org.apache.flink.runtime.clusterframework.types.ResourceID)3 TaskDeploymentDescriptor (org.apache.flink.runtime.deployment.TaskDeploymentDescriptor)3 NettyShuffleEnvironmentBuilder (org.apache.flink.runtime.io.network.NettyShuffleEnvironmentBuilder)3 TestingJobMasterGateway (org.apache.flink.runtime.jobmaster.utils.TestingJobMasterGateway)3 TestingResourceManagerGateway (org.apache.flink.runtime.resourcemanager.utils.TestingResourceManagerGateway)3 Test (org.junit.Test)3 File (java.io.File)2 IOException (java.io.IOException)2 InetAddress (java.net.InetAddress)2 ArrayDeque (java.util.ArrayDeque)2 Arrays (java.util.Arrays)2 Collection (java.util.Collection)2 Collections (java.util.Collections)2 List (java.util.List)2