zhijiangW commented on a change in pull request #9984: [FLINK-9495][kubernetes] Implement ResourceManager for Kubernetes. URL: https://github.com/apache/flink/pull/9984#discussion_r354682441
########## File path: flink-kubernetes/src/main/java/org/apache/flink/kubernetes/KubernetesResourceManager.java ########## @@ -0,0 +1,340 @@ +/* + * 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.kubernetes; + +import org.apache.flink.annotation.VisibleForTesting; +import org.apache.flink.configuration.Configuration; +import org.apache.flink.configuration.GlobalConfiguration; +import org.apache.flink.kubernetes.configuration.KubernetesConfigOptions; +import org.apache.flink.kubernetes.kubeclient.FlinkKubeClient; +import org.apache.flink.kubernetes.kubeclient.KubeClientFactory; +import org.apache.flink.kubernetes.kubeclient.TaskManagerPodParameter; +import org.apache.flink.kubernetes.kubeclient.resources.KubernetesPod; +import org.apache.flink.kubernetes.taskmanager.KubernetesTaskExecutorRunner; +import org.apache.flink.kubernetes.utils.Constants; +import org.apache.flink.kubernetes.utils.KubernetesUtils; +import org.apache.flink.runtime.clusterframework.ApplicationStatus; +import org.apache.flink.runtime.clusterframework.BootstrapTools; +import org.apache.flink.runtime.clusterframework.ContaineredTaskManagerParameters; +import org.apache.flink.runtime.clusterframework.types.ResourceID; +import org.apache.flink.runtime.clusterframework.types.ResourceProfile; +import org.apache.flink.runtime.entrypoint.ClusterInformation; +import org.apache.flink.runtime.entrypoint.parser.CommandLineOptions; +import org.apache.flink.runtime.heartbeat.HeartbeatServices; +import org.apache.flink.runtime.highavailability.HighAvailabilityServices; +import org.apache.flink.runtime.metrics.groups.ResourceManagerMetricGroup; +import org.apache.flink.runtime.resourcemanager.ActiveResourceManager; +import org.apache.flink.runtime.resourcemanager.JobLeaderIdService; +import org.apache.flink.runtime.resourcemanager.ResourceManager; +import org.apache.flink.runtime.resourcemanager.exceptions.ResourceManagerException; +import org.apache.flink.runtime.resourcemanager.slotmanager.SlotManager; +import org.apache.flink.runtime.rpc.FatalErrorHandler; +import org.apache.flink.runtime.rpc.RpcService; + +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import javax.annotation.Nullable; + +import java.io.File; +import java.util.Arrays; +import java.util.Collection; +import java.util.Collections; +import java.util.HashMap; +import java.util.List; +import java.util.Map; +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.atomic.AtomicLong; + +/** + * Kubernetes specific implementation of the {@link ResourceManager}. + */ +public class KubernetesResourceManager extends ActiveResourceManager<KubernetesWorkerNode> + implements FlinkKubeClient.PodCallbackHandler { + + private static final Logger LOG = LoggerFactory.getLogger(KubernetesResourceManager.class); + + /** The taskmanager pod name pattern is {clusterId}-{taskmanager}-{attemptId}-{podIndex}. */ + private static final String TASK_MANAGER_POD_FORMAT = "%s-taskmanager-%d-%d"; + + private final Map<ResourceID, KubernetesWorkerNode> workerNodes = new HashMap<>(); + + private final double defaultCpus; + + /** When ResourceManager failover, the max attempt should recover. */ + private final AtomicLong currentMaxAttemptId = new AtomicLong(0); + + /** Current max pod index. When creating a new pod, it should increase one. */ + private final AtomicLong currentMaxPodId = new AtomicLong(0); + + private final String clusterId; + + private final FlinkKubeClient kubeClient; + + private final ContaineredTaskManagerParameters taskManagerParameters; + + private final List<String> taskManagerStartCommand; + + /** The number of pods requested, but not yet granted. */ + private int numPendingPodRequests = 0; + + public KubernetesResourceManager( + RpcService rpcService, + String resourceManagerEndpointId, + ResourceID resourceId, + Configuration flinkConfig, + HighAvailabilityServices highAvailabilityServices, + HeartbeatServices heartbeatServices, + SlotManager slotManager, + JobLeaderIdService jobLeaderIdService, + ClusterInformation clusterInformation, + FatalErrorHandler fatalErrorHandler, + ResourceManagerMetricGroup resourceManagerMetricGroup) { + super( + flinkConfig, + System.getenv(), + rpcService, + resourceManagerEndpointId, + resourceId, + highAvailabilityServices, + heartbeatServices, + slotManager, + jobLeaderIdService, + clusterInformation, + fatalErrorHandler, + resourceManagerMetricGroup); + this.clusterId = flinkConfig.getString(KubernetesConfigOptions.CLUSTER_ID); + this.defaultCpus = flinkConfig.getDouble(KubernetesConfigOptions.TASK_MANAGER_CPU, numSlotsPerTaskManager); + + this.kubeClient = createFlinkKubeClient(); + + this.taskManagerParameters = + ContaineredTaskManagerParameters.create(flinkConfig, taskExecutorResourceSpec, numSlotsPerTaskManager); + + this.taskManagerStartCommand = getTaskManagerStartCommand(); + } + + @Override + protected Configuration loadClientConfiguration() { + return GlobalConfiguration.loadConfiguration(); + } + + @Override + protected void initialize() throws ResourceManagerException { + recoverWorkerNodesFromPreviousAttempts(); + + kubeClient.watchPodsAndDoCallback(getTaskManagerLabels(), this); + } + + @Override + public CompletableFuture<Void> onStop() { + // shut down all components + Throwable exception = null; + + try { + kubeClient.close(); + } catch (Throwable t) { + exception = t; + } + + return getStopTerminationFutureOrCompletedExceptionally(exception); + } + + @Override + protected void internalDeregisterApplication(ApplicationStatus finalStatus, @Nullable String diagnostics) { + LOG.info( + "Stopping kubernetes cluster, clusterId: {}, diagnostics: {}", + clusterId, + diagnostics == null ? "" : diagnostics); + kubeClient.stopAndCleanupCluster(clusterId); + } + + @Override + public Collection<ResourceProfile> startNewWorker(ResourceProfile resourceProfile) { + LOG.info("Starting new worker with resource profile, {}", resourceProfile); + if (!resourceProfilesPerWorker.iterator().next().isMatching(resourceProfile)) { + return Collections.emptyList(); + } + requestKubernetesPod(); + return resourceProfilesPerWorker; + } + + @Override + protected KubernetesWorkerNode workerStarted(ResourceID resourceID) { + return workerNodes.get(resourceID); + } + + @Override + public boolean stopWorker(final KubernetesWorkerNode worker) { + LOG.info("Stopping Worker {}.", worker.getResourceID()); + workerNodes.remove(worker.getResourceID()); + try { + kubeClient.stopPod(worker.getResourceID().toString()); + } catch (Exception e) { + kubeClient.handleException(e); + return false; + } + return true; + } + + @Override + public void onAdded(List<KubernetesPod> pods) { + runAsync(() -> { + for (KubernetesPod pod : pods) { + if (numPendingPodRequests > 0) { + numPendingPodRequests--; + final KubernetesWorkerNode worker = new KubernetesWorkerNode(new ResourceID(pod.getName())); + workerNodes.putIfAbsent(worker.getResourceID(), worker); + } + + log.info("Received new TaskManager pod: {} - Remaining pending pod requests: {}", + pod.getName(), numPendingPodRequests); + } + }); + } + + @Override + public void onModified(List<KubernetesPod> pods) { + runAsync(() -> pods.forEach(this::removePodIfTerminated)); + } + + @Override + public void onDeleted(List<KubernetesPod> pods) { + runAsync(() -> pods.forEach(this::removePodIfTerminated)); + } + + @Override + public void onError(List<KubernetesPod> pods) { + runAsync(() -> pods.forEach(this::removePodIfTerminated)); + } + + @VisibleForTesting + Map<ResourceID, KubernetesWorkerNode> getWorkerNodes() { + return workerNodes; + } + + private void recoverWorkerNodesFromPreviousAttempts() throws ResourceManagerException { + final List<KubernetesPod> podList = kubeClient.getPodsWithLabels(getTaskManagerLabels()); + for (KubernetesPod pod : podList) { + final KubernetesWorkerNode worker = new KubernetesWorkerNode(new ResourceID(pod.getName())); + workerNodes.put(worker.getResourceID(), worker); + final long attempt = worker.getAttempt(); + if (attempt > currentMaxAttemptId.get()) { + currentMaxAttemptId.set(attempt); + } + } + + log.info("Recovered {} pods from previous attempts, current attempt id is {}.", + workerNodes.size(), + currentMaxAttemptId.addAndGet(1)); + } + + private void requestKubernetesPod() { + numPendingPodRequests++; + + log.info("Requesting new TaskManager pod with <{},{}>. Number pending requests {}.", + defaultMemoryMB, + defaultCpus, + numPendingPodRequests); + + final String podName = String.format( + TASK_MANAGER_POD_FORMAT, + clusterId, + currentMaxAttemptId.get(), + currentMaxPodId.addAndGet(1)); + + final HashMap<String, String> env = new HashMap<>(); + env.put(Constants.ENV_FLINK_POD_NAME, podName); + env.putAll(taskManagerParameters.taskManagerEnv()); + + final TaskManagerPodParameter parameter = new TaskManagerPodParameter( + podName, + taskManagerStartCommand, + defaultMemoryMB, + defaultCpus, + env); + + log.info("TaskManager {} will be started with {}.", podName, taskExecutorResourceSpec); + kubeClient.createTaskManagerPod(parameter); + } + + /** + * Request new pod if pending pods cannot satisfy pending slot requests. + */ + private void requestKubernetesPodIfRequired() { + final int requiredTaskManagerSlots = getNumberRequiredTaskManagerSlots(); + final int pendingTaskManagerSlots = numPendingPodRequests * numSlotsPerTaskManager; + + if (requiredTaskManagerSlots > pendingTaskManagerSlots) { + requestKubernetesPod(); + } + } + + private void removePodIfTerminated(KubernetesPod pod) { + if (pod.isTerminated()) { + kubeClient.stopPod(pod.getName()); + final KubernetesWorkerNode kubernetesWorkerNode = workerNodes.remove(new ResourceID(pod.getName())); + if (kubernetesWorkerNode != null) { + requestKubernetesPodIfRequired(); + } + } + } + + private List<String> getTaskManagerStartCommand() { + + final String confDir = flinkConfig.getString(KubernetesConfigOptions.FLINK_CONF_DIR); + final boolean hasLogback = new File(confDir, Constants.CONFIG_FILE_LOGBACK_NAME).exists(); + final boolean hasLog4j = new File(confDir, Constants.CONFIG_FILE_LOG4J_NAME).exists(); + + final String logDir = flinkConfig.getString(KubernetesConfigOptions.FLINK_LOG_DIR); + + final String mainClassArgs = "--" + CommandLineOptions.CONFIG_DIR_OPTION.getLongOpt() + " " + + flinkConfig.getString(KubernetesConfigOptions.FLINK_CONF_DIR) + " " + + BootstrapTools.getDynamicProperties(flinkClientConfig, flinkConfig); + + final String command = KubernetesUtils.getTaskManagerStartCommand( + flinkConfig, + taskManagerParameters, + confDir, + logDir, + hasLogback, + hasLog4j, + KubernetesTaskExecutorRunner.class.getCanonicalName(), + mainClassArgs); + + return Arrays.asList("/bin/bash", "-c", command); + } + + /** + * Get task manager labels for the current Flink cluster. They could be used to watch the pods status. + * + * @return Task manager labels. + */ + private Map<String, String> getTaskManagerLabels() { + final Map<String, String> labels = new HashMap<>(); + labels.put(Constants.LABEL_TYPE_KEY, Constants.LABEL_TYPE_NATIVE_TYPE); + labels.put(Constants.LABEL_APP_KEY, clusterId); + labels.put(Constants.LABEL_COMPONENT_KEY, Constants.LABEL_COMPONENT_TASK_MANAGER); + return labels; + } + + protected FlinkKubeClient createFlinkKubeClient() { Review comment: Maybe not need this separate method with only one sentence. If you prefer to maintaining, it should be private method. ---------------------------------------------------------------- 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
