List of usage examples for org.apache.hadoop.security Credentials writeTokenStorageToStream
public void writeTokenStorageToStream(DataOutputStream os) throws IOException
From source file:io.hops.tensorflow.Client.java
License:Apache License
private ContainerLaunchContext createContainerLaunchContext(GetNewApplicationResponse appResponse) throws IOException { FileSystem fs = FileSystem.get(conf); ApplicationId appId = appResponse.getApplicationId(); DistributedCacheList dcl = populateDistributedCache(fs, appId); Map<String, LocalResource> localResources = prepareLocalResources(fs, appId, dcl); Map<String, String> launchEnv = setupLaunchEnv(); // Set the executable command for the application master Vector<CharSequence> vargs = new Vector<>(30); LOG.info("Setting up app master command"); vargs.add(Environment.JAVA_HOME.$$() + "/bin/java"); vargs.add("-Xmx" + amMemory + "m"); vargs.add(appMasterMainClass);//from w w w. ja v a 2s . c om if (python != null) { vargs.add(newArg(PYTHON, python)); } vargs.add(newArg(MEMORY, String.valueOf(memory))); vargs.add(newArg(VCORES, String.valueOf(vcores))); vargs.add(newArg(GPUS, String.valueOf(gpus))); if (protocol != null) { vargs.add(newArg(PROTOCOL, protocol)); } // vargs.add(newArg(PRIORITY, String.valueOf(priority))); vargs.add(newArg(ALLOCATION_TIMEOUT, String.valueOf(allocationTimeout / 1000))); vargs.add(newArg(ApplicationMasterArguments.MAIN_RELATIVE, mainRelativePath)); if (arguments != null) { vargs.add(newArg(ARGS, StringUtils.join(arguments, " "))); } vargs.add(newArg(WORKERS, Integer.toString(numWorkers))); vargs.add(newArg(PSES, Integer.toString(numPses))); for (Map.Entry<String, String> entry : environment.entrySet()) { vargs.add(newArg(ENV, entry.getKey() + "=" + entry.getValue())); } if (tensorboard) { vargs.add("--" + TENSORBOARD); } if (debugFlag) { vargs.add("--" + DEBUG); } // Add log redirect params vargs.add("1>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/AppMaster.stdout"); vargs.add("2>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/AppMaster.stderr"); // Get final command StringBuilder command = new StringBuilder(); for (CharSequence str : vargs) { command.append(str).append(" "); } LOG.info("Completed setting up app master command " + command.toString()); List<String> commands = new ArrayList<>(); commands.add(command.toString()); // Set up the container launch context for the application master ContainerLaunchContext amContainer = ContainerLaunchContext.newInstance(localResources, launchEnv, commands, null, null, null); // Setup security tokens if (UserGroupInformation.isSecurityEnabled()) { Credentials credentials = new Credentials(); String tokenRenewer = conf.get(YarnConfiguration.RM_PRINCIPAL); if (tokenRenewer == null || tokenRenewer.length() == 0) { throw new IOException("Can't get Master Kerberos principal for the RM to use as renewer"); } // For now: only getting tokens for the default file-system. final Token<?> tokens[] = fs.addDelegationTokens(tokenRenewer, credentials); if (tokens != null) { for (Token<?> token : tokens) { LOG.info("Got dt for " + fs.getUri() + "; " + token); } } DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); ByteBuffer fsTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); amContainer.setTokens(fsTokens); } return amContainer; }
From source file:me.haosdent.noya.ApplicationMaster.java
License:Apache License
/** * Main run function for the application master * * @throws org.apache.hadoop.yarn.exceptions.YarnException * @throws java.io.IOException/*w ww. j av a2s . c o m*/ */ @SuppressWarnings({ "unchecked" }) public void run() throws YarnException, IOException { LOG.info("Starting ApplicationMaster"); try { publishApplicationAttemptEvent(timelineClient, appAttemptID.toString(), DSEvent.DS_APP_ATTEMPT_START); } catch (Exception e) { LOG.error("App Attempt start event coud not be pulished for " + appAttemptID.toString(), e); } Credentials credentials = UserGroupInformation.getCurrentUser().getCredentials(); DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); // Now remove the AM->RM token so that containers cannot access it. Iterator<Token<?>> iter = credentials.getAllTokens().iterator(); LOG.info("Executing with tokens:"); while (iter.hasNext()) { Token<?> token = iter.next(); LOG.info(token); if (token.getKind().equals(AMRMTokenIdentifier.KIND_NAME)) { iter.remove(); } } allTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); // Create appSubmitterUgi and add original tokens to it String appSubmitterUserName = System.getenv(ApplicationConstants.Environment.USER.name()); appSubmitterUgi = UserGroupInformation.createRemoteUser(appSubmitterUserName); appSubmitterUgi.addCredentials(credentials); AMRMClientAsync.CallbackHandler allocListener = new RMCallbackHandler(); amRMClient = AMRMClientAsync.createAMRMClientAsync(1000, allocListener); amRMClient.init(conf); amRMClient.start(); containerListener = createNMCallbackHandler(); nmClientAsync = new NMClientAsyncImpl(containerListener); nmClientAsync.init(conf); nmClientAsync.start(); // Setup local RPC Server to accept status requests directly from clients // TODO need to setup a protocol for client to be able to communicate to // the RPC server // TODO use the rpc port info to register with the RM for the client to // send requests to this app master // Register self with ResourceManager // This will start heartbeating to the RM appMasterHostname = NetUtils.getHostname(); RegisterApplicationMasterResponse response = amRMClient.registerApplicationMaster(appMasterHostname, appMasterRpcPort, appMasterTrackingUrl); // Dump out information about cluster capability as seen by the // resource manager int maxMem = response.getMaximumResourceCapability().getMemory(); LOG.info("Max mem capabililty of resources in this cluster " + maxMem); int maxVCores = response.getMaximumResourceCapability().getVirtualCores(); LOG.info("Max vcores capabililty of resources in this cluster " + maxVCores); // A resource ask cannot exceed the max. if (containerMemory > maxMem) { LOG.info("Container memory specified above max threshold of cluster." + " Using max value." + ", specified=" + containerMemory + ", max=" + maxMem); containerMemory = maxMem; } if (containerVirtualCores > maxVCores) { LOG.info("Container virtual cores specified above max threshold of cluster." + " Using max value." + ", specified=" + containerVirtualCores + ", max=" + maxVCores); containerVirtualCores = maxVCores; } List<Container> previousAMRunningContainers = response.getContainersFromPreviousAttempts(); LOG.info("Received " + previousAMRunningContainers.size() + " previous AM's running containers on AM registration."); numAllocatedContainers.addAndGet(previousAMRunningContainers.size()); int numTotalContainersToRequest = numTotalContainers - previousAMRunningContainers.size(); // Setup ask for containers from RM // Send request for containers to RM // Until we get our fully allocated quota, we keep on polling RM for // containers // Keep looping until all the containers are launched and shell script // executed on them ( regardless of success/failure). for (int i = 0; i < numTotalContainersToRequest; ++i) { ContainerRequest containerAsk = setupContainerAskForRM(); amRMClient.addContainerRequest(containerAsk); } numRequestedContainers.set(numTotalContainersToRequest); try { publishApplicationAttemptEvent(timelineClient, appAttemptID.toString(), DSEvent.DS_APP_ATTEMPT_END); } catch (Exception e) { LOG.error("App Attempt start event coud not be pulished for " + appAttemptID.toString(), e); } }
From source file:me.haosdent.noya.Client.java
License:Apache License
/** * Main run function for the client/*from w ww .j a va 2 s.co m*/ * * @return true if application completed successfully * * @throws java.io.IOException * @throws org.apache.hadoop.yarn.exceptions.YarnException */ public boolean run() throws IOException, YarnException { LOG.info("Running Client"); yarnClient.start(); YarnClusterMetrics clusterMetrics = yarnClient.getYarnClusterMetrics(); LOG.info("Got Cluster metric info from ASM" + ", numNodeManagers=" + clusterMetrics.getNumNodeManagers()); List<NodeReport> clusterNodeReports = yarnClient.getNodeReports(NodeState.RUNNING); LOG.info("Got Cluster node info from ASM"); for (NodeReport node : clusterNodeReports) { LOG.info("Got node report from ASM for" + ", nodeId=" + node.getNodeId() + ", nodeAddress" + node.getHttpAddress() + ", nodeRackName" + node.getRackName() + ", nodeNumContainers" + node.getNumContainers()); } QueueInfo queueInfo = yarnClient.getQueueInfo(this.amQueue); LOG.info("Queue info" + ", queueName=" + queueInfo.getQueueName() + ", queueCurrentCapacity=" + queueInfo.getCurrentCapacity() + ", queueMaxCapacity=" + queueInfo.getMaximumCapacity() + ", queueApplicationCount=" + queueInfo.getApplications().size() + ", queueChildQueueCount=" + queueInfo.getChildQueues().size()); List<QueueUserACLInfo> listAclInfo = yarnClient.getQueueAclsInfo(); for (QueueUserACLInfo aclInfo : listAclInfo) { for (QueueACL userAcl : aclInfo.getUserAcls()) { LOG.info("User ACL Info for Queue" + ", queueName=" + aclInfo.getQueueName() + ", userAcl=" + userAcl.name()); } } // Get a new application id YarnClientApplication app = yarnClient.createApplication(); GetNewApplicationResponse appResponse = app.getNewApplicationResponse(); // TODO get min/max resource capabilities from RM and change memory ask if needed // If we do not have min/max, we may not be able to correctly request // the required resources from the RM for the app master // Memory ask has to be a multiple of min and less than max. // Dump out information about cluster capability as seen by the resource manager int maxMem = appResponse.getMaximumResourceCapability().getMemory(); LOG.info("Max mem capabililty of resources in this cluster " + maxMem); // A resource ask cannot exceed the max. if (amMemory > maxMem) { LOG.info("AM memory specified above max threshold of cluster. Using max value." + ", specified=" + amMemory + ", max=" + maxMem); amMemory = maxMem; } int maxVCores = appResponse.getMaximumResourceCapability().getVirtualCores(); LOG.info("Max virtual cores capabililty of resources in this cluster " + maxVCores); if (amVCores > maxVCores) { LOG.info("AM virtual cores specified above max threshold of cluster. " + "Using max value." + ", specified=" + amVCores + ", max=" + maxVCores); amVCores = maxVCores; } // set the application name ApplicationSubmissionContext appContext = app.getApplicationSubmissionContext(); ApplicationId appId = appContext.getApplicationId(); //appContext.setKeepContainersAcrossApplicationAttempts(keepContainers); appContext.setApplicationName(appName); // Set up the container launch context for the application master ContainerLaunchContext amContainer = Records.newRecord(ContainerLaunchContext.class); // set local resources for the application master // local files or archives as needed // In this scenario, the jar file for the application master is part of the local resources Map<String, LocalResource> localResources = new HashMap<String, LocalResource>(); LOG.info("Copy App Master jar from local filesystem and add to local environment"); // Copy the application master jar to the filesystem // Create a local resource to point to the destination jar path FileSystem fs = FileSystem.get(conf); addToLocalResources(fs, appMasterJar, appMasterJarPath, appId.toString(), localResources, null); // Set the log4j properties if needed if (!log4jPropFile.isEmpty()) { addToLocalResources(fs, log4jPropFile, log4jPath, appId.toString(), localResources, null); } // The shell script has to be made available on the final container(s) // where it will be executed. // To do this, we need to first copy into the filesystem that is visible // to the yarn framework. // We do not need to set this as a local resource for the application // master as the application master does not need it. String hdfsShellScriptLocation = ""; long hdfsShellScriptLen = 0; long hdfsShellScriptTimestamp = 0; if (!shellScriptPath.isEmpty()) { Path shellSrc = new Path(shellScriptPath); String shellPathSuffix = appName + "/" + appId.toString() + "/" + SCRIPT_PATH; Path shellDst = new Path(fs.getHomeDirectory(), shellPathSuffix); fs.copyFromLocalFile(false, true, shellSrc, shellDst); hdfsShellScriptLocation = shellDst.toUri().toString(); FileStatus shellFileStatus = fs.getFileStatus(shellDst); hdfsShellScriptLen = shellFileStatus.getLen(); hdfsShellScriptTimestamp = shellFileStatus.getModificationTime(); } if (!shellCommand.isEmpty()) { addToLocalResources(fs, null, shellCommandPath, appId.toString(), localResources, shellCommand); } if (shellArgs.length > 0) { addToLocalResources(fs, null, shellArgsPath, appId.toString(), localResources, StringUtils.join(shellArgs, " ")); } // Set local resource info into app master container launch context amContainer.setLocalResources(localResources); // Set the necessary security tokens as needed //amContainer.setContainerTokens(containerToken); // Set the env variables to be setup in the env where the application master will be run LOG.info("Set the environment for the application master"); Map<String, String> env = new HashMap<String, String>(); // put location of shell script into env // using the env info, the application master will create the correct local resource for the // eventual containers that will be launched to execute the shell scripts env.put(DSConstants.DISTRIBUTEDSHELLSCRIPTLOCATION, hdfsShellScriptLocation); env.put(DSConstants.DISTRIBUTEDSHELLSCRIPTTIMESTAMP, Long.toString(hdfsShellScriptTimestamp)); env.put(DSConstants.DISTRIBUTEDSHELLSCRIPTLEN, Long.toString(hdfsShellScriptLen)); // Add AppMaster.jar location to classpath // At some point we should not be required to add // the hadoop specific classpaths to the env. // It should be provided out of the box. // For now setting all required classpaths including // the classpath to "." for the application jar StringBuilder classPathEnv = new StringBuilder(ApplicationConstants.Environment.CLASSPATH.$$()) .append(ApplicationConstants.CLASS_PATH_SEPARATOR).append("./*"); for (String c : conf.getStrings(YarnConfiguration.YARN_APPLICATION_CLASSPATH, ApplicationConstants.DEFAULT_YARN_CROSS_PLATFORM_APPLICATION_CLASSPATH)) { classPathEnv.append(ApplicationConstants.CLASS_PATH_SEPARATOR); classPathEnv.append(c.trim()); } classPathEnv.append(ApplicationConstants.CLASS_PATH_SEPARATOR).append("./log4j.properties"); // add the runtime classpath needed for tests to work if (conf.getBoolean(YarnConfiguration.IS_MINI_YARN_CLUSTER, false)) { classPathEnv.append(':'); classPathEnv.append(System.getProperty("java.class.path")); } env.put("CLASSPATH", classPathEnv.toString()); amContainer.setEnvironment(env); // Set the necessary command to execute the application master Vector<CharSequence> vargs = new Vector<CharSequence>(30); // Set java executable command LOG.info("Setting up app master command"); vargs.add(ApplicationConstants.Environment.JAVA_HOME.$$() + "/bin/java"); // Set Xmx based on am memory size vargs.add("-Xmx" + amMemory + "m"); // Set class name vargs.add(appMasterMainClass); // Set params for Application Master vargs.add("--container_memory " + String.valueOf(containerMemory)); vargs.add("--container_vcores " + String.valueOf(containerVirtualCores)); vargs.add("--num_containers " + String.valueOf(numContainers)); vargs.add("--priority " + String.valueOf(shellCmdPriority)); for (Map.Entry<String, String> entry : shellEnv.entrySet()) { vargs.add("--shell_env " + entry.getKey() + "=" + entry.getValue()); } if (debugFlag) { vargs.add("--debug"); } vargs.add("1>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/AppMaster.stdout"); vargs.add("2>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/AppMaster.stderr"); // Get final commmand StringBuilder command = new StringBuilder(); for (CharSequence str : vargs) { command.append(str).append(" "); } LOG.info("Completed setting up app master command " + command.toString()); List<String> commands = new ArrayList<String>(); //commands.add(command.toString()); commands.add("echo 'hello' >/tmp/yarn_test"); amContainer.setCommands(commands); // Set up resource type requirements // For now, both memory and vcores are supported, so we set memory and // vcores requirements Resource capability = Records.newRecord(Resource.class); capability.setMemory(amMemory); capability.setVirtualCores(amVCores); appContext.setResource(capability); // Service data is a binary blob that can be passed to the application // Not needed in this scenario // amContainer.setServiceData(serviceData); // Setup security tokens if (UserGroupInformation.isSecurityEnabled()) { Credentials credentials = new Credentials(); String tokenRenewer = conf.get(YarnConfiguration.RM_PRINCIPAL); if (tokenRenewer == null || tokenRenewer.length() == 0) { throw new IOException("Can't get Master Kerberos principal for the RM to use as renewer"); } // For now, only getting tokens for the default file-system. final Token<?> tokens[] = fs.addDelegationTokens(tokenRenewer, credentials); if (tokens != null) { for (Token<?> token : tokens) { LOG.info("Got dt for " + fs.getUri() + "; " + token); } } DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); ByteBuffer fsTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); amContainer.setTokens(fsTokens); } appContext.setAMContainerSpec(amContainer); // Set the priority for the application master Priority pri = Records.newRecord(Priority.class); // TODO - what is the range for priority? how to decide? pri.setPriority(amPriority); appContext.setPriority(pri); // Set the queue to which this application is to be submitted in the RM appContext.setQueue(amQueue); // Submit the application to the applications manager // SubmitApplicationResponse submitResp = applicationsManager.submitApplication(appRequest); // Ignore the response as either a valid response object is returned on success // or an exception thrown to denote some form of a failure LOG.info("Submitting application to ASM"); yarnClient.submitApplication(appContext); // TODO // Try submitting the same request again // app submission failure? // Monitor the application return monitorApplication(appId); }
From source file:ml.shifu.guagua.yarn.GuaguaAppMaster.java
License:Apache License
/** * Populate allTokens with the tokens received *///from www.j a va2 s. co m private void getAllTokens() throws IOException { Credentials credentials = UserGroupInformation.getCurrentUser().getCredentials(); DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); // Now remove the AM->RM token so that containers cannot access it. Iterator<Token<?>> iter = credentials.getAllTokens().iterator(); while (iter.hasNext()) { Token<?> token = iter.next(); if (LOG.isDebugEnabled()) { LOG.debug("Token type : {}", token.getKind()); } if (token.getKind().equals(AMRMTokenIdentifier.KIND_NAME)) { iter.remove(); } } this.allTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); }
From source file:ml.shifu.guagua.yarn.GuaguaYarnClient.java
License:Apache License
/** * Set delegation tokens for AM container * //from w w w .jav a2s .c o m * @param amContainer * AM container */ private void setToken(ContainerLaunchContext amContainer) throws IOException { // Setup security tokens if (UserGroupInformation.isSecurityEnabled()) { Credentials credentials = new Credentials(); String tokenRenewer = getConf().get(YarnConfiguration.RM_PRINCIPAL); if (tokenRenewer == null || tokenRenewer.length() == 0) { throw new IOException("Can't get Master Kerberos principal for the RM to use as renewer"); } FileSystem fs = FileSystem.get(getConf()); // For now, only getting tokens for the default file-system. final Token<?>[] tokens = fs.addDelegationTokens(tokenRenewer, credentials); if (tokens != null) { for (Token<?> token : tokens) { LOG.info("Got dt for " + fs.getUri() + "; " + token); } } DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); ByteBuffer fsTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); amContainer.setTokens(fsTokens); } }
From source file:org.apache.flink.yarn.ApplicationMaster.java
License:Apache License
private void run() throws Exception { //Utils.logFilesInCurrentDirectory(LOG); // Initialize clients to ResourceManager and NodeManagers Configuration conf = Utils.initializeYarnConfiguration(); FileSystem fs = FileSystem.get(conf); Map<String, String> envs = System.getenv(); final String currDir = envs.get(Environment.PWD.key()); final String logDirs = envs.get(Environment.LOG_DIRS.key()); final String ownHostname = envs.get(Environment.NM_HOST.key()); final String appId = envs.get(Client.ENV_APP_ID); final String clientHomeDir = envs.get(Client.ENV_CLIENT_HOME_DIR); final String applicationMasterHost = envs.get(Environment.NM_HOST.key()); final String remoteFlinkJarPath = envs.get(Client.FLINK_JAR_PATH); final String shipListString = envs.get(Client.ENV_CLIENT_SHIP_FILES); final String yarnClientUsername = envs.get(Client.ENV_CLIENT_USERNAME); final int taskManagerCount = Integer.valueOf(envs.get(Client.ENV_TM_COUNT)); final int memoryPerTaskManager = Integer.valueOf(envs.get(Client.ENV_TM_MEMORY)); final int coresPerTaskManager = Integer.valueOf(envs.get(Client.ENV_TM_CORES)); int heapLimit = Utils.calculateHeapSize(memoryPerTaskManager); if (currDir == null) { throw new RuntimeException("Current directory unknown"); }// www .j av a 2 s.c om if (ownHostname == null) { throw new RuntimeException("Own hostname (" + Environment.NM_HOST + ") not set."); } LOG.info("Working directory " + currDir); // load Flink configuration. Utils.getFlinkConfiguration(currDir); final String localWebInterfaceDir = currDir + "/resources/" + ConfigConstants.DEFAULT_JOB_MANAGER_WEB_PATH_NAME; // Update yaml conf -> set jobManager address to this machine's address. FileInputStream fis = new FileInputStream(currDir + "/flink-conf.yaml"); BufferedReader br = new BufferedReader(new InputStreamReader(fis)); Writer output = new BufferedWriter(new FileWriter(currDir + "/flink-conf-modified.yaml")); String line; while ((line = br.readLine()) != null) { if (line.contains(ConfigConstants.JOB_MANAGER_IPC_ADDRESS_KEY)) { output.append(ConfigConstants.JOB_MANAGER_IPC_ADDRESS_KEY + ": " + ownHostname + "\n"); } else if (line.contains(ConfigConstants.JOB_MANAGER_WEB_ROOT_PATH_KEY)) { output.append(ConfigConstants.JOB_MANAGER_WEB_ROOT_PATH_KEY + ": " + "\n"); } else { output.append(line + "\n"); } } // just to make sure. output.append(ConfigConstants.JOB_MANAGER_IPC_ADDRESS_KEY + ": " + ownHostname + "\n"); output.append(ConfigConstants.JOB_MANAGER_WEB_ROOT_PATH_KEY + ": " + localWebInterfaceDir + "\n"); output.append(ConfigConstants.JOB_MANAGER_WEB_LOG_PATH_KEY + ": " + logDirs + "\n"); output.close(); br.close(); File newConf = new File(currDir + "/flink-conf-modified.yaml"); if (!newConf.exists()) { LOG.warn("modified yaml does not exist!"); } Utils.copyJarContents("resources/" + ConfigConstants.DEFAULT_JOB_MANAGER_WEB_PATH_NAME, ApplicationMaster.class.getProtectionDomain().getCodeSource().getLocation().getPath()); JobManager jm; { String pathToNepheleConfig = currDir + "/flink-conf-modified.yaml"; String[] args = { "-executionMode", "cluster", "-configDir", pathToNepheleConfig }; // start the job manager jm = JobManager.initialize(args); // Start info server for jobmanager jm.startInfoServer(); } AMRMClient<ContainerRequest> rmClient = AMRMClient.createAMRMClient(); rmClient.init(conf); rmClient.start(); NMClient nmClient = NMClient.createNMClient(); nmClient.init(conf); nmClient.start(); // Register with ResourceManager LOG.info("registering ApplicationMaster"); rmClient.registerApplicationMaster(applicationMasterHost, 0, "http://" + applicationMasterHost + ":" + GlobalConfiguration.getString(ConfigConstants.JOB_MANAGER_WEB_PORT_KEY, "undefined")); // Priority for worker containers - priorities are intra-application Priority priority = Records.newRecord(Priority.class); priority.setPriority(0); // Resource requirements for worker containers Resource capability = Records.newRecord(Resource.class); capability.setMemory(memoryPerTaskManager); capability.setVirtualCores(coresPerTaskManager); // Make container requests to ResourceManager for (int i = 0; i < taskManagerCount; ++i) { ContainerRequest containerAsk = new ContainerRequest(capability, null, null, priority); LOG.info("Requesting TaskManager container " + i); rmClient.addContainerRequest(containerAsk); } LocalResource flinkJar = Records.newRecord(LocalResource.class); LocalResource flinkConf = Records.newRecord(LocalResource.class); // register Flink Jar with remote HDFS final Path remoteJarPath = new Path(remoteFlinkJarPath); Utils.registerLocalResource(fs, remoteJarPath, flinkJar); // register conf with local fs. Path remoteConfPath = Utils.setupLocalResource(conf, fs, appId, new Path("file://" + currDir + "/flink-conf-modified.yaml"), flinkConf, new Path(clientHomeDir)); LOG.info("Prepared localresource for modified yaml: " + flinkConf); boolean hasLog4j = new File(currDir + "/log4j.properties").exists(); // prepare the files to ship LocalResource[] remoteShipRsc = null; String[] remoteShipPaths = shipListString.split(","); if (!shipListString.isEmpty()) { remoteShipRsc = new LocalResource[remoteShipPaths.length]; { // scope for i int i = 0; for (String remoteShipPathStr : remoteShipPaths) { if (remoteShipPathStr == null || remoteShipPathStr.isEmpty()) { continue; } remoteShipRsc[i] = Records.newRecord(LocalResource.class); Path remoteShipPath = new Path(remoteShipPathStr); Utils.registerLocalResource(fs, remoteShipPath, remoteShipRsc[i]); i++; } } } // respect custom JVM options in the YAML file final String javaOpts = GlobalConfiguration.getString(ConfigConstants.FLINK_JVM_OPTIONS, ""); // Obtain allocated containers and launch int allocatedContainers = 0; int completedContainers = 0; while (allocatedContainers < taskManagerCount) { AllocateResponse response = rmClient.allocate(0); for (Container container : response.getAllocatedContainers()) { LOG.info("Got new Container for TM " + container.getId() + " on host " + container.getNodeId().getHost()); ++allocatedContainers; // Launch container by create ContainerLaunchContext ContainerLaunchContext ctx = Records.newRecord(ContainerLaunchContext.class); String tmCommand = "$JAVA_HOME/bin/java -Xmx" + heapLimit + "m " + javaOpts; if (hasLog4j) { tmCommand += " -Dlog.file=\"" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/taskmanager-log4j.log\" -Dlog4j.configuration=file:log4j.properties"; } tmCommand += " org.apache.flink.yarn.YarnTaskManagerRunner -configDir . " + " 1>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/taskmanager-stdout.log" + " 2>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/taskmanager-stderr.log"; ctx.setCommands(Collections.singletonList(tmCommand)); LOG.info("Starting TM with command=" + tmCommand); // copy resources to the TaskManagers. Map<String, LocalResource> localResources = new HashMap<String, LocalResource>(2); localResources.put("flink.jar", flinkJar); localResources.put("flink-conf.yaml", flinkConf); // add ship resources if (!shipListString.isEmpty()) { Preconditions.checkNotNull(remoteShipRsc); for (int i = 0; i < remoteShipPaths.length; i++) { localResources.put(new Path(remoteShipPaths[i]).getName(), remoteShipRsc[i]); } } ctx.setLocalResources(localResources); // Setup CLASSPATH for Container (=TaskTracker) Map<String, String> containerEnv = new HashMap<String, String>(); Utils.setupEnv(conf, containerEnv); //add flink.jar to class path. containerEnv.put(Client.ENV_CLIENT_USERNAME, yarnClientUsername); ctx.setEnvironment(containerEnv); UserGroupInformation user = UserGroupInformation.getCurrentUser(); try { Credentials credentials = user.getCredentials(); DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); ByteBuffer securityTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); ctx.setTokens(securityTokens); } catch (IOException e) { LOG.warn("Getting current user info failed when trying to launch the container" + e.getMessage()); } LOG.info("Launching container " + allocatedContainers); nmClient.startContainer(container, ctx); } for (ContainerStatus status : response.getCompletedContainersStatuses()) { ++completedContainers; LOG.info("Completed container (while allocating) " + status.getContainerId() + ". Total Completed:" + completedContainers); LOG.info("Diagnostics " + status.getDiagnostics()); } Thread.sleep(100); } // Now wait for containers to complete while (completedContainers < taskManagerCount) { AllocateResponse response = rmClient.allocate(completedContainers / taskManagerCount); for (ContainerStatus status : response.getCompletedContainersStatuses()) { ++completedContainers; LOG.info("Completed container " + status.getContainerId() + ". Total Completed:" + completedContainers); LOG.info("Diagnostics " + status.getDiagnostics()); } Thread.sleep(5000); } LOG.info("Shutting down JobManager"); jm.shutdown(); // Un-register with ResourceManager rmClient.unregisterApplicationMaster(FinalApplicationStatus.SUCCEEDED, "", ""); }
From source file:org.apache.flink.yarn.appMaster.ApplicationMaster.java
License:Apache License
/** * Run a Thread to allocate new containers until taskManagerCount * is correct again.// w w w. j a va 2 s .c om */ private void allocateOutstandingContainer(StringBuffer containerDiag) throws Exception { // respect custom JVM options in the YAML file final String javaOpts = GlobalConfiguration.getString(ConfigConstants.FLINK_JVM_OPTIONS, ""); int allocatedContainers = 0; while (allocatedContainers < taskManagerCount) { AllocateResponse response = rmClient.allocate(0); for (Container container : response.getAllocatedContainers()) { LOG.info("Got new Container for TM " + container.getId() + " on host " + container.getNodeId().getHost()); ++allocatedContainers; // Launch container by create ContainerLaunchContext ContainerLaunchContext ctx = Records.newRecord(ContainerLaunchContext.class); String tmCommand = "$JAVA_HOME/bin/java -Xmx" + heapLimit + "m " + javaOpts; if (hasLogback) { tmCommand += " -Dlog.file=\"" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/taskmanager-logback" + ".log\" -Dlogback.configurationFile=file:logback.xml"; } tmCommand += " " + YarnTaskManagerRunner.class.getName() + " -configDir . " + " 1>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/taskmanager-stdout.log" + " 2>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + "/taskmanager-stderr.log"; ctx.setCommands(Collections.singletonList(tmCommand)); LOG.info("Starting TM with command=" + tmCommand); ctx.setLocalResources(taskManagerLocalResources); // Setup CLASSPATH for Container (=TaskTracker) Map<String, String> containerEnv = new HashMap<String, String>(); Utils.setupEnv(conf, containerEnv); //add flink.jar to class path. containerEnv.put(Client.ENV_CLIENT_USERNAME, yarnClientUsername); ctx.setEnvironment(containerEnv); UserGroupInformation user = UserGroupInformation.getCurrentUser(); try { Credentials credentials = user.getCredentials(); DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); ByteBuffer securityTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); ctx.setTokens(securityTokens); } catch (IOException e) { LOG.warn("Getting current user info failed when trying to launch the container", e); } LOG.info("Launching container " + allocatedContainers); nmClient.startContainer(container, ctx); } for (ContainerStatus status : response.getCompletedContainersStatuses()) { ++completedContainers; LOG.info("Completed container (while allocating) " + status.getContainerId() + ". Total Completed:" + completedContainers); LOG.info("Diagnostics " + status.getDiagnostics()); // status. logDeadContainer(status, containerDiag); } Thread.sleep(100); } }
From source file:org.apache.flink.yarn.YarnApplicationMasterRunner.java
License:Apache License
/** * Creates the launch context, which describes how to bring up a TaskManager process in * an allocated YARN container./* w w w.j a va 2s.c o m*/ * * <p>This code is extremely YARN specific and registers all the resources that the TaskManager * needs (such as JAR file, config file, ...) and all environment variables in a YARN * container launch context. The launch context then ensures that those resources will be * copied into the containers transient working directory. * * <p>We do this work before we start the ResourceManager actor in order to fail early if * any of the operations here fail. * * @param flinkConfig * The Flink configuration object. * @param yarnConfig * The YARN configuration object. * @param env * The environment variables. * @param tmParams * The TaskManager container memory parameters. * @param taskManagerConfig * The configuration for the TaskManagers. * @param workingDirectory * The current application master container's working directory. * @param taskManagerMainClass * The class with the main method. * @param log * The logger. * * @return The launch context for the TaskManager processes. * * @throws Exception Thrown if teh launch context could not be created, for example if * the resources could not be copied. */ public static ContainerLaunchContext createTaskManagerContext(Configuration flinkConfig, YarnConfiguration yarnConfig, Map<String, String> env, ContaineredTaskManagerParameters tmParams, Configuration taskManagerConfig, String workingDirectory, Class<?> taskManagerMainClass, Logger log) throws Exception { log.info("Setting up resources for TaskManagers"); // get and validate all relevant variables String remoteFlinkJarPath = env.get(YarnConfigKeys.FLINK_JAR_PATH); require(remoteFlinkJarPath != null, "Environment variable %s not set", YarnConfigKeys.FLINK_JAR_PATH); String appId = env.get(YarnConfigKeys.ENV_APP_ID); require(appId != null, "Environment variable %s not set", YarnConfigKeys.ENV_APP_ID); String clientHomeDir = env.get(YarnConfigKeys.ENV_CLIENT_HOME_DIR); require(clientHomeDir != null, "Environment variable %s not set", YarnConfigKeys.ENV_CLIENT_HOME_DIR); String shipListString = env.get(YarnConfigKeys.ENV_CLIENT_SHIP_FILES); require(shipListString != null, "Environment variable %s not set", YarnConfigKeys.ENV_CLIENT_SHIP_FILES); String yarnClientUsername = env.get(YarnConfigKeys.ENV_CLIENT_USERNAME); require(yarnClientUsername != null, "Environment variable %s not set", YarnConfigKeys.ENV_CLIENT_USERNAME); // obtain a handle to the file system used by YARN final org.apache.hadoop.fs.FileSystem yarnFileSystem; try { yarnFileSystem = org.apache.hadoop.fs.FileSystem.get(yarnConfig); } catch (IOException e) { throw new Exception("Could not access YARN's default file system", e); } // register Flink Jar with remote HDFS LocalResource flinkJar = Records.newRecord(LocalResource.class); { Path remoteJarPath = new Path(remoteFlinkJarPath); Utils.registerLocalResource(yarnFileSystem, remoteJarPath, flinkJar); } // register conf with local fs LocalResource flinkConf = Records.newRecord(LocalResource.class); { // write the TaskManager configuration to a local file final File taskManagerConfigFile = new File(workingDirectory, UUID.randomUUID() + "-taskmanager-conf.yaml"); LOG.debug("Writing TaskManager configuration to {}", taskManagerConfigFile.getAbsolutePath()); BootstrapTools.writeConfiguration(taskManagerConfig, taskManagerConfigFile); Utils.setupLocalResource(yarnFileSystem, appId, new Path(taskManagerConfigFile.toURI()), flinkConf, new Path(clientHomeDir)); log.info("Prepared local resource for modified yaml: {}", flinkConf); } Map<String, LocalResource> taskManagerLocalResources = new HashMap<>(); taskManagerLocalResources.put("flink.jar", flinkJar); taskManagerLocalResources.put("flink-conf.yaml", flinkConf); // prepare additional files to be shipped for (String pathStr : shipListString.split(",")) { if (!pathStr.isEmpty()) { LocalResource resource = Records.newRecord(LocalResource.class); Path path = new Path(pathStr); Utils.registerLocalResource(yarnFileSystem, path, resource); taskManagerLocalResources.put(path.getName(), resource); } } // now that all resources are prepared, we can create the launch context log.info("Creating container launch context for TaskManagers"); boolean hasLogback = new File(workingDirectory, "logback.xml").exists(); boolean hasLog4j = new File(workingDirectory, "log4j.properties").exists(); String launchCommand = BootstrapTools.getTaskManagerShellCommand(flinkConfig, tmParams, ".", ApplicationConstants.LOG_DIR_EXPANSION_VAR, hasLogback, hasLog4j, taskManagerMainClass); log.info("Starting TaskManagers with command: " + launchCommand); ContainerLaunchContext ctx = Records.newRecord(ContainerLaunchContext.class); ctx.setCommands(Collections.singletonList(launchCommand)); ctx.setLocalResources(taskManagerLocalResources); Map<String, String> containerEnv = new HashMap<>(); containerEnv.putAll(tmParams.taskManagerEnv()); // add YARN classpath, etc to the container environment Utils.setupEnv(yarnConfig, containerEnv); containerEnv.put(YarnConfigKeys.ENV_CLIENT_USERNAME, yarnClientUsername); ctx.setEnvironment(containerEnv); try { UserGroupInformation user = UserGroupInformation.getCurrentUser(); Credentials credentials = user.getCredentials(); DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); ByteBuffer securityTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); ctx.setTokens(securityTokens); } catch (Throwable t) { log.error("Getting current user info failed when trying to launch the container", t); } return ctx; }
From source file:org.apache.giraph.yarn.GiraphApplicationMaster.java
License:Apache License
/** * Populate allTokens with the tokens received * @return//from ww w.jav a 2 s.c o m */ private void getAllTokens() throws IOException { Credentials credentials = UserGroupInformation.getCurrentUser().getCredentials(); DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); // Now remove the AM->RM token so that containers cannot access it. Iterator<Token<?>> iter = credentials.getAllTokens().iterator(); while (iter.hasNext()) { Token<?> token = iter.next(); if (LOG.isDebugEnabled()) { LOG.debug("Token type :" + token.getKind()); } if (token.getKind().equals(AMRMTokenIdentifier.KIND_NAME)) { iter.remove(); } } allTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); }
From source file:org.apache.giraph.yarn.GiraphYarnClient.java
License:Apache License
/** * Set delegation tokens for AM container * @param amContainer AM container//from w ww . jav a 2 s. c om * @return */ private void setToken(ContainerLaunchContext amContainer) throws IOException { // Setup security tokens if (UserGroupInformation.isSecurityEnabled()) { Credentials credentials = new Credentials(); String tokenRenewer = giraphConf.get(YarnConfiguration.RM_PRINCIPAL); if (tokenRenewer == null || tokenRenewer.length() == 0) { throw new IOException("Can't get Master Kerberos principal for the RM to use as renewer"); } FileSystem fs = FileSystem.get(giraphConf); // For now, only getting tokens for the default file-system. final Token<?>[] tokens = fs.addDelegationTokens(tokenRenewer, credentials); if (tokens != null) { for (Token<?> token : tokens) { LOG.info("Got dt for " + fs.getUri() + "; " + token); } } DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); ByteBuffer fsTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); amContainer.setTokens(fsTokens); } }