List of usage examples for org.apache.hadoop.security Credentials writeTokenStorageToStream
public void writeTokenStorageToStream(DataOutputStream os) throws IOException
From source file:edu.cmu.graphchi.toolkits.collaborative_filtering.yarn.ApplicationMaster.java
License:Apache License
/** * Main run function for the application master * * @throws YarnException//from w w w.j a v a 2s.com * @throws IOException */ @SuppressWarnings({ "unchecked" }) public boolean run() throws YarnException, IOException { yarnClient.start(); LOG.info("Starting ApplicationMaster"); 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 (token.getKind().equals(AMRMTokenIdentifier.KIND_NAME)) { iter.remove(); } } allTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); amRMClient = AMRMClient.createAMRMClient(); amRMClient.init(conf); amRMClient.start(); containerListener = createNMCallbackHandler(); nmClientAsync = new NMClientAsyncImpl(containerListener); nmClientAsync.init(conf); nmClientAsync.start(); // Register self with ResourceManager // This will start heartbeating to the RM appMasterHostname = NetUtils.getHostname(); RegisterApplicationMasterResponse response = amRMClient.registerApplicationMaster(appMasterHostname, appMasterRpcPort, appMasterTrackingUrl); //TODO: Figure out how to do this. List<NodeReport> reports = this.yarnClient.getNodeReports(); LOG.info("Cluster Status"); List<Resource> availableResources = new ArrayList<Resource>(); for (NodeReport nr : reports) { LOG.info(" NodeId: " + nr.getNodeId() + " Capabilities " + nr.getCapability() + " Used Resources " + nr.getUsed()); int availableMem = nr.getCapability().getMemory() - nr.getUsed().getMemory(); int availableVCores = nr.getCapability().getVirtualCores() - nr.getUsed().getVirtualCores(); Resource resource = Resource.newInstance(availableMem, availableVCores); availableResources.add(resource); } /*Based on available resources scheduler should decide the best allocation of recommenders to resources and return a list of resources that should be requested from the ResourceManager*/ DataSetDescription datasetDesc = new DataSetDescription(this.setup.dataMetadataFile); List<RecommenderPool> recommenderPools = RecommenderScheduler.splitRecommenderPool(availableResources, recommenders, datasetDesc, this.setup.nShards); for (RecommenderPool res : recommenderPools) { ContainerRequest containerAsk = setupContainerAskForRM(res.getTotalMemory(), requestPriority); LOG.info("CONTAINER ASK: " + containerAsk); amRMClient.addContainerRequest(containerAsk); } float progress = 0; List<RecommenderPool> pendingPools = new ArrayList<RecommenderPool>(); for (RecommenderPool p : recommenderPools) pendingPools.add(p); this.numTotalContainers = recommenderPools.size(); this.numCompletedContainers.set(0); this.numAllocatedContainers.set(0); while (numCompletedContainers.get() != numTotalContainers) { try { Thread.sleep(200); AllocateResponse allocResp = amRMClient.allocate(progress); List<Container> newContainers = allocResp.getAllocatedContainers(); List<ContainerStatus> completedContainers = allocResp.getCompletedContainersStatuses(); if (this.numAllocatedContainers.get() >= this.numTotalContainers && pendingPools.size() != 0) { //Ask for new containers for pending pools LOG.warn("The number of allocated containers has exceeded number of total containers, but " + "the pending pool size is still not 0. Asking for new containers from RM"); for (RecommenderPool res : pendingPools) { ContainerRequest containerAsk = setupContainerAskForRM(res.getTotalMemory(), requestPriority); LOG.info("NEW CONTAINER ASK: " + containerAsk); amRMClient.addContainerRequest(containerAsk); } } if (newContainers.size() > 0) { LOG.info("Allocated " + newContainers.size() + " new containers"); numAllocatedContainers.addAndGet(newContainers.size()); for (Container container : newContainers) { //Find matching recommender pool from pendingRecommender pools. RecommenderPool pool = null; for (RecommenderPool p : pendingPools) { if (p.getTotalMemory() == container.getResource().getMemory()) { pool = p; break; } } if (pool == null) { LOG.warn("No Takers for Container " + container + " Releasing container"); amRMClient.releaseAssignedContainer(container.getId()); } else { startContainer(container, pool); //This pool has now got a container. Remove it from pending pools pendingPools.remove(pool); } } } onContainersCompleted(completedContainers); } catch (InterruptedException ex) { } } finish(); return success; }
From source file:eu.stratosphere.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 remoteStratosphereJarPath = envs.get(Client.STRATOSPHERE_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"); }/*from w w w . j ava 2 s .c o m*/ if (ownHostname == null) { throw new RuntimeException("Own hostname (" + Environment.NM_HOST + ") not set."); } LOG.info("Working directory " + currDir); // load Stratosphere configuration. Utils.getStratosphereConfiguration(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 + "/stratosphere-conf.yaml"); BufferedReader br = new BufferedReader(new InputStreamReader(fis)); Writer output = new BufferedWriter(new FileWriter(currDir + "/stratosphere-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 + "/stratosphere-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 + "/stratosphere-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 stratosphereJar = Records.newRecord(LocalResource.class); LocalResource stratosphereConf = Records.newRecord(LocalResource.class); // register Stratosphere Jar with remote HDFS final Path remoteJarPath = new Path(remoteStratosphereJarPath); Utils.registerLocalResource(fs, remoteJarPath, stratosphereJar); // register conf with local fs. Path remoteConfPath = Utils.setupLocalResource(conf, fs, appId, new Path("file://" + currDir + "/stratosphere-conf-modified.yaml"), stratosphereConf, new Path(clientHomeDir)); LOG.info("Prepared localresource for modified yaml: " + stratosphereConf); 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.STRATOSPHERE_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 += " eu.stratosphere.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("stratosphere.jar", stratosphereJar); localResources.put("stratosphere-conf.yaml", stratosphereConf); // 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 stratosphere.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:eu.stratosphere.yarn.Utils.java
License:Apache License
public static void setTokensFor(ContainerLaunchContext amContainer, Path[] paths, Configuration conf) throws IOException { Credentials credentials = new Credentials(); // for HDFS/*from ww w . j a v a2s. co m*/ TokenCache.obtainTokensForNamenodes(credentials, paths, conf); // for user UserGroupInformation currUsr = UserGroupInformation.getCurrentUser(); Collection<Token<? extends TokenIdentifier>> usrTok = currUsr.getTokens(); for (Token<? extends TokenIdentifier> token : usrTok) { final Text id = new Text(token.getIdentifier()); LOG.info("Adding user token " + id + " with " + token); credentials.addToken(id, token); } DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); LOG.debug("Wrote tokens. Credentials buffer length: " + dob.getLength()); ByteBuffer securityTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); amContainer.setTokens(securityTokens); }
From source file:gobblin.hadoop.token.TokenUtils.java
License:Open Source License
private static void persistTokens(Credentials cred, File tokenFile) throws IOException { FileOutputStream fos = null;/*from w ww . j av a 2s . c om*/ DataOutputStream dos = null; try { fos = new FileOutputStream(tokenFile); dos = new DataOutputStream(fos); cred.writeTokenStorageToStream(dos); } finally { if (dos != null) { try { dos.close(); } catch (Throwable t) { LOG.error("encountered exception while closing DataOutputStream of the tokenFile", t); } } if (fos != null) { fos.close(); } } LOG.info("Tokens loaded in " + tokenFile.getAbsolutePath()); }
From source file:gobblin.util.hadoop.TokenUtils.java
License:Apache License
private static void persistTokens(Credentials cred, File tokenFile) throws IOException { try (FileOutputStream fos = new FileOutputStream(tokenFile); DataOutputStream dos = new DataOutputStream(fos)) { cred.writeTokenStorageToStream(dos); }/*from www .j a v a2 s . com*/ LOG.info("Tokens loaded in " + tokenFile.getAbsolutePath()); }
From source file:gobblin.yarn.GobblinYarnAppLauncher.java
License:Apache License
private void setupSecurityTokens(ContainerLaunchContext containerLaunchContext) throws IOException { Credentials credentials = UserGroupInformation.getCurrentUser().getCredentials(); String tokenRenewer = this.yarnConfiguration.get(YarnConfiguration.RM_PRINCIPAL); if (tokenRenewer == null || tokenRenewer.length() == 0) { throw new IOException("Failed to get master Kerberos principal for the RM to use as renewer"); }// w ww . j a v a2 s.c o m // For now, only getting tokens for the default file-system. Token<?> tokens[] = this.fs.addDelegationTokens(tokenRenewer, credentials); if (tokens != null) { for (Token<?> token : tokens) { LOGGER.info("Got delegation token for " + this.fs.getUri() + "; " + token); } } Closer closer = Closer.create(); try { DataOutputBuffer dataOutputBuffer = closer.register(new DataOutputBuffer()); credentials.writeTokenStorageToStream(dataOutputBuffer); ByteBuffer fsTokens = ByteBuffer.wrap(dataOutputBuffer.getData(), 0, dataOutputBuffer.getLength()); containerLaunchContext.setTokens(fsTokens); } catch (Throwable t) { throw closer.rethrow(t); } finally { closer.close(); } }
From source file:gobblin.yarn.YarnService.java
License:Apache License
private ByteBuffer getSecurityTokens() throws IOException { Credentials credentials = UserGroupInformation.getCurrentUser().getCredentials(); Closer closer = Closer.create();/*from www . j a va2 s. c o m*/ try { DataOutputBuffer dataOutputBuffer = closer.register(new DataOutputBuffer()); credentials.writeTokenStorageToStream(dataOutputBuffer); // Remove the AM->RM token so that containers cannot access it Iterator<Token<?>> tokenIterator = credentials.getAllTokens().iterator(); while (tokenIterator.hasNext()) { Token<?> token = tokenIterator.next(); if (token.getKind().equals(AMRMTokenIdentifier.KIND_NAME)) { tokenIterator.remove(); } } return ByteBuffer.wrap(dataOutputBuffer.getData(), 0, dataOutputBuffer.getLength()); } catch (Throwable t) { throw closer.rethrow(t); } finally { closer.close(); } }
From source file:hadoop.yarn.distributedshell.DshellClient.java
License:Apache License
/** * Main run function for the client//from w w w . ja v a 2s . com * * @return true if application completed successfully * @throws IOException * @throws 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(DshellDSConstants.DISTRIBUTEDSHELLSCRIPTLOCATION, hdfsShellScriptLocation); env.put(DshellDSConstants.DISTRIBUTEDSHELLSCRIPTTIMESTAMP, Long.toString(hdfsShellScriptTimestamp)); env.put(DshellDSConstants.DISTRIBUTEDSHELLSCRIPTLEN, Long.toString(hdfsShellScriptLen)); // ========================================jar? if (containerJarPaths.length != 0) { for (int i = 0; i < containerJarPaths.length; i++) { String hdfsJarLocation = ""; String[] jarNameSplit = containerJarPaths[i].split("/"); String jarName = jarNameSplit[jarNameSplit.length - 1]; long hdfsJarLen = 0; long hdfsJarTimestamp = 0; if (!containerJarPaths[i].isEmpty()) { Path jarSrc = new Path(containerJarPaths[i]); String jarPathSuffix = appName + "/" + appId.toString() + "/" + jarName; Path jarDst = new Path(fs.getHomeDirectory(), jarPathSuffix); fs.copyFromLocalFile(false, true, jarSrc, jarDst); hdfsJarLocation = jarDst.toUri().toString(); FileStatus jarFileStatus = fs.getFileStatus(jarDst); hdfsJarLen = jarFileStatus.getLen(); hdfsJarTimestamp = jarFileStatus.getModificationTime(); env.put(DshellDSConstants.DISTRIBUTEDJARLOCATION + i, hdfsJarLocation); env.put(DshellDSConstants.DISTRIBUTEDJARTIMESTAMP + i, Long.toString(hdfsJarTimestamp)); env.put(DshellDSConstants.DISTRIBUTEDJARLEN + i, Long.toString(hdfsJarLen)); } } } // ========================================jar? // ========================================archive? if (containerArchivePaths.length != 0) { for (int i = 0; i < containerArchivePaths.length; i++) { String hdfsArchiveLocation = ""; String[] archiveNameSplit = containerArchivePaths[i].split("/"); String archiveName = archiveNameSplit[archiveNameSplit.length - 1]; long hdfsArchiveLen = 0; long hdfsArchiveTimestamp = 0; if (!containerArchivePaths[i].isEmpty()) { Path archiveSrc = new Path(containerArchivePaths[i]); String archivePathSuffix = appName + "/" + appId.toString() + "/" + archiveName; Path archiveDst = new Path(fs.getHomeDirectory(), archivePathSuffix); fs.copyFromLocalFile(false, true, archiveSrc, archiveDst); hdfsArchiveLocation = archiveDst.toUri().toString(); FileStatus archiveFileStatus = fs.getFileStatus(archiveDst); hdfsArchiveLen = archiveFileStatus.getLen(); hdfsArchiveTimestamp = archiveFileStatus.getModificationTime(); env.put(DshellDSConstants.DISTRIBUTEDARCHIVELOCATION + i, hdfsArchiveLocation); env.put(DshellDSConstants.DISTRIBUTEDARCHIVETIMESTAMP + i, Long.toString(hdfsArchiveTimestamp)); env.put(DshellDSConstants.DISTRIBUTEDARCHIVELEN + i, Long.toString(hdfsArchiveLen)); } } } // ========================================archive? // 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(Environment.CLASSPATH.$$()) .append(ApplicationConstants.CLASS_PATH_SEPARATOR).append("./*"); for (String c : conf.getStrings(YarnConfiguration.YARN_APPLICATION_CLASSPATH, YarnConfiguration.DEFAULT_YARN_CROSS_PLATFORM_APPLICATION_CLASSPATH)) { classPathEnv.append(ApplicationConstants.CLASS_PATH_SEPARATOR); classPathEnv.append(c.trim()); } classPathEnv.append(ApplicationConstants.CLASS_PATH_SEPARATOR).append("./log4j.properties"); // add the runtime classpath needed for tests to work if (conf.getBoolean(YarnConfiguration.IS_MINI_YARN_CLUSTER, false)) { classPathEnv.append(':'); classPathEnv.append(System.getProperty("java.class.path")); } env.put("CLASSPATH", classPathEnv.toString()); amContainer.setEnvironment(env); // Set the necessary command to execute the application master Vector<CharSequence> vargs = new Vector<CharSequence>(30); // Set java executable command LOG.info("Setting up app master command"); vargs.add(Environment.JAVA_HOME.$$() + "/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()); 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:io.hops.ha.common.TransactionStateImpl.java
License:Apache License
private void persistAppAttempt() throws IOException { if (!appAttempts.isEmpty()) { ApplicationAttemptStateDataAccess DA = (ApplicationAttemptStateDataAccess) RMStorageFactory .getDataAccess(ApplicationAttemptStateDataAccess.class); List<ApplicationAttemptState> toAdd = new ArrayList<ApplicationAttemptState>(); for (String appAttemptIdStr : appAttempts.keySet()) { RMAppAttempt appAttempt = appAttempts.get(appAttemptIdStr); String appIdStr = appAttempt.getAppAttemptId().getApplicationId().toString(); Credentials credentials = appAttempt.getCredentials(); ByteBuffer appAttemptTokens = null; if (credentials != null) { DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); appAttemptTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength()); }/*from w ww . j ava 2 s .co m*/ ApplicationAttemptStateDataPBImpl attemptStateData = (ApplicationAttemptStateDataPBImpl) ApplicationAttemptStateDataPBImpl .newApplicationAttemptStateData(appAttempt.getAppAttemptId(), appAttempt.getMasterContainer(), appAttemptTokens, appAttempt.getStartTime(), appAttempt.getState(), appAttempt.getOriginalTrackingUrl(), appAttempt.getDiagnostics(), appAttempt.getFinalApplicationStatus(), appAttempt.getRanNodes(), appAttempt.getJustFinishedContainers(), appAttempt.getProgress(), appAttempt.getHost(), appAttempt.getRpcPort()); byte[] attemptIdByteArray = attemptStateData.getProto().toByteArray(); LOG.debug("adding appAttempt : " + appAttempt.getAppAttemptId() + " with state " + appAttempt.getState()); toAdd.add(new ApplicationAttemptState(appIdStr, appAttemptIdStr, attemptIdByteArray, appAttempt.getHost(), appAttempt.getRpcPort(), appAttemptTokens, appAttempt.getTrackingUrl())); } DA.addAll(toAdd); } }
From source file:io.hops.tensorflow.ApplicationMaster.java
License:Apache License
/** * Main run function for the application master * * @throws YarnException//from w w w. j av a2 s . co m * @throws IOException */ public void run() throws YarnException, IOException, InterruptedException { LOG.info("Starting ApplicationMaster. " + "Workers: " + numWorkers + ", Parameter servers: " + numPses); clusterSpecServer = new ClusterSpecGeneratorServer(appAttemptID.getApplicationId().toString(), numTotalContainers, numWorkers); LOG.info("Starting ClusterSpecGeneratorServer"); int port = 2222; while (true) { try { clusterSpecServer.start(port); break; } catch (IOException e) { port++; } } environment.put("YARNTF_AM_ADDRESS", InetAddress.getLocalHost().getHostName() + ":" + port); environment.put("YARNTF_APPLICATION_ID", appAttemptID.getApplicationId().toString()); // Note: Credentials, Token, UserGroupInformation, DataOutputBuffer class // are marked as LimitedPrivate 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(Environment.USER.name()); appSubmitterUgi = UserGroupInformation.createRemoteUser(appSubmitterUserName); appSubmitterUgi.addCredentials(credentials); rmWrapper = new RMWrapper(this); rmWrapper.getClient().init(conf); rmWrapper.getClient().start(); nmWrapper = new NMWrapper(this); nmWrapper.getClient().init(conf); nmWrapper.getClient().start(); timelineHandler = new TimelineHandler(appAttemptID.toString(), domainId, appSubmitterUgi); timelineHandler.startClient(conf); if (timelineHandler.isClientNotNull()) { timelineHandler.publishApplicationAttemptEvent(YarntfEvent.YARNTF_APP_ATTEMPT_START); } // Register self with ResourceManager // This will start heartbeating to the RM appMasterHostname = NetUtils.getHostname(); appMasterTrackingUrl = InetAddress.getLocalHost().getHostName() + ":" + TensorBoardServer.spawn(this); RegisterApplicationMasterResponse response = rmWrapper.getClient() .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); int maxGPUS = response.getMaximumResourceCapability().getGPUs(); LOG.info("Max gpus capabililty of resources in this cluster " + maxGPUS); // 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; } if (containerGPUs > maxGPUS) { LOG.info("Container gpus specified above max threshold of cluster." + " Using max value." + ", specified=" + containerGPUs + ", max=" + maxGPUS); containerGPUs = maxGPUS; } List<Container> previousAMRunningContainers = response.getContainersFromPreviousAttempts(); LOG.info(appAttemptID + " received " + previousAMRunningContainers.size() + " previous attempts' running containers on AM registration."); numAllocatedContainers.addAndGet(previousAMRunningContainers.size()); // Stop eventual containers from previous attempts for (Container prevContainer : previousAMRunningContainers) { LOG.info("Releasing YARN container " + prevContainer.getId()); rmWrapper.getClient().releaseAssignedContainer(prevContainer.getId()); } // Send request for containers to RM for (int i = 0; i < numWorkers; i++) { ContainerRequest workerRequest = setupContainerAskForRM(true); rmWrapper.getClient().addContainerRequest(workerRequest); } numRequestedContainers.addAndGet(numWorkers); for (int i = 0; i < numPses; i++) { ContainerRequest psRequest = setupContainerAskForRM(false); rmWrapper.getClient().addContainerRequest(psRequest); } numRequestedContainers.addAndGet(numPses); }