List of usage examples for org.apache.hadoop.yarn.api.protocolrecords RegisterApplicationMasterResponse getContainersFromPreviousAttempts
@Public @Unstable public abstract List<Container> getContainersFromPreviousAttempts();
Get the list of running containers as viewed by ResourceManager from previous application attempts.
From source file:org.hdl.tensorflow.yarn.appmaster.ApplicationMaster.java
License:Apache License
private void setupPreviousRunningContainers(RegisterApplicationMasterResponse response) { String containerIdStr = System.getenv(Environment.CONTAINER_ID.name()); ContainerId containerId = ContainerId.fromString(containerIdStr); appAttemptId = containerId.getApplicationAttemptId(); List<Container> previousAMRunningContainers = response.getContainersFromPreviousAttempts(); LOG.info(appAttemptId + " received " + previousAMRunningContainers.size() + " previous attempts' running containers on AM registration."); for (Container container : previousAMRunningContainers) { launchedContainers.add(container.getId()); }//from w w w . ja v a 2 s. co m allocatedContainerNum.addAndGet(previousAMRunningContainers.size()); }
From source file:org.starschema.hadoop.yarn.applications.distributedshell.ApplicationMaster.java
License:Apache License
/** * Main run function for the application master * * @throws YarnException//from ww w .j av a 2 s. co m * @throws IOException */ @SuppressWarnings({ "unchecked" }) public void run() throws YarnException, IOException, InterruptedException { LOG.info("Starting ApplicationMaster"); // 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(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(); startTimelineClient(conf); if (timelineClient != null) { publishApplicationAttemptEvent(timelineClient, appAttemptID.toString(), DSEvent.DS_APP_ATTEMPT_START, domainId, appSubmitterUgi); } // 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(appAttemptID + " received " + previousAMRunningContainers.size() + " previous attempts' running containers on AM registration."); for (Container container : previousAMRunningContainers) { launchedContainers.add(container.getId()); } 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(numTotalContainers); }
From source file:origin.hadoop.yarn.distributedshell.ApplicationMaster.java
License:Apache License
/** * Main run function for the application master * * @throws YarnException//from w ww . j av a 2 s.com * @throws IOException */ @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); } // 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); 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(appAttemptID + " received " + previousAMRunningContainers.size() + " previous attempts' 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(numTotalContainers); 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:x10.x10rt.yarn.ApplicationMaster.java
License:Open Source License
private void setup() throws IOException, YarnException { LOG.info("Starting ApplicationMaster"); // Remove the AM->RM token so that containers cannot access it. Credentials credentials = UserGroupInformation.getCurrentUser().getCredentials(); DataOutputBuffer dob = new DataOutputBuffer(); credentials.writeTokenStorageToStream(dob); Iterator<Token<?>> iter = credentials.getAllTokens().iterator(); LOG.info("Executing with tokens:"); while (iter.hasNext()) { Token<?> token = iter.next(); LOG.info(token);/*from ww w . ja v a 2s . c om*/ 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()); UserGroupInformation appSubmitterUgi = UserGroupInformation.createRemoteUser(appSubmitterUserName); appSubmitterUgi.addCredentials(credentials); resourceManager = AMRMClientAsync.createAMRMClientAsync(1000, new RMCallbackHandler()); resourceManager.init(conf); resourceManager.start(); nodeManager = new NMClientAsyncImpl(new NMCallbackHandler(this)); nodeManager.init(conf); nodeManager.start(); // Register self with ResourceManager // This will start heartbeating to the RM appMasterHostname = NetUtils.getHostname(); RegisterApplicationMasterResponse response = resourceManager.registerApplicationMaster(appMasterHostname, appMasterRpcPort, appMasterTrackingUrl); { int slash = appMasterHostname.indexOf('/'); if (slash != -1) appMasterHostname = appMasterHostname.substring(0, slash); } // 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. // TODO: should we reject instead of modifying to fit? if (memoryPerPlaceInMb > maxMem) { LOG.info("Container memory specified above max threshold of cluster." + " Using max value." + ", specified=" + memoryPerPlaceInMb + ", max=" + maxMem); memoryPerPlaceInMb = maxMem; } if (coresPerPlace > maxVCores) { LOG.info("Container virtual cores specified above max threshold of cluster." + " Using max value." + ", specified=" + coresPerPlace + ", max=" + maxVCores); coresPerPlace = maxVCores; } else if (coresPerPlace == 0) { LOG.info("Container virtual cores specified as auto (X10_NTHREADS=0)." + " Using max value." + ", specified=" + coresPerPlace + ", max=" + maxVCores); coresPerPlace = maxVCores; } List<Container> previousAMRunningContainers = response.getContainersFromPreviousAttempts(); LOG.info(appAttemptID + " received " + previousAMRunningContainers.size() + " previous attempts' running containers on AM registration."); numAllocatedContainers.addAndGet(previousAMRunningContainers.size()); int numTotalContainersToRequest = initialNumPlaces - previousAMRunningContainers.size(); // open a local port for X10rt management, and register it with the selector launcherChannel = ServerSocketChannel.open(); //launcherChannel.bind(new InetSocketAddress(appMasterHostname, 0)); // bind to the visible network hostname and random port launcherChannel.bind(null); launcherChannel.configureBlocking(false); appMasterPort = launcherChannel.socket().getLocalPort(); launcherChannel.register(selector, SelectionKey.OP_ACCEPT); numRequestedContainers.set(initialNumPlaces); // Send request for containers to RM for (int i = 0; i < numTotalContainersToRequest; ++i) { Resource capability = Resource.newInstance(memoryPerPlaceInMb, coresPerPlace); ContainerRequest request = new ContainerRequest(capability, null, null, Priority.newInstance(0)); LOG.info("Requested container ask: " + request.toString()); resourceManager.addContainerRequest(request); pendingRequests.add(request); } }