Search in sources :

Example 31 with NodeReport

use of org.apache.hadoop.yarn.api.records.NodeReport in project hadoop by apache.

the class TestClientRMService method testGetClusterNodes.

@Test
public void testGetClusterNodes() throws Exception {
    MockRM rm = new MockRM() {

        protected ClientRMService createClientRMService() {
            return new ClientRMService(this.rmContext, scheduler, this.rmAppManager, this.applicationACLsManager, this.queueACLsManager, this.getRMContext().getRMDelegationTokenSecretManager());
        }

        ;
    };
    rm.start();
    RMNodeLabelsManager labelsMgr = rm.getRMContext().getNodeLabelManager();
    labelsMgr.addToCluserNodeLabelsWithDefaultExclusivity(ImmutableSet.of("x", "y"));
    // Add a healthy node with label = x
    MockNM node = rm.registerNode("host1:1234", 1024);
    Map<NodeId, Set<String>> map = new HashMap<NodeId, Set<String>>();
    map.put(node.getNodeId(), ImmutableSet.of("x"));
    labelsMgr.replaceLabelsOnNode(map);
    rm.sendNodeStarted(node);
    node.nodeHeartbeat(true);
    // Add and lose a node with label = y
    MockNM lostNode = rm.registerNode("host2:1235", 1024);
    rm.sendNodeStarted(lostNode);
    lostNode.nodeHeartbeat(true);
    rm.waitForState(lostNode.getNodeId(), NodeState.RUNNING);
    rm.sendNodeLost(lostNode);
    // Create a client.
    Configuration conf = new Configuration();
    YarnRPC rpc = YarnRPC.create(conf);
    InetSocketAddress rmAddress = rm.getClientRMService().getBindAddress();
    LOG.info("Connecting to ResourceManager at " + rmAddress);
    ApplicationClientProtocol client = (ApplicationClientProtocol) rpc.getProxy(ApplicationClientProtocol.class, rmAddress, conf);
    // Make call
    GetClusterNodesRequest request = GetClusterNodesRequest.newInstance(EnumSet.of(NodeState.RUNNING));
    List<NodeReport> nodeReports = client.getClusterNodes(request).getNodeReports();
    Assert.assertEquals(1, nodeReports.size());
    Assert.assertNotSame("Node is expected to be healthy!", NodeState.UNHEALTHY, nodeReports.get(0).getNodeState());
    // Check node's label = x
    Assert.assertTrue(nodeReports.get(0).getNodeLabels().contains("x"));
    // Now make the node unhealthy.
    node.nodeHeartbeat(false);
    rm.waitForState(node.getNodeId(), NodeState.UNHEALTHY);
    // Call again
    nodeReports = client.getClusterNodes(request).getNodeReports();
    Assert.assertEquals("Unhealthy nodes should not show up by default", 0, nodeReports.size());
    // Change label of host1 to y
    map = new HashMap<NodeId, Set<String>>();
    map.put(node.getNodeId(), ImmutableSet.of("y"));
    labelsMgr.replaceLabelsOnNode(map);
    // Now query for UNHEALTHY nodes
    request = GetClusterNodesRequest.newInstance(EnumSet.of(NodeState.UNHEALTHY));
    nodeReports = client.getClusterNodes(request).getNodeReports();
    Assert.assertEquals(1, nodeReports.size());
    Assert.assertEquals("Node is expected to be unhealthy!", NodeState.UNHEALTHY, nodeReports.get(0).getNodeState());
    Assert.assertTrue(nodeReports.get(0).getNodeLabels().contains("y"));
    // Remove labels of host1
    map = new HashMap<NodeId, Set<String>>();
    map.put(node.getNodeId(), ImmutableSet.of("y"));
    labelsMgr.removeLabelsFromNode(map);
    // Query all states should return all nodes
    rm.registerNode("host3:1236", 1024);
    request = GetClusterNodesRequest.newInstance(EnumSet.allOf(NodeState.class));
    nodeReports = client.getClusterNodes(request).getNodeReports();
    Assert.assertEquals(3, nodeReports.size());
    // All host1-3's label should be empty (instead of null)
    for (NodeReport report : nodeReports) {
        Assert.assertTrue(report.getNodeLabels() != null && report.getNodeLabels().isEmpty());
    }
    rpc.stopProxy(client, conf);
    rm.close();
}
Also used : EnumSet(java.util.EnumSet) Set(java.util.Set) ImmutableSet(com.google.common.collect.ImmutableSet) HashSet(java.util.HashSet) CapacitySchedulerConfiguration(org.apache.hadoop.yarn.server.resourcemanager.scheduler.capacity.CapacitySchedulerConfiguration) Configuration(org.apache.hadoop.conf.Configuration) YarnConfiguration(org.apache.hadoop.yarn.conf.YarnConfiguration) ConcurrentHashMap(java.util.concurrent.ConcurrentHashMap) HashMap(java.util.HashMap) InetSocketAddress(java.net.InetSocketAddress) Matchers.anyString(org.mockito.Matchers.anyString) YarnRPC(org.apache.hadoop.yarn.ipc.YarnRPC) ApplicationClientProtocol(org.apache.hadoop.yarn.api.ApplicationClientProtocol) NodeId(org.apache.hadoop.yarn.api.records.NodeId) GetClusterNodesRequest(org.apache.hadoop.yarn.api.protocolrecords.GetClusterNodesRequest) NodeReport(org.apache.hadoop.yarn.api.records.NodeReport) RMNodeLabelsManager(org.apache.hadoop.yarn.server.resourcemanager.nodelabels.RMNodeLabelsManager) Test(org.junit.Test)

Example 32 with NodeReport

use of org.apache.hadoop.yarn.api.records.NodeReport in project jstorm by alibaba.

the class JstormOnYarn method run.

/**
     * Main run function for the client
     *
     * @return true if application completed successfully
     * @throws IOException
     * @throws YarnException
     */
public boolean run() throws IOException, YarnException {
    LOG.info("Running Client");
    jstormClientContext.yarnClient.start();
    YarnClusterMetrics clusterMetrics = jstormClientContext.yarnClient.getYarnClusterMetrics();
    LOG.info("Got Cluster metric info from ASM" + ", numNodeManagers=" + clusterMetrics.getNumNodeManagers());
    List<NodeReport> clusterNodeReports = jstormClientContext.yarnClient.getNodeReports(NodeState.RUNNING);
    LOG.info("Got Cluster node info from ASM, count is " + String.valueOf(clusterNodeReports.size()));
    QueueInfo queueInfo = jstormClientContext.yarnClient.getQueueInfo(this.jstormClientContext.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 = jstormClientContext.yarnClient.getQueueAclsInfo();
    for (QueueUserACLInfo aclInfo : listAclInfo) {
        for (QueueACL userAcl : aclInfo.getUserAcls()) {
            LOG.info("User ACL Info for Queue" + ", queueName=" + aclInfo.getQueueName() + ", userAcl=" + userAcl.name());
        }
    }
    if (jstormClientContext.domainId != null && jstormClientContext.domainId.length() > 0 && jstormClientContext.toCreateDomain) {
        prepareTimelineDomain();
    }
    // Get a new application id
    YarnClientApplication app = jstormClientContext.yarnClient.createApplication();
    GetNewApplicationResponse appResponse = app.getNewApplicationResponse();
    int maxMem = appResponse.getMaximumResourceCapability().getMemory();
    LOG.info("Max mem capabililty of resources in this cluster " + maxMem);
    // A resource ask cannot exceed the max.
    if (jstormClientContext.amMemory > maxMem) {
        LOG.info("AM memory specified above max threshold of cluster. Using max value." + ", specified=" + jstormClientContext.amMemory + ", max=" + maxMem);
        jstormClientContext.amMemory = maxMem;
    }
    int maxVCores = appResponse.getMaximumResourceCapability().getVirtualCores();
    LOG.info("Max virtual cores capabililty of resources in this cluster " + maxVCores);
    if (jstormClientContext.amVCores > maxVCores) {
        LOG.info("AM virtual cores specified above max threshold of cluster. " + "Using max value." + ", specified=" + jstormClientContext.amVCores + ", max=" + maxVCores);
        jstormClientContext.amVCores = maxVCores;
    }
    // set the application name
    ApplicationSubmissionContext appContext = app.getApplicationSubmissionContext();
    ApplicationId appId = appContext.getApplicationId();
    appContext.setKeepContainersAcrossApplicationAttempts(jstormClientContext.keepContainers);
    if (!StringUtils.isBlank(jstormClientContext.instanceName) && JOYConstants.CLIIENT_CLASS.equals(jstormClientContext.appName)) {
        appContext.setApplicationName(jstormClientContext.appName + JOYConstants.DOT + jstormClientContext.instanceName);
    } else {
        appContext.setApplicationName(jstormClientContext.appName);
    }
    if (jstormClientContext.attemptFailuresValidityInterval >= 0) {
        appContext.setAttemptFailuresValidityInterval(jstormClientContext.attemptFailuresValidityInterval);
    }
    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
    FileSystem fs = FileSystem.get(jstormClientContext.conf);
    addToLocalResources(fs, jstormClientContext.appMasterJar, JOYConstants.appMasterJarPath, appId.toString(), localResources, null);
    //add jstormOnYarn's configuration to resources which path is jstorm-yarn.xml
    if (jstormClientContext.confFile == null || jstormClientContext.confFile.isEmpty()) {
        addToLocalResources(fs, JOYConstants.CONF_NAME, JOYConstants.CONF_NAME, appId.toString(), localResources, null);
    } else {
        addToLocalResources(fs, jstormClientContext.confFile, JOYConstants.CONF_NAME, appId.toString(), localResources, null);
    }
    if (jstormClientContext.libJars != null && !jstormClientContext.libJars.isEmpty()) {
        for (String libPath : jstormClientContext.libJars.split(JOYConstants.COMMA)) {
            String[] strArr = libPath.split(JOYConstants.BACKLASH);
            String libName = strArr[strArr.length - 1];
            addToLocalResources(fs, libPath, libName, appId.toString(), localResources, null);
        }
    }
    // Set the log4j properties if needed
    if (!jstormClientContext.log4jPropFile.isEmpty()) {
        addToLocalResources(fs, jstormClientContext.log4jPropFile, JOYConstants.log4jPath, appId.toString(), localResources, null);
    }
    // The shell script has to be made available on the final container(s)
    // where it will be executed.
    String hdfsShellScriptLocation = JOYConstants.EMPTY;
    long hdfsShellScriptLen = 0;
    long hdfsShellScriptTimestamp = 0;
    if (!jstormClientContext.shellScriptPath.isEmpty()) {
        Path shellSrc = new Path(jstormClientContext.shellScriptPath);
        String shellPathSuffix = jstormClientContext.appName + JOYConstants.BACKLASH + appId.toString() + JOYConstants.BACKLASH + JOYConstants.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 (!jstormClientContext.shellCommand.isEmpty()) {
        addToLocalResources(fs, null, JOYConstants.shellCommandPath, appId.toString(), localResources, jstormClientContext.shellCommand);
    }
    if (jstormClientContext.shellArgs.length > 0) {
        addToLocalResources(fs, null, JOYConstants.shellArgsPath, appId.toString(), localResources, StringUtils.join(jstormClientContext.shellArgs, JOYConstants.BLANK));
    }
    // 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>();
    Path appMaterJar = new Path(fs.getHomeDirectory(), jstormClientContext.appName + JOYConstants.BACKLASH + appId.toString() + JOYConstants.BACKLASH + JOYConstants.appMasterJarPath);
    FileStatus jarFIleStatus = fs.getFileStatus(appMaterJar);
    // 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(JOYConstants.DISTRIBUTEDSHELLSCRIPTLOCATION, hdfsShellScriptLocation);
    env.put(JOYConstants.DISTRIBUTEDSHELLSCRIPTLOCATION, hdfsShellScriptLocation);
    env.put(JOYConstants.DISTRIBUTEDSHELLSCRIPTTIMESTAMP, Long.toString(hdfsShellScriptTimestamp));
    env.put(JOYConstants.DISTRIBUTEDSHELLSCRIPTLEN, Long.toString(hdfsShellScriptLen));
    if (jstormClientContext.domainId != null && jstormClientContext.domainId.length() > 0) {
        env.put(JOYConstants.DISTRIBUTEDSHELLTIMELINEDOMAIN, jstormClientContext.domainId);
    }
    env.put(JOYConstants.APPMASTERJARSCRIPTLOCATION, appMaterJar.toUri().toString());
    env.put(JOYConstants.APPMASTERLEN, Long.toString(jarFIleStatus.getLen()));
    env.put(JOYConstants.APPMASTERTIMESTAMP, Long.toString(jarFIleStatus.getModificationTime()));
    env.put(JOYConstants.APPMASTERJARSCRIPTLOCATION, appMaterJar.toUri().toString());
    env.put(JOYConstants.BINARYFILEDEPLOYPATH, jstormClientContext.deployPath);
    env.put(JOYConstants.INSTANCENAME, jstormClientContext.instanceName);
    // Add AppMaster.jar location to classpath
    StringBuilder classPathEnv = new StringBuilder(Environment.CLASSPATH.$$()).append(ApplicationConstants.CLASS_PATH_SEPARATOR).append(JOYConstants.DOT + JOYConstants.BACKLASH + JOYConstants.ASTERISK);
    for (String c : jstormClientContext.conf.getStrings(YarnConfiguration.YARN_APPLICATION_CLASSPATH, YarnConfiguration.DEFAULT_YARN_CROSS_PLATFORM_APPLICATION_CLASSPATH)) {
        classPathEnv.append(ApplicationConstants.CLASS_PATH_SEPARATOR);
        classPathEnv.append(c.trim());
    }
    classPathEnv.append(ApplicationConstants.CLASS_PATH_SEPARATOR).append(JOYConstants.DOT + JOYConstants.BACKLASH + JOYConstants.LOG_PROPERTIES);
    // add the runtime classpath needed for tests to work
    if (jstormClientContext.conf.getBoolean(YarnConfiguration.IS_MINI_YARN_CLUSTER, false)) {
        classPathEnv.append(JOYConstants.COLON);
        classPathEnv.append(System.getProperty(JOYConstants.JAVA_CLASS_PATH));
    }
    env.put(JOYConstants.CLASS_PATH, classPathEnv.toString());
    // Set the necessary command to execute the application master
    Vector<CharSequence> vargs = new Vector<CharSequence>(30);
    // Set java executable command
    vargs.add(Environment.JAVA_HOME.$$() + JOYConstants.JAVA);
    // Set Xmx based on am memory size
    vargs.add(JOYConstants.XMX + jstormClientContext.amMemory + JOYConstants.MB);
    // Set class name
    vargs.add(appMasterMainClass);
    // Set params for Application Master
    vargs.add(JOYConstants.CLI_PREFIX + JOYConstants.CONTAINER_MEMORY + JOYConstants.BLANK + String.valueOf(jstormClientContext.containerMemory));
    vargs.add(JOYConstants.CLI_PREFIX + JOYConstants.CONTAINER_VCORES + JOYConstants.BLANK + String.valueOf(jstormClientContext.containerVirtualCores));
    vargs.add(JOYConstants.CLI_PREFIX + JOYConstants.NUM_CONTAINERS + JOYConstants.BLANK + String.valueOf(jstormClientContext.numContainers));
    if (null != jstormClientContext.nodeLabelExpression) {
        appContext.setNodeLabelExpression(jstormClientContext.nodeLabelExpression);
    }
    vargs.add(JOYConstants.CLI_PREFIX + JOYConstants.PRIORITY + JOYConstants.BLANK + String.valueOf(jstormClientContext.shellCmdPriority));
    for (Map.Entry<String, String> entry : jstormClientContext.shellEnv.entrySet()) {
        vargs.add(JOYConstants.CLI_PREFIX + JOYConstants.SHELL_ENV + JOYConstants.BLANK + entry.getKey() + JOYConstants.EQUAL + entry.getValue());
    }
    if (jstormClientContext.debugFlag) {
        vargs.add(JOYConstants.CLI_PREFIX + JOYConstants.DEBUG);
    }
    vargs.add("1>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + JOYConstants.APPMASTER_STDOUT);
    vargs.add("2>" + ApplicationConstants.LOG_DIR_EXPANSION_VAR + JOYConstants.APPMASTER_STDERR);
    // Get final commmand
    StringBuilder command = new StringBuilder();
    for (CharSequence str : vargs) {
        command.append(str).append(JOYConstants.BLANK);
    }
    LOG.info("Completed setting up app master command " + command.toString());
    List<String> commands = new ArrayList<String>();
    commands.add(command.toString());
    // Set up the container launch context for the application master
    ContainerLaunchContext amContainer = ContainerLaunchContext.newInstance(localResources, env, commands, null, null, null);
    // Set up resource type requirements
    Resource capability = Resource.newInstance(jstormClientContext.amMemory, jstormClientContext.amVCores);
    appContext.setResource(capability);
    // Setup security tokens
    if (UserGroupInformation.isSecurityEnabled()) {
        // Note: Credentials class is marked as LimitedPrivate for HDFS and MapReduce
        Credentials credentials = new Credentials();
        String tokenRenewer = jstormClientContext.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 = Priority.newInstance(jstormClientContext.amPriority);
    appContext.setPriority(pri);
    // Set the queue to which this application is to be submitted in the RM
    appContext.setQueue(jstormClientContext.amQueue);
    LOG.info("Submitting application to ASM");
    //check configuration
    if (JstormYarnUtils.isUnset(jstormClientContext.conf.get(JOYConstants.INSTANCE_NAME_KEY))) {
        throw new IOException(JOYConstants.INSTANCE_NAME_KEY + " is not set");
    }
    if (JstormYarnUtils.isUnset(jstormClientContext.conf.get(JOYConstants.INSTANCE_DEPLOY_DIR_KEY))) {
        throw new IOException(JOYConstants.INSTANCE_DEPLOY_DIR_KEY + " is not set");
    }
    jstormClientContext.yarnClient.submitApplication(appContext);
    // Monitor the application
    return monitorApplication(appId);
}
Also used : QueueInfo(org.apache.hadoop.yarn.api.records.QueueInfo) GetNewApplicationResponse(org.apache.hadoop.yarn.api.protocolrecords.GetNewApplicationResponse) FileStatus(org.apache.hadoop.fs.FileStatus) HashMap(java.util.HashMap) ArrayList(java.util.ArrayList) Token(org.apache.hadoop.security.token.Token) FileSystem(org.apache.hadoop.fs.FileSystem) ApplicationSubmissionContext(org.apache.hadoop.yarn.api.records.ApplicationSubmissionContext) DataOutputBuffer(org.apache.hadoop.io.DataOutputBuffer) Vector(java.util.Vector) Path(org.apache.hadoop.fs.Path) YarnClusterMetrics(org.apache.hadoop.yarn.api.records.YarnClusterMetrics) YarnClientApplication(org.apache.hadoop.yarn.client.api.YarnClientApplication) Priority(org.apache.hadoop.yarn.api.records.Priority) QueueACL(org.apache.hadoop.yarn.api.records.QueueACL) Resource(org.apache.hadoop.yarn.api.records.Resource) LocalResource(org.apache.hadoop.yarn.api.records.LocalResource) QueueUserACLInfo(org.apache.hadoop.yarn.api.records.QueueUserACLInfo) ContainerLaunchContext(org.apache.hadoop.yarn.api.records.ContainerLaunchContext) ByteBuffer(java.nio.ByteBuffer) LocalResource(org.apache.hadoop.yarn.api.records.LocalResource) ApplicationId(org.apache.hadoop.yarn.api.records.ApplicationId) Map(java.util.Map) HashMap(java.util.HashMap) NodeReport(org.apache.hadoop.yarn.api.records.NodeReport) Credentials(org.apache.hadoop.security.Credentials)

Example 33 with NodeReport

use of org.apache.hadoop.yarn.api.records.NodeReport in project hadoop by apache.

the class BuilderUtils method newNodeReport.

public static NodeReport newNodeReport(NodeId nodeId, NodeState nodeState, String httpAddress, String rackName, Resource used, Resource capability, int numContainers, String healthReport, long lastHealthReportTime, Set<String> nodeLabels, ResourceUtilization containersUtilization, ResourceUtilization nodeUtilization) {
    NodeReport nodeReport = recordFactory.newRecordInstance(NodeReport.class);
    nodeReport.setNodeId(nodeId);
    nodeReport.setNodeState(nodeState);
    nodeReport.setHttpAddress(httpAddress);
    nodeReport.setRackName(rackName);
    nodeReport.setUsed(used);
    nodeReport.setCapability(capability);
    nodeReport.setNumContainers(numContainers);
    nodeReport.setHealthReport(healthReport);
    nodeReport.setLastHealthReportTime(lastHealthReportTime);
    nodeReport.setNodeLabels(nodeLabels);
    nodeReport.setAggregatedContainersUtilization(containersUtilization);
    nodeReport.setNodeUtilization(nodeUtilization);
    return nodeReport;
}
Also used : NodeReport(org.apache.hadoop.yarn.api.records.NodeReport)

Example 34 with NodeReport

use of org.apache.hadoop.yarn.api.records.NodeReport in project hadoop by apache.

the class ApplicationMasterService method allocateInternal.

protected void allocateInternal(ApplicationAttemptId appAttemptId, AllocateRequest request, AllocateResponse allocateResponse) throws YarnException {
    //filter illegal progress values
    float filteredProgress = request.getProgress();
    if (Float.isNaN(filteredProgress) || filteredProgress == Float.NEGATIVE_INFINITY || filteredProgress < 0) {
        request.setProgress(0);
    } else if (filteredProgress > 1 || filteredProgress == Float.POSITIVE_INFINITY) {
        request.setProgress(1);
    }
    // Send the status update to the appAttempt.
    this.rmContext.getDispatcher().getEventHandler().handle(new RMAppAttemptStatusupdateEvent(appAttemptId, request.getProgress()));
    List<ResourceRequest> ask = request.getAskList();
    List<ContainerId> release = request.getReleaseList();
    ResourceBlacklistRequest blacklistRequest = request.getResourceBlacklistRequest();
    List<String> blacklistAdditions = (blacklistRequest != null) ? blacklistRequest.getBlacklistAdditions() : Collections.EMPTY_LIST;
    List<String> blacklistRemovals = (blacklistRequest != null) ? blacklistRequest.getBlacklistRemovals() : Collections.EMPTY_LIST;
    RMApp app = this.rmContext.getRMApps().get(appAttemptId.getApplicationId());
    // set label expression for Resource Requests if resourceName=ANY
    ApplicationSubmissionContext asc = app.getApplicationSubmissionContext();
    for (ResourceRequest req : ask) {
        if (null == req.getNodeLabelExpression() && ResourceRequest.ANY.equals(req.getResourceName())) {
            req.setNodeLabelExpression(asc.getNodeLabelExpression());
        }
    }
    Resource maximumCapacity = rScheduler.getMaximumResourceCapability();
    // sanity check
    try {
        RMServerUtils.normalizeAndValidateRequests(ask, maximumCapacity, app.getQueue(), rScheduler, rmContext);
    } catch (InvalidResourceRequestException e) {
        LOG.warn("Invalid resource ask by application " + appAttemptId, e);
        throw e;
    }
    try {
        RMServerUtils.validateBlacklistRequest(blacklistRequest);
    } catch (InvalidResourceBlacklistRequestException e) {
        LOG.warn("Invalid blacklist request by application " + appAttemptId, e);
        throw e;
    }
    // AM to release containers from the earlier attempt.
    if (!app.getApplicationSubmissionContext().getKeepContainersAcrossApplicationAttempts()) {
        try {
            RMServerUtils.validateContainerReleaseRequest(release, appAttemptId);
        } catch (InvalidContainerReleaseException e) {
            LOG.warn("Invalid container release by application " + appAttemptId, e);
            throw e;
        }
    }
    // Split Update Resource Requests into increase and decrease.
    // No Exceptions are thrown here. All update errors are aggregated
    // and returned to the AM.
    List<UpdateContainerError> updateErrors = new ArrayList<>();
    ContainerUpdates containerUpdateRequests = RMServerUtils.validateAndSplitUpdateResourceRequests(rmContext, request, maximumCapacity, updateErrors);
    // Send new requests to appAttempt.
    Allocation allocation;
    RMAppAttemptState state = app.getRMAppAttempt(appAttemptId).getAppAttemptState();
    if (state.equals(RMAppAttemptState.FINAL_SAVING) || state.equals(RMAppAttemptState.FINISHING) || app.isAppFinalStateStored()) {
        LOG.warn(appAttemptId + " is in " + state + " state, ignore container allocate request.");
        allocation = EMPTY_ALLOCATION;
    } else {
        allocation = this.rScheduler.allocate(appAttemptId, ask, release, blacklistAdditions, blacklistRemovals, containerUpdateRequests);
    }
    if (!blacklistAdditions.isEmpty() || !blacklistRemovals.isEmpty()) {
        LOG.info("blacklist are updated in Scheduler." + "blacklistAdditions: " + blacklistAdditions + ", " + "blacklistRemovals: " + blacklistRemovals);
    }
    RMAppAttempt appAttempt = app.getRMAppAttempt(appAttemptId);
    if (allocation.getNMTokens() != null && !allocation.getNMTokens().isEmpty()) {
        allocateResponse.setNMTokens(allocation.getNMTokens());
    }
    // Notify the AM of container update errors
    addToUpdateContainerErrors(allocateResponse, updateErrors);
    // update the response with the deltas of node status changes
    List<RMNode> updatedNodes = new ArrayList<RMNode>();
    if (app.pullRMNodeUpdates(updatedNodes) > 0) {
        List<NodeReport> updatedNodeReports = new ArrayList<NodeReport>();
        for (RMNode rmNode : updatedNodes) {
            SchedulerNodeReport schedulerNodeReport = rScheduler.getNodeReport(rmNode.getNodeID());
            Resource used = BuilderUtils.newResource(0, 0);
            int numContainers = 0;
            if (schedulerNodeReport != null) {
                used = schedulerNodeReport.getUsedResource();
                numContainers = schedulerNodeReport.getNumContainers();
            }
            NodeId nodeId = rmNode.getNodeID();
            NodeReport report = BuilderUtils.newNodeReport(nodeId, rmNode.getState(), rmNode.getHttpAddress(), rmNode.getRackName(), used, rmNode.getTotalCapability(), numContainers, rmNode.getHealthReport(), rmNode.getLastHealthReportTime(), rmNode.getNodeLabels());
            updatedNodeReports.add(report);
        }
        allocateResponse.setUpdatedNodes(updatedNodeReports);
    }
    addToAllocatedContainers(allocateResponse, allocation.getContainers());
    allocateResponse.setCompletedContainersStatuses(appAttempt.pullJustFinishedContainers());
    allocateResponse.setAvailableResources(allocation.getResourceLimit());
    addToContainerUpdates(appAttemptId, allocateResponse, allocation);
    allocateResponse.setNumClusterNodes(this.rScheduler.getNumClusterNodes());
    // add collector address for this application
    if (YarnConfiguration.timelineServiceV2Enabled(getConfig())) {
        allocateResponse.setCollectorAddr(this.rmContext.getRMApps().get(appAttemptId.getApplicationId()).getCollectorAddr());
    }
    // add preemption to the allocateResponse message (if any)
    allocateResponse.setPreemptionMessage(generatePreemptionMessage(allocation));
    // Set application priority
    allocateResponse.setApplicationPriority(app.getApplicationPriority());
}
Also used : RMApp(org.apache.hadoop.yarn.server.resourcemanager.rmapp.RMApp) RMAppAttempt(org.apache.hadoop.yarn.server.resourcemanager.rmapp.attempt.RMAppAttempt) ResourceBlacklistRequest(org.apache.hadoop.yarn.api.records.ResourceBlacklistRequest) ContainerUpdates(org.apache.hadoop.yarn.server.resourcemanager.scheduler.ContainerUpdates) ArrayList(java.util.ArrayList) InvalidResourceRequestException(org.apache.hadoop.yarn.exceptions.InvalidResourceRequestException) RMNode(org.apache.hadoop.yarn.server.resourcemanager.rmnode.RMNode) Allocation(org.apache.hadoop.yarn.server.resourcemanager.scheduler.Allocation) ContainerId(org.apache.hadoop.yarn.api.records.ContainerId) InvalidContainerReleaseException(org.apache.hadoop.yarn.exceptions.InvalidContainerReleaseException) SchedulerNodeReport(org.apache.hadoop.yarn.server.resourcemanager.scheduler.SchedulerNodeReport) ApplicationSubmissionContext(org.apache.hadoop.yarn.api.records.ApplicationSubmissionContext) RMAppAttemptStatusupdateEvent(org.apache.hadoop.yarn.server.resourcemanager.rmapp.attempt.event.RMAppAttemptStatusupdateEvent) Resource(org.apache.hadoop.yarn.api.records.Resource) UpdateContainerError(org.apache.hadoop.yarn.api.records.UpdateContainerError) NodeId(org.apache.hadoop.yarn.api.records.NodeId) InvalidResourceBlacklistRequestException(org.apache.hadoop.yarn.exceptions.InvalidResourceBlacklistRequestException) RMAppAttemptState(org.apache.hadoop.yarn.server.resourcemanager.rmapp.attempt.RMAppAttemptState) PreemptionResourceRequest(org.apache.hadoop.yarn.api.records.PreemptionResourceRequest) ResourceRequest(org.apache.hadoop.yarn.api.records.ResourceRequest) NodeReport(org.apache.hadoop.yarn.api.records.NodeReport) SchedulerNodeReport(org.apache.hadoop.yarn.server.resourcemanager.scheduler.SchedulerNodeReport)

Example 35 with NodeReport

use of org.apache.hadoop.yarn.api.records.NodeReport in project hadoop by apache.

the class ClientRMService method createNodeReports.

private NodeReport createNodeReports(RMNode rmNode) {
    SchedulerNodeReport schedulerNodeReport = scheduler.getNodeReport(rmNode.getNodeID());
    Resource used = BuilderUtils.newResource(0, 0);
    int numContainers = 0;
    if (schedulerNodeReport != null) {
        used = schedulerNodeReport.getUsedResource();
        numContainers = schedulerNodeReport.getNumContainers();
    }
    NodeReport report = BuilderUtils.newNodeReport(rmNode.getNodeID(), rmNode.getState(), rmNode.getHttpAddress(), rmNode.getRackName(), used, rmNode.getTotalCapability(), numContainers, rmNode.getHealthReport(), rmNode.getLastHealthReportTime(), rmNode.getNodeLabels(), rmNode.getAggregatedContainersUtilization(), rmNode.getNodeUtilization());
    return report;
}
Also used : SchedulerNodeReport(org.apache.hadoop.yarn.server.resourcemanager.scheduler.SchedulerNodeReport) Resource(org.apache.hadoop.yarn.api.records.Resource) SchedulerNodeReport(org.apache.hadoop.yarn.server.resourcemanager.scheduler.SchedulerNodeReport) NodeReport(org.apache.hadoop.yarn.api.records.NodeReport)

Aggregations

NodeReport (org.apache.hadoop.yarn.api.records.NodeReport)49 Test (org.junit.Test)18 ArrayList (java.util.ArrayList)17 Resource (org.apache.hadoop.yarn.api.records.Resource)14 GenericTestOperator (com.datatorrent.stram.engine.GenericTestOperator)10 LogicalPlan (com.datatorrent.stram.plan.logical.LogicalPlan)10 ContainerStartRequest (com.datatorrent.stram.StreamingContainerAgent.ContainerStartRequest)9 MemoryStorageAgent (com.datatorrent.stram.support.StramTestSupport.MemoryStorageAgent)9 NodeId (org.apache.hadoop.yarn.api.records.NodeId)9 File (java.io.File)8 NodeState (org.apache.hadoop.yarn.api.records.NodeState)7 HashMap (java.util.HashMap)6 HashSet (java.util.HashSet)6 Priority (org.apache.hadoop.yarn.api.records.Priority)6 IOException (java.io.IOException)5 Map (java.util.Map)5 Container (org.apache.hadoop.yarn.api.records.Container)5 LocalResource (org.apache.hadoop.yarn.api.records.LocalResource)5 PrintWriter (java.io.PrintWriter)4 Configuration (org.apache.hadoop.conf.Configuration)4