Examples with ResourceBlacklistRequest org.apache.hadoop.yarn.api.records.ResourceBlacklistRequest used on opensource projects

Example 6 with ResourceBlacklistRequest

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

the class TestSchedulerUtils method testValidateResourceBlacklistRequest.

@Test
public void testValidateResourceBlacklistRequest() throws Exception {
    MyContainerManager containerManager = new MyContainerManager();
    final MockRMWithAMS rm = new MockRMWithAMS(new YarnConfiguration(), containerManager);
    rm.start();
    MockNM nm1 = rm.registerNode("localhost:1234", 5120);
    Map<ApplicationAccessType, String> acls = new HashMap<ApplicationAccessType, String>(2);
    acls.put(ApplicationAccessType.VIEW_APP, "*");
    RMApp app = rm.submitApp(1024, "appname", "appuser", acls);
    nm1.nodeHeartbeat(true);
    RMAppAttempt attempt = app.getCurrentAppAttempt();
    ApplicationAttemptId applicationAttemptId = attempt.getAppAttemptId();
    waitForLaunchedState(attempt);
    // Create a client to the RM.
    final Configuration conf = rm.getConfig();
    final YarnRPC rpc = YarnRPC.create(conf);
    UserGroupInformation currentUser = UserGroupInformation.createRemoteUser(applicationAttemptId.toString());
    Credentials credentials = containerManager.getContainerCredentials();
    final InetSocketAddress rmBindAddress = rm.getApplicationMasterService().getBindAddress();
    Token<? extends TokenIdentifier> amRMToken = MockRMWithAMS.setupAndReturnAMRMToken(rmBindAddress, credentials.getAllTokens());
    currentUser.addToken(amRMToken);
    ApplicationMasterProtocol client = currentUser.doAs(new PrivilegedAction<ApplicationMasterProtocol>() {

        @Override
        public ApplicationMasterProtocol run() {
            return (ApplicationMasterProtocol) rpc.getProxy(ApplicationMasterProtocol.class, rmBindAddress, conf);
        }
    });
    RegisterApplicationMasterRequest request = Records.newRecord(RegisterApplicationMasterRequest.class);
    client.registerApplicationMaster(request);
    ResourceBlacklistRequest blacklistRequest = ResourceBlacklistRequest.newInstance(Collections.singletonList(ResourceRequest.ANY), null);
    AllocateRequest allocateRequest = AllocateRequest.newInstance(0, 0.0f, null, null, blacklistRequest);
    boolean error = false;
    try {
        client.allocate(allocateRequest);
    } catch (InvalidResourceBlacklistRequestException e) {
        error = true;
    }
    rm.stop();
    Assert.assertTrue("Didn't not catch InvalidResourceBlacklistRequestException", error);
}

Example 7 with ResourceBlacklistRequest

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

the class TestBlacklistManager method testSimpleBlacklistAboveFailureThreshold.

@Test
public void testSimpleBlacklistAboveFailureThreshold() {
    // Create a threshold of 0.5 * 3 i.e at 1.5 node failures.
    BlacklistManager manager = new SimpleBlacklistManager(3, 0.5);
    String anyNode = "foo";
    String anyNode2 = "bar";
    manager.addNode(anyNode);
    ResourceBlacklistRequest blacklist = manager.getBlacklistUpdates();
    List<String> blacklistAdditions = blacklist.getBlacklistAdditions();
    Collections.sort(blacklistAdditions);
    List<String> blacklistRemovals = blacklist.getBlacklistRemovals();
    String[] expectedBlacklistAdditions = new String[] { anyNode };
    Assert.assertArrayEquals("Blacklist additions was not as expected", expectedBlacklistAdditions, blacklistAdditions.toArray());
    Assert.assertTrue("Blacklist removals should be empty but was " + blacklistRemovals, blacklistRemovals.isEmpty());
    manager.addNode(anyNode2);
    blacklist = manager.getBlacklistUpdates();
    blacklistAdditions = blacklist.getBlacklistAdditions();
    Collections.sort(blacklistAdditions);
    blacklistRemovals = blacklist.getBlacklistRemovals();
    Collections.sort(blacklistRemovals);
    String[] expectedBlacklistRemovals = new String[] { anyNode2, anyNode };
    Assert.assertTrue("Blacklist additions should be empty but was " + blacklistAdditions, blacklistAdditions.isEmpty());
    Assert.assertArrayEquals("Blacklist removals was not as expected", expectedBlacklistRemovals, blacklistRemovals.toArray());
}

Example 8 with ResourceBlacklistRequest

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

the class TestBlacklistManager method testDisabledBlacklist.

@Test
public void testDisabledBlacklist() {
    BlacklistManager disabled = new DisabledBlacklistManager();
    String anyNode = "foo";
    disabled.addNode(anyNode);
    ResourceBlacklistRequest blacklist = disabled.getBlacklistUpdates();
    List<String> blacklistAdditions = blacklist.getBlacklistAdditions();
    List<String> blacklistRemovals = blacklist.getBlacklistRemovals();
    Assert.assertTrue("Blacklist additions should be empty but was " + blacklistAdditions, blacklistAdditions.isEmpty());
    Assert.assertTrue("Blacklist removals should be empty but was " + blacklistRemovals, blacklistRemovals.isEmpty());
}

Example 9 with ResourceBlacklistRequest

use of org.apache.hadoop.yarn.api.records.ResourceBlacklistRequest 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());
}

Example 10 with ResourceBlacklistRequest

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

the class SimpleBlacklistManager method getBlacklistUpdates.

@Override
public ResourceBlacklistRequest getBlacklistUpdates() {
    ResourceBlacklistRequest ret;
    List<String> blacklist = new ArrayList<>(blacklistNodes);
    final int currentBlacklistSize = blacklist.size();
    final double failureThreshold = this.blacklistDisableFailureThreshold * numberOfNodeManagerHosts;
    if (currentBlacklistSize < failureThreshold) {
        if (LOG.isDebugEnabled()) {
            LOG.debug("blacklist size " + currentBlacklistSize + " is less than " + "failure threshold ratio " + blacklistDisableFailureThreshold + " out of total usable nodes " + numberOfNodeManagerHosts);
        }
        ret = ResourceBlacklistRequest.newInstance(blacklist, EMPTY_LIST);
    } else {
        LOG.warn("Ignoring Blacklists, blacklist size " + currentBlacklistSize + " is more than failure threshold ratio " + blacklistDisableFailureThreshold + " out of total usable nodes " + numberOfNodeManagerHosts);
        // TODO: After the threshold hits, we will keep sending a long list
        // every time a new AM is to be scheduled.
        ret = ResourceBlacklistRequest.newInstance(EMPTY_LIST, blacklist);
    }
    return ret;
}