Search in sources :

Example 6 with TaskAttemptId

use of org.apache.hyracks.api.dataflow.TaskAttemptId in project asterixdb by apache.

the class JobExecutor method assignTaskLocations.

private void assignTaskLocations(TaskCluster tc, Map<String, List<TaskAttemptDescriptor>> taskAttemptMap) throws HyracksException {
    ActivityClusterGraph acg = jobRun.getActivityClusterGraph();
    Task[] tasks = tc.getTasks();
    List<TaskClusterAttempt> tcAttempts = tc.getAttempts();
    int attempts = tcAttempts.size();
    TaskClusterAttempt tcAttempt = new TaskClusterAttempt(tc, attempts);
    Map<TaskId, TaskAttempt> taskAttempts = new HashMap<>();
    Map<TaskId, LValueConstraintExpression> locationMap = new HashMap<>();
    for (int i = 0; i < tasks.length; ++i) {
        Task ts = tasks[i];
        TaskId tid = ts.getTaskId();
        TaskAttempt taskAttempt = new TaskAttempt(tcAttempt, new TaskAttemptId(new TaskId(tid.getActivityId(), tid.getPartition()), attempts), ts);
        taskAttempt.setStatus(TaskAttempt.TaskStatus.INITIALIZED, null);
        locationMap.put(tid, new PartitionLocationExpression(tid.getActivityId().getOperatorDescriptorId(), tid.getPartition()));
        taskAttempts.put(tid, taskAttempt);
    }
    tcAttempt.setTaskAttempts(taskAttempts);
    solver.solve(locationMap.values());
    for (int i = 0; i < tasks.length; ++i) {
        Task ts = tasks[i];
        TaskId tid = ts.getTaskId();
        TaskAttempt taskAttempt = taskAttempts.get(tid);
        String nodeId = assignLocation(acg, locationMap, tid, taskAttempt);
        taskAttempt.setNodeId(nodeId);
        taskAttempt.setStatus(TaskAttempt.TaskStatus.RUNNING, null);
        taskAttempt.setStartTime(System.currentTimeMillis());
        List<TaskAttemptDescriptor> tads = taskAttemptMap.get(nodeId);
        if (tads == null) {
            tads = new ArrayList<>();
            taskAttemptMap.put(nodeId, tads);
        }
        OperatorDescriptorId opId = tid.getActivityId().getOperatorDescriptorId();
        jobRun.registerOperatorLocation(opId, tid.getPartition(), nodeId);
        ActivityPartitionDetails apd = ts.getActivityPlan().getActivityPartitionDetails();
        TaskAttemptDescriptor tad = new TaskAttemptDescriptor(taskAttempt.getTaskAttemptId(), apd.getPartitionCount(), apd.getInputPartitionCounts(), apd.getOutputPartitionCounts());
        tads.add(tad);
    }
    tcAttempt.initializePendingTaskCounter();
    tcAttempts.add(tcAttempt);
    /**
         * Improvement for reducing master/slave message communications, for each TaskAttemptDescriptor,
         * we set the NetworkAddress[][] partitionLocations, in which each row is for an incoming connector descriptor
         * and each column is for an input channel of the connector.
         */
    INodeManager nodeManager = ccs.getNodeManager();
    for (Map.Entry<String, List<TaskAttemptDescriptor>> e : taskAttemptMap.entrySet()) {
        List<TaskAttemptDescriptor> tads = e.getValue();
        for (TaskAttemptDescriptor tad : tads) {
            TaskAttemptId taid = tad.getTaskAttemptId();
            int attempt = taid.getAttempt();
            TaskId tid = taid.getTaskId();
            ActivityId aid = tid.getActivityId();
            List<IConnectorDescriptor> inConnectors = acg.getActivityInputs(aid);
            int[] inPartitionCounts = tad.getInputPartitionCounts();
            if (inPartitionCounts == null) {
                continue;
            }
            NetworkAddress[][] partitionLocations = new NetworkAddress[inPartitionCounts.length][];
            for (int i = 0; i < inPartitionCounts.length; ++i) {
                ConnectorDescriptorId cdId = inConnectors.get(i).getConnectorId();
                IConnectorPolicy policy = jobRun.getConnectorPolicyMap().get(cdId);
                /**
                     * carry sender location information into a task
                     * when it is not the case that it is an re-attempt and the send-side
                     * is materialized blocking.
                     */
                if (attempt > 0 && policy.materializeOnSendSide() && policy.consumerWaitsForProducerToFinish()) {
                    continue;
                }
                ActivityId producerAid = acg.getProducerActivity(cdId);
                partitionLocations[i] = new NetworkAddress[inPartitionCounts[i]];
                for (int j = 0; j < inPartitionCounts[i]; ++j) {
                    TaskId producerTaskId = new TaskId(producerAid, j);
                    String nodeId = findTaskLocation(producerTaskId);
                    partitionLocations[i][j] = nodeManager.getNodeControllerState(nodeId).getDataPort();
                }
            }
            tad.setInputPartitionLocations(partitionLocations);
        }
    }
    tcAttempt.setStatus(TaskClusterAttempt.TaskClusterStatus.RUNNING);
    tcAttempt.setStartTime(System.currentTimeMillis());
    inProgressTaskClusters.add(tc);
}
Also used : INodeManager(org.apache.hyracks.control.cc.cluster.INodeManager) Task(org.apache.hyracks.control.cc.job.Task) TaskId(org.apache.hyracks.api.dataflow.TaskId) TaskClusterAttempt(org.apache.hyracks.control.cc.job.TaskClusterAttempt) HashMap(java.util.HashMap) ActivityId(org.apache.hyracks.api.dataflow.ActivityId) ConnectorDescriptorId(org.apache.hyracks.api.dataflow.ConnectorDescriptorId) NetworkAddress(org.apache.hyracks.api.comm.NetworkAddress) ArrayList(java.util.ArrayList) List(java.util.List) TaskAttempt(org.apache.hyracks.control.cc.job.TaskAttempt) PartitionLocationExpression(org.apache.hyracks.api.constraints.expressions.PartitionLocationExpression) IConnectorDescriptor(org.apache.hyracks.api.dataflow.IConnectorDescriptor) OperatorDescriptorId(org.apache.hyracks.api.dataflow.OperatorDescriptorId) TaskAttemptId(org.apache.hyracks.api.dataflow.TaskAttemptId) IConnectorPolicy(org.apache.hyracks.api.dataflow.connectors.IConnectorPolicy) Constraint(org.apache.hyracks.api.constraints.Constraint) LValueConstraintExpression(org.apache.hyracks.api.constraints.expressions.LValueConstraintExpression) TaskAttemptDescriptor(org.apache.hyracks.control.common.job.TaskAttemptDescriptor) ActivityClusterGraph(org.apache.hyracks.api.job.ActivityClusterGraph) HashMap(java.util.HashMap) Map(java.util.Map)

Example 7 with TaskAttemptId

use of org.apache.hyracks.api.dataflow.TaskAttemptId in project asterixdb by apache.

the class JobExecutor method notifyTaskComplete.

public void notifyTaskComplete(TaskAttempt ta) throws HyracksException {
    TaskAttemptId taId = ta.getTaskAttemptId();
    TaskCluster tc = ta.getTask().getTaskCluster();
    TaskClusterAttempt lastAttempt = findLastTaskClusterAttempt(tc);
    if (lastAttempt == null || taId.getAttempt() != lastAttempt.getAttempt()) {
        LOGGER.warning("Ignoring task complete notification: " + taId + " -- Current last attempt = " + lastAttempt);
        return;
    }
    TaskAttempt.TaskStatus taStatus = ta.getStatus();
    if (taStatus != TaskAttempt.TaskStatus.RUNNING) {
        LOGGER.warning("Spurious task complete notification: " + taId + " Current state = " + taStatus);
        return;
    }
    ta.setStatus(TaskAttempt.TaskStatus.COMPLETED, null);
    ta.setEndTime(System.currentTimeMillis());
    if (lastAttempt.decrementPendingTasksCounter() == 0) {
        lastAttempt.setStatus(TaskClusterAttempt.TaskClusterStatus.COMPLETED);
        lastAttempt.setEndTime(System.currentTimeMillis());
        inProgressTaskClusters.remove(tc);
        startRunnableActivityClusters();
    }
}
Also used : TaskClusterAttempt(org.apache.hyracks.control.cc.job.TaskClusterAttempt) TaskAttemptId(org.apache.hyracks.api.dataflow.TaskAttemptId) TaskCluster(org.apache.hyracks.control.cc.job.TaskCluster) TaskAttempt(org.apache.hyracks.control.cc.job.TaskAttempt)

Example 8 with TaskAttemptId

use of org.apache.hyracks.api.dataflow.TaskAttemptId in project asterixdb by apache.

the class JobletProfile method readFields.

@Override
public void readFields(DataInput input) throws IOException {
    super.readFields(input);
    nodeId = input.readUTF();
    int size = input.readInt();
    taskProfiles = new HashMap<TaskAttemptId, TaskProfile>();
    for (int i = 0; i < size; i++) {
        TaskAttemptId key = TaskAttemptId.create(input);
        TaskProfile value = TaskProfile.create(input);
        taskProfiles.put(key, value);
    }
}
Also used : TaskAttemptId(org.apache.hyracks.api.dataflow.TaskAttemptId)

Example 9 with TaskAttemptId

use of org.apache.hyracks.api.dataflow.TaskAttemptId in project asterixdb by apache.

the class TestUtils method create.

public static IHyracksTaskContext create(int frameSize) {
    try {
        IOManager ioManager = createIoManager();
        INCServiceContext serviceCtx = new TestNCServiceContext(ioManager, null);
        TestJobletContext jobletCtx = new TestJobletContext(frameSize, serviceCtx, new JobId(0));
        TaskAttemptId tid = new TaskAttemptId(new TaskId(new ActivityId(new OperatorDescriptorId(0), 0), 0), 0);
        IHyracksTaskContext taskCtx = new TestTaskContext(jobletCtx, tid);
        return taskCtx;
    } catch (HyracksException e) {
        throw new RuntimeException(e);
    }
}
Also used : OperatorDescriptorId(org.apache.hyracks.api.dataflow.OperatorDescriptorId) TaskId(org.apache.hyracks.api.dataflow.TaskId) IOManager(org.apache.hyracks.control.nc.io.IOManager) TaskAttemptId(org.apache.hyracks.api.dataflow.TaskAttemptId) ActivityId(org.apache.hyracks.api.dataflow.ActivityId) HyracksException(org.apache.hyracks.api.exceptions.HyracksException) INCServiceContext(org.apache.hyracks.api.application.INCServiceContext) IHyracksTaskContext(org.apache.hyracks.api.context.IHyracksTaskContext) JobId(org.apache.hyracks.api.job.JobId)

Aggregations

TaskAttemptId (org.apache.hyracks.api.dataflow.TaskAttemptId)9 ActivityId (org.apache.hyracks.api.dataflow.ActivityId)4 TaskId (org.apache.hyracks.api.dataflow.TaskId)4 HyracksException (org.apache.hyracks.api.exceptions.HyracksException)4 ArrayList (java.util.ArrayList)3 List (java.util.List)3 OperatorDescriptorId (org.apache.hyracks.api.dataflow.OperatorDescriptorId)3 TaskAttempt (org.apache.hyracks.control.cc.job.TaskAttempt)3 TaskCluster (org.apache.hyracks.control.cc.job.TaskCluster)3 TaskClusterAttempt (org.apache.hyracks.control.cc.job.TaskClusterAttempt)3 HashMap (java.util.HashMap)2 Map (java.util.Map)2 INCServiceContext (org.apache.hyracks.api.application.INCServiceContext)2 IConnectorDescriptor (org.apache.hyracks.api.dataflow.IConnectorDescriptor)2 IConnectorPolicy (org.apache.hyracks.api.dataflow.connectors.IConnectorPolicy)2 ActivityClusterGraph (org.apache.hyracks.api.job.ActivityClusterGraph)2 JobId (org.apache.hyracks.api.job.JobId)2 INodeManager (org.apache.hyracks.control.cc.cluster.INodeManager)2 TaskAttemptDescriptor (org.apache.hyracks.control.common.job.TaskAttemptDescriptor)2 Joblet (org.apache.hyracks.control.nc.Joblet)2