Examples with TaskId org.apache.hyracks.api.dataflow.TaskId used on opensource projects

Example 6 with TaskId

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

the class ActivityClusterPlanner method buildConnectorPolicyAwareTaskClusters.

private TaskCluster[] buildConnectorPolicyAwareTaskClusters(ActivityCluster ac, Map<ActivityId, ActivityPlan> activityPlanMap, Map<TaskId, List<Pair<TaskId, ConnectorDescriptorId>>> taskConnectivity) {
    Map<TaskId, Set<TaskId>> taskClusterMap = new HashMap<>();
    for (ActivityId anId : ac.getActivityMap().keySet()) {
        ActivityPlan ap = activityPlanMap.get(anId);
        Task[] tasks = ap.getTasks();
        for (Task t : tasks) {
            Set<TaskId> cluster = new HashSet<>();
            TaskId tid = t.getTaskId();
            cluster.add(tid);
            taskClusterMap.put(tid, cluster);
        }
    }
    JobRun jobRun = executor.getJobRun();
    Map<ConnectorDescriptorId, IConnectorPolicy> connectorPolicies = jobRun.getConnectorPolicyMap();
    for (Map.Entry<TaskId, List<Pair<TaskId, ConnectorDescriptorId>>> e : taskConnectivity.entrySet()) {
        Set<TaskId> cluster = taskClusterMap.get(e.getKey());
        for (Pair<TaskId, ConnectorDescriptorId> p : e.getValue()) {
            IConnectorPolicy cPolicy = connectorPolicies.get(p.getRight());
            if (cPolicy.requiresProducerConsumerCoscheduling()) {
                cluster.add(p.getLeft());
            }
        }
    }
    /*
         * We compute the transitive closure of this (producer-consumer) relation to find the largest set of
         * tasks that need to be co-scheduled.
         */
    int counter = 0;
    TaskId[] ordinalList = new TaskId[taskClusterMap.size()];
    Map<TaskId, Integer> ordinalMap = new HashMap<>();
    for (TaskId tid : taskClusterMap.keySet()) {
        ordinalList[counter] = tid;
        ordinalMap.put(tid, counter);
        ++counter;
    }
    int n = ordinalList.length;
    BitSet[] paths = new BitSet[n];
    for (Map.Entry<TaskId, Set<TaskId>> e : taskClusterMap.entrySet()) {
        int i = ordinalMap.get(e.getKey());
        BitSet bsi = paths[i];
        if (bsi == null) {
            bsi = new BitSet(n);
            paths[i] = bsi;
        }
        for (TaskId ttid : e.getValue()) {
            int j = ordinalMap.get(ttid);
            paths[i].set(j);
            BitSet bsj = paths[j];
            if (bsj == null) {
                bsj = new BitSet(n);
                paths[j] = bsj;
            }
            bsj.set(i);
        }
    }
    for (int k = 0; k < n; ++k) {
        for (int i = paths[k].nextSetBit(0); i >= 0; i = paths[k].nextSetBit(i + 1)) {
            for (int j = paths[i].nextClearBit(0); j < n && j >= 0; j = paths[i].nextClearBit(j + 1)) {
                paths[i].set(j, paths[k].get(j));
                paths[j].set(i, paths[i].get(j));
            }
        }
    }
    BitSet pending = new BitSet(n);
    pending.set(0, n);
    List<List<TaskId>> clusters = new ArrayList<>();
    for (int i = pending.nextSetBit(0); i >= 0; i = pending.nextSetBit(i)) {
        List<TaskId> cluster = new ArrayList<>();
        for (int j = paths[i].nextSetBit(0); j >= 0; j = paths[i].nextSetBit(j + 1)) {
            cluster.add(ordinalList[j]);
            pending.clear(j);
        }
        clusters.add(cluster);
    }
    List<TaskCluster> tcSet = new ArrayList<>();
    counter = 0;
    for (List<TaskId> cluster : clusters) {
        List<Task> taskStates = new ArrayList<>();
        for (TaskId tid : cluster) {
            taskStates.add(activityPlanMap.get(tid.getActivityId()).getTasks()[tid.getPartition()]);
        }
        TaskCluster tc = new TaskCluster(new TaskClusterId(ac.getId(), counter++), ac, taskStates.toArray(new Task[taskStates.size()]));
        tcSet.add(tc);
        for (TaskId tid : cluster) {
            activityPlanMap.get(tid.getActivityId()).getTasks()[tid.getPartition()].setTaskCluster(tc);
        }
    }
    return tcSet.toArray(new TaskCluster[tcSet.size()]);
}

Example 7 with TaskId

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

the class ActivityClusterPlanner method computeTaskClusters.

private TaskCluster[] computeTaskClusters(ActivityCluster ac, JobRun jobRun, Map<ActivityId, ActivityPlan> activityPlanMap) {
    Set<ActivityId> activities = ac.getActivityMap().keySet();
    Map<TaskId, List<Pair<TaskId, ConnectorDescriptorId>>> taskConnectivity = computeTaskConnectivity(jobRun, activityPlanMap, activities);
    TaskCluster[] taskClusters = ac.getActivityClusterGraph().isUseConnectorPolicyForScheduling() ? buildConnectorPolicyAwareTaskClusters(ac, activityPlanMap, taskConnectivity) : buildConnectorPolicyUnawareTaskClusters(ac, activityPlanMap);
    for (TaskCluster tc : taskClusters) {
        Set<TaskCluster> tcDependencyTaskClusters = tc.getDependencyTaskClusters();
        for (Task ts : tc.getTasks()) {
            TaskId tid = ts.getTaskId();
            List<Pair<TaskId, ConnectorDescriptorId>> cInfoList = taskConnectivity.get(tid);
            if (cInfoList != null) {
                for (Pair<TaskId, ConnectorDescriptorId> p : cInfoList) {
                    Task targetTS = activityPlanMap.get(p.getLeft().getActivityId()).getTasks()[p.getLeft().getPartition()];
                    TaskCluster targetTC = targetTS.getTaskCluster();
                    if (targetTC != tc) {
                        ConnectorDescriptorId cdId = p.getRight();
                        PartitionId pid = new PartitionId(jobRun.getJobId(), cdId, tid.getPartition(), p.getLeft().getPartition());
                        tc.getProducedPartitions().add(pid);
                        targetTC.getRequiredPartitions().add(pid);
                        partitionProducingTaskClusterMap.put(pid, tc);
                    }
                }
            }
            for (TaskId dTid : ts.getDependencies()) {
                TaskCluster dTC = getTaskCluster(dTid);
                dTC.getDependentTaskClusters().add(tc);
                tcDependencyTaskClusters.add(dTC);
            }
        }
    }
    return taskClusters;
}

Example 8 with TaskId

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

the class ActivityClusterPlanner method computeTaskConnectivity.

private Map<TaskId, List<Pair<TaskId, ConnectorDescriptorId>>> computeTaskConnectivity(JobRun jobRun, Map<ActivityId, ActivityPlan> activityPlanMap, Set<ActivityId> activities) {
    Map<TaskId, List<Pair<TaskId, ConnectorDescriptorId>>> taskConnectivity = new HashMap<>();
    ActivityClusterGraph acg = jobRun.getActivityClusterGraph();
    BitSet targetBitmap = new BitSet();
    for (ActivityId ac1 : activities) {
        ActivityCluster ac = acg.getActivityMap().get(ac1);
        Task[] ac1TaskStates = activityPlanMap.get(ac1).getTasks();
        int nProducers = ac1TaskStates.length;
        List<IConnectorDescriptor> outputConns = ac.getActivityOutputMap().get(ac1);
        if (outputConns == null) {
            continue;
        }
        for (IConnectorDescriptor c : outputConns) {
            ConnectorDescriptorId cdId = c.getConnectorId();
            ActivityId ac2 = ac.getConsumerActivity(cdId);
            Task[] ac2TaskStates = activityPlanMap.get(ac2).getTasks();
            int nConsumers = ac2TaskStates.length;
            if (c.allProducersToAllConsumers()) {
                List<Pair<TaskId, ConnectorDescriptorId>> cInfoList = new ArrayList<>();
                for (int j = 0; j < nConsumers; j++) {
                    TaskId targetTID = ac2TaskStates[j].getTaskId();
                    cInfoList.add(Pair.of(targetTID, cdId));
                }
                for (int i = 0; i < nProducers; ++i) {
                    taskConnectivity.put(ac1TaskStates[i].getTaskId(), cInfoList);
                }
                continue;
            }
            for (int i = 0; i < nProducers; ++i) {
                c.indicateTargetPartitions(nProducers, nConsumers, i, targetBitmap);
                List<Pair<TaskId, ConnectorDescriptorId>> cInfoList = taskConnectivity.get(ac1TaskStates[i].getTaskId());
                if (cInfoList == null) {
                    cInfoList = new ArrayList<>();
                    taskConnectivity.put(ac1TaskStates[i].getTaskId(), cInfoList);
                }
                for (int j = targetBitmap.nextSetBit(0); j >= 0; j = targetBitmap.nextSetBit(j + 1)) {
                    TaskId targetTID = ac2TaskStates[j].getTaskId();
                    cInfoList.add(Pair.of(targetTID, cdId));
                }
            }
        }
    }
    return taskConnectivity;
}

Example 9 with TaskId

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

the class AbstractTaskLifecycleWork method runWork.

@Override
public final void runWork() {
    IJobManager jobManager = ccs.getJobManager();
    JobRun run = jobManager.get(jobId);
    if (run != null) {
        TaskId tid = taId.getTaskId();
        Map<ActivityId, ActivityCluster> activityClusterMap = run.getActivityClusterGraph().getActivityMap();
        ActivityCluster ac = activityClusterMap.get(tid.getActivityId());
        if (ac != null) {
            Map<ActivityId, ActivityPlan> taskStateMap = run.getActivityClusterPlanMap().get(ac.getId()).getActivityPlanMap();
            Task[] taskStates = taskStateMap.get(tid.getActivityId()).getTasks();
            if (taskStates != null && taskStates.length > tid.getPartition()) {
                Task ts = taskStates[tid.getPartition()];
                TaskCluster tc = ts.getTaskCluster();
                List<TaskClusterAttempt> taskClusterAttempts = tc.getAttempts();
                if (taskClusterAttempts != null && taskClusterAttempts.size() > taId.getAttempt()) {
                    TaskClusterAttempt tca = taskClusterAttempts.get(taId.getAttempt());
                    TaskAttempt ta = tca.getTaskAttempts().get(tid);
                    if (ta != null) {
                        performEvent(ta);
                    }
                }
            }
        }
    }
}

Example 10 with TaskId

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

the class JobRun method toJSON.

public ObjectNode toJSON() {
    ObjectMapper om = new ObjectMapper();
    ObjectNode result = om.createObjectNode();
    result.put("job-id", jobId.toString());
    result.putPOJO("status", getStatus());
    result.put("create-time", getCreateTime());
    result.put("start-time", getStartTime());
    result.put("end-time", getEndTime());
    ArrayNode aClusters = om.createArrayNode();
    for (ActivityCluster ac : acg.getActivityClusterMap().values()) {
        ObjectNode acJSON = om.createObjectNode();
        acJSON.put("activity-cluster-id", String.valueOf(ac.getId()));
        ArrayNode activitiesJSON = om.createArrayNode();
        for (ActivityId aid : ac.getActivityMap().keySet()) {
            activitiesJSON.addPOJO(aid);
        }
        acJSON.set("activities", activitiesJSON);
        ArrayNode dependenciesJSON = om.createArrayNode();
        for (ActivityCluster dependency : ac.getDependencies()) {
            dependenciesJSON.add(String.valueOf(dependency.getId()));
        }
        acJSON.set("dependencies", dependenciesJSON);
        ActivityClusterPlan acp = activityClusterPlanMap.get(ac.getId());
        if (acp == null) {
            acJSON.putNull("plan");
        } else {
            ObjectNode planJSON = om.createObjectNode();
            ArrayNode acTasks = om.createArrayNode();
            for (Map.Entry<ActivityId, ActivityPlan> e : acp.getActivityPlanMap().entrySet()) {
                ActivityPlan acPlan = e.getValue();
                ObjectNode entry = om.createObjectNode();
                entry.put("activity-id", e.getKey().toString());
                ActivityPartitionDetails apd = acPlan.getActivityPartitionDetails();
                entry.put("partition-count", apd.getPartitionCount());
                ArrayNode inPartCountsJSON = om.createArrayNode();
                int[] inPartCounts = apd.getInputPartitionCounts();
                if (inPartCounts != null) {
                    for (int i : inPartCounts) {
                        inPartCountsJSON.add(i);
                    }
                }
                entry.set("input-partition-counts", inPartCountsJSON);
                ArrayNode outPartCountsJSON = om.createArrayNode();
                int[] outPartCounts = apd.getOutputPartitionCounts();
                if (outPartCounts != null) {
                    for (int o : outPartCounts) {
                        outPartCountsJSON.add(o);
                    }
                }
                entry.set("output-partition-counts", outPartCountsJSON);
                ArrayNode tasks = om.createArrayNode();
                for (Task t : acPlan.getTasks()) {
                    ObjectNode task = om.createObjectNode();
                    task.put("task-id", t.getTaskId().toString());
                    ArrayNode dependentTasksJSON = om.createArrayNode();
                    for (TaskId dependent : t.getDependents()) {
                        dependentTasksJSON.add(dependent.toString());
                        task.set("dependents", dependentTasksJSON);
                        ArrayNode dependencyTasksJSON = om.createArrayNode();
                        for (TaskId dependency : t.getDependencies()) {
                            dependencyTasksJSON.add(dependency.toString());
                        }
                        task.set("dependencies", dependencyTasksJSON);
                        tasks.add(task);
                    }
                    entry.set("tasks", tasks);
                    acTasks.add(entry);
                }
            }
            planJSON.set("activities", acTasks);
            ArrayNode tClusters = om.createArrayNode();
            for (TaskCluster tc : acp.getTaskClusters()) {
                ObjectNode c = om.createObjectNode();
                c.put("task-cluster-id", String.valueOf(tc.getTaskClusterId()));
                ArrayNode tasksAry = om.createArrayNode();
                for (Task t : tc.getTasks()) {
                    tasksAry.add(t.getTaskId().toString());
                }
                c.set("tasks", tasksAry);
                ArrayNode prodParts = om.createArrayNode();
                for (PartitionId p : tc.getProducedPartitions()) {
                    prodParts.add(p.toString());
                }
                c.set("produced-partitions", prodParts);
                ArrayNode reqdParts = om.createArrayNode();
                for (PartitionId p : tc.getRequiredPartitions()) {
                    reqdParts.add(p.toString());
                }
                c.set("required-partitions", reqdParts);
                ArrayNode attempts = om.createArrayNode();
                List<TaskClusterAttempt> tcAttempts = tc.getAttempts();
                if (tcAttempts != null) {
                    for (TaskClusterAttempt tca : tcAttempts) {
                        ObjectNode attempt = om.createObjectNode();
                        attempt.put("attempt", tca.getAttempt());
                        attempt.putPOJO("status", tca.getStatus());
                        attempt.put("start-time", tca.getStartTime());
                        attempt.put("end-time", tca.getEndTime());
                        ArrayNode taskAttempts = om.createArrayNode();
                        for (TaskAttempt ta : tca.getTaskAttempts().values()) {
                            ObjectNode taskAttempt = om.createObjectNode();
                            taskAttempt.putPOJO("task-id", ta.getTaskAttemptId().getTaskId());
                            taskAttempt.putPOJO("task-attempt-id", ta.getTaskAttemptId());
                            taskAttempt.putPOJO("status", ta.getStatus());
                            taskAttempt.put("node-id", ta.getNodeId());
                            taskAttempt.put("start-time", ta.getStartTime());
                            taskAttempt.put("end-time", ta.getEndTime());
                            List<Exception> exceptions = ta.getExceptions();
                            if (exceptions != null && !exceptions.isEmpty()) {
                                List<Exception> filteredExceptions = ExceptionUtils.getActualExceptions(exceptions);
                                for (Exception exception : filteredExceptions) {
                                    StringWriter exceptionWriter = new StringWriter();
                                    exception.printStackTrace(new PrintWriter(exceptionWriter));
                                    taskAttempt.put("failure-details", exceptionWriter.toString());
                                }
                            }
                            taskAttempts.add(taskAttempt);
                        }
                        attempt.set("task-attempts", taskAttempts);
                        attempts.add(attempt);
                    }
                }
                c.set("attempts", attempts);
                tClusters.add(c);
            }
            planJSON.set("task-clusters", tClusters);
            acJSON.set("plan", planJSON);
        }
        aClusters.add(acJSON);
    }
    result.set("activity-clusters", aClusters);
    result.set("profile", profile.toJSON());
    return result;
}