private TaskCluster getTaskCluster(TaskId tid) {
JobRun run = scheduler.getJobRun();
ActivityCluster ac = run.getActivityClusterGraph().getActivityMap().get(tid.getActivityId());
ActivityClusterPlan acp = run.getActivityClusterPlanMap().get(ac.getId());
Task[] tasks = acp.getActivityPlanMap().get(tid.getActivityId()).getTasks();
Task task = tasks[tid.getPartition()];
assert task.getTaskId().equals(tid);
return task.getTaskCluster();
}
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;
}
private TaskCluster[] buildConnectorPolicyAwareTaskClusters(
ActivityCluster ac,
Map<ActivityId, ActivityPlan> activityPlanMap,
Map<TaskId, List<Pair<TaskId, ConnectorDescriptorId>>> taskConnectivity) {
Map<TaskId, Set<TaskId>> taskClusterMap = new HashMap<TaskId, Set<TaskId>>();
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>();
TaskId tid = t.getTaskId();
cluster.add(tid);
taskClusterMap.put(tid, cluster);
}
}
JobRun jobRun = scheduler.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());
}
}
}
/*
* taskClusterMap contains for every TID x, x -> { coscheduled consumer TIDs U x }
* We compute the transitive closure of this 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<TaskId, Integer>();
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<List<TaskId>>();
for (int i = pending.nextSetBit(0); i >= 0; i = pending.nextSetBit(i)) {
List<TaskId> cluster = new ArrayList<TaskId>();
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<TaskCluster>();
counter = 0;
for (List<TaskId> cluster : clusters) {
List<Task> taskStates = new ArrayList<Task>();
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);
}
}
TaskCluster[] taskClusters = tcSet.toArray(new TaskCluster[tcSet.size()]);
return taskClusters;
}