use of org.apache.hyracks.api.constraints.expressions.LValueConstraintExpression in project asterixdb by apache.
the class FeedOperations method combineIntakeCollectJobs.
private static JobSpecification combineIntakeCollectJobs(MetadataProvider metadataProvider, Feed feed, JobSpecification intakeJob, List<JobSpecification> jobsList, List<FeedConnection> feedConnections, String[] intakeLocations) throws AlgebricksException, HyracksDataException {
JobSpecification jobSpec = new JobSpecification(intakeJob.getFrameSize());
// copy ingestor
FeedIntakeOperatorDescriptor firstOp = (FeedIntakeOperatorDescriptor) intakeJob.getOperatorMap().get(new OperatorDescriptorId(0));
FeedIntakeOperatorDescriptor ingestionOp;
if (firstOp.getAdaptorFactory() == null) {
ingestionOp = new FeedIntakeOperatorDescriptor(jobSpec, feed, firstOp.getAdaptorLibraryName(), firstOp.getAdaptorFactoryClassName(), firstOp.getAdapterOutputType(), firstOp.getPolicyAccessor(), firstOp.getOutputRecordDescriptors()[0]);
} else {
ingestionOp = new FeedIntakeOperatorDescriptor(jobSpec, feed, firstOp.getAdaptorFactory(), firstOp.getAdapterOutputType(), firstOp.getPolicyAccessor(), firstOp.getOutputRecordDescriptors()[0]);
}
// create replicator
ReplicateOperatorDescriptor replicateOp = new ReplicateOperatorDescriptor(jobSpec, ingestionOp.getOutputRecordDescriptors()[0], jobsList.size());
jobSpec.connect(new OneToOneConnectorDescriptor(jobSpec), ingestionOp, 0, replicateOp, 0);
PartitionConstraintHelper.addAbsoluteLocationConstraint(jobSpec, ingestionOp, intakeLocations);
PartitionConstraintHelper.addAbsoluteLocationConstraint(jobSpec, replicateOp, intakeLocations);
// Loop over the jobs to copy operators and connections
Map<OperatorDescriptorId, OperatorDescriptorId> operatorIdMapping = new HashMap<>();
Map<ConnectorDescriptorId, ConnectorDescriptorId> connectorIdMapping = new HashMap<>();
Map<OperatorDescriptorId, List<LocationConstraint>> operatorLocations = new HashMap<>();
Map<OperatorDescriptorId, Integer> operatorCounts = new HashMap<>();
List<JobId> jobIds = new ArrayList<>();
FeedMetaOperatorDescriptor metaOp;
for (int iter1 = 0; iter1 < jobsList.size(); iter1++) {
FeedConnection curFeedConnection = feedConnections.get(iter1);
JobSpecification subJob = jobsList.get(iter1);
operatorIdMapping.clear();
Map<OperatorDescriptorId, IOperatorDescriptor> operatorsMap = subJob.getOperatorMap();
String datasetName = feedConnections.get(iter1).getDatasetName();
FeedConnectionId feedConnectionId = new FeedConnectionId(ingestionOp.getEntityId(), datasetName);
FeedPolicyEntity feedPolicyEntity = FeedMetadataUtil.validateIfPolicyExists(curFeedConnection.getDataverseName(), curFeedConnection.getPolicyName(), metadataProvider.getMetadataTxnContext());
for (Map.Entry<OperatorDescriptorId, IOperatorDescriptor> entry : operatorsMap.entrySet()) {
IOperatorDescriptor opDesc = entry.getValue();
OperatorDescriptorId oldId = opDesc.getOperatorId();
OperatorDescriptorId opId = null;
if (opDesc instanceof LSMTreeInsertDeleteOperatorDescriptor && ((LSMTreeInsertDeleteOperatorDescriptor) opDesc).isPrimary()) {
metaOp = new FeedMetaOperatorDescriptor(jobSpec, feedConnectionId, opDesc, feedPolicyEntity.getProperties(), FeedRuntimeType.STORE);
opId = metaOp.getOperatorId();
opDesc.setOperatorId(opId);
} else {
if (opDesc instanceof AlgebricksMetaOperatorDescriptor) {
AlgebricksMetaOperatorDescriptor algOp = (AlgebricksMetaOperatorDescriptor) opDesc;
IPushRuntimeFactory[] runtimeFactories = algOp.getPipeline().getRuntimeFactories();
// Tweak AssignOp to work with messages
if (runtimeFactories[0] instanceof AssignRuntimeFactory && runtimeFactories.length > 1) {
IConnectorDescriptor connectorDesc = subJob.getOperatorInputMap().get(opDesc.getOperatorId()).get(0);
// anything on the network interface needs to be message compatible
if (connectorDesc instanceof MToNPartitioningConnectorDescriptor) {
metaOp = new FeedMetaOperatorDescriptor(jobSpec, feedConnectionId, opDesc, feedPolicyEntity.getProperties(), FeedRuntimeType.COMPUTE);
opId = metaOp.getOperatorId();
opDesc.setOperatorId(opId);
}
}
}
if (opId == null) {
opId = jobSpec.createOperatorDescriptorId(opDesc);
}
}
operatorIdMapping.put(oldId, opId);
}
// copy connectors
connectorIdMapping.clear();
for (Entry<ConnectorDescriptorId, IConnectorDescriptor> entry : subJob.getConnectorMap().entrySet()) {
IConnectorDescriptor connDesc = entry.getValue();
ConnectorDescriptorId newConnId;
if (connDesc instanceof MToNPartitioningConnectorDescriptor) {
MToNPartitioningConnectorDescriptor m2nConn = (MToNPartitioningConnectorDescriptor) connDesc;
connDesc = new MToNPartitioningWithMessageConnectorDescriptor(jobSpec, m2nConn.getTuplePartitionComputerFactory());
newConnId = connDesc.getConnectorId();
} else {
newConnId = jobSpec.createConnectorDescriptor(connDesc);
}
connectorIdMapping.put(entry.getKey(), newConnId);
}
// make connections between operators
for (Entry<ConnectorDescriptorId, Pair<Pair<IOperatorDescriptor, Integer>, Pair<IOperatorDescriptor, Integer>>> entry : subJob.getConnectorOperatorMap().entrySet()) {
ConnectorDescriptorId newId = connectorIdMapping.get(entry.getKey());
IConnectorDescriptor connDesc = jobSpec.getConnectorMap().get(newId);
Pair<IOperatorDescriptor, Integer> leftOp = entry.getValue().getLeft();
Pair<IOperatorDescriptor, Integer> rightOp = entry.getValue().getRight();
IOperatorDescriptor leftOpDesc = jobSpec.getOperatorMap().get(leftOp.getLeft().getOperatorId());
IOperatorDescriptor rightOpDesc = jobSpec.getOperatorMap().get(rightOp.getLeft().getOperatorId());
if (leftOp.getLeft() instanceof FeedCollectOperatorDescriptor) {
jobSpec.connect(new OneToOneConnectorDescriptor(jobSpec), replicateOp, iter1, leftOpDesc, leftOp.getRight());
}
jobSpec.connect(connDesc, leftOpDesc, leftOp.getRight(), rightOpDesc, rightOp.getRight());
}
// prepare for setting partition constraints
operatorLocations.clear();
operatorCounts.clear();
for (Constraint constraint : subJob.getUserConstraints()) {
LValueConstraintExpression lexpr = constraint.getLValue();
ConstraintExpression cexpr = constraint.getRValue();
OperatorDescriptorId opId;
switch(lexpr.getTag()) {
case PARTITION_COUNT:
opId = ((PartitionCountExpression) lexpr).getOperatorDescriptorId();
operatorCounts.put(operatorIdMapping.get(opId), (int) ((ConstantExpression) cexpr).getValue());
break;
case PARTITION_LOCATION:
opId = ((PartitionLocationExpression) lexpr).getOperatorDescriptorId();
IOperatorDescriptor opDesc = jobSpec.getOperatorMap().get(operatorIdMapping.get(opId));
List<LocationConstraint> locations = operatorLocations.get(opDesc.getOperatorId());
if (locations == null) {
locations = new ArrayList<>();
operatorLocations.put(opDesc.getOperatorId(), locations);
}
String location = (String) ((ConstantExpression) cexpr).getValue();
LocationConstraint lc = new LocationConstraint(location, ((PartitionLocationExpression) lexpr).getPartition());
locations.add(lc);
break;
default:
break;
}
}
// set absolute location constraints
for (Entry<OperatorDescriptorId, List<LocationConstraint>> entry : operatorLocations.entrySet()) {
IOperatorDescriptor opDesc = jobSpec.getOperatorMap().get(entry.getKey());
// why do we need to sort?
Collections.sort(entry.getValue(), (LocationConstraint o1, LocationConstraint o2) -> {
return o1.partition - o2.partition;
});
String[] locations = new String[entry.getValue().size()];
for (int j = 0; j < locations.length; ++j) {
locations[j] = entry.getValue().get(j).location;
}
PartitionConstraintHelper.addAbsoluteLocationConstraint(jobSpec, opDesc, locations);
}
// set count constraints
for (Entry<OperatorDescriptorId, Integer> entry : operatorCounts.entrySet()) {
IOperatorDescriptor opDesc = jobSpec.getOperatorMap().get(entry.getKey());
if (!operatorLocations.keySet().contains(entry.getKey())) {
PartitionConstraintHelper.addPartitionCountConstraint(jobSpec, opDesc, entry.getValue());
}
}
// roots
for (OperatorDescriptorId root : subJob.getRoots()) {
jobSpec.addRoot(jobSpec.getOperatorMap().get(operatorIdMapping.get(root)));
}
jobIds.add(((JobEventListenerFactory) subJob.getJobletEventListenerFactory()).getJobId());
}
// jobEventListenerFactory
jobSpec.setJobletEventListenerFactory(new MultiTransactionJobletEventListenerFactory(jobIds, true));
// useConnectorSchedulingPolicy
jobSpec.setUseConnectorPolicyForScheduling(jobsList.get(0).isUseConnectorPolicyForScheduling());
// connectorAssignmentPolicy
jobSpec.setConnectorPolicyAssignmentPolicy(jobsList.get(0).getConnectorPolicyAssignmentPolicy());
return jobSpec;
}
use of org.apache.hyracks.api.constraints.expressions.LValueConstraintExpression in project asterixdb by apache.
the class JobExecutor method assignLocation.
private String assignLocation(ActivityClusterGraph acg, Map<TaskId, LValueConstraintExpression> locationMap, TaskId tid, TaskAttempt taskAttempt) throws HyracksException {
ActivityId aid = tid.getActivityId();
ActivityCluster ac = acg.getActivityMap().get(aid);
Set<ActivityId> blockers = ac.getBlocked2BlockerMap().get(aid);
String nodeId = null;
if (blockers != null) {
for (ActivityId blocker : blockers) {
nodeId = findTaskLocation(new TaskId(blocker, tid.getPartition()));
if (nodeId != null) {
break;
}
}
}
INodeManager nodeManager = ccs.getNodeManager();
Collection<String> liveNodes = nodeManager.getAllNodeIds();
if (nodeId == null) {
LValueConstraintExpression pLocationExpr = locationMap.get(tid);
Object location = solver.getValue(pLocationExpr);
if (location == null) {
// pick any
nodeId = liveNodes.toArray(new String[liveNodes.size()])[random.nextInt(1) % liveNodes.size()];
} else if (location instanceof String) {
nodeId = (String) location;
} else if (location instanceof String[]) {
for (String choice : (String[]) location) {
if (liveNodes.contains(choice)) {
nodeId = choice;
break;
}
}
if (nodeId == null) {
throw new HyracksException("No satisfiable location found for " + taskAttempt.getTaskAttemptId());
}
} else {
throw new HyracksException("Unknown type of value for " + pLocationExpr + ": " + location + "(" + location.getClass() + ")");
}
}
if (nodeId == null) {
throw new HyracksException("No satisfiable location found for " + taskAttempt.getTaskAttemptId());
}
if (!liveNodes.contains(nodeId)) {
throw new HyracksException("Node " + nodeId + " not live");
}
return nodeId;
}
use of org.apache.hyracks.api.constraints.expressions.LValueConstraintExpression in project asterixdb by apache.
the class PartitionConstraintSolver method solveLValue.
private Solution solveLValue(LValueConstraintExpression lv, Set<LValueConstraintExpression> inProcess) {
if (inProcess.contains(lv)) {
return new Solution(null, Solution.Status.CYCLE);
}
Solution result = null;
inProcess.add(lv);
Set<ConstraintExpression> rValues = constraints.get(lv);
if (rValues == null) {
return new Solution(null, Solution.Status.NOT_BOUND);
}
for (ConstraintExpression ce : rValues) {
Solution solution = solve(ce, inProcess);
if (solution != null && solution.status == Solution.Status.FOUND) {
result = solution;
break;
}
}
if (result != null) {
rValues.clear();
rValues.add(new ConstantExpression(result.value));
}
inProcess.remove(lv);
return result;
}
use of org.apache.hyracks.api.constraints.expressions.LValueConstraintExpression in project asterixdb by apache.
the class ActivityClusterPlanner method computePartitionCounts.
private Map<ActivityId, ActivityPartitionDetails> computePartitionCounts(ActivityCluster ac) throws HyracksException {
PartitionConstraintSolver solver = executor.getSolver();
Set<LValueConstraintExpression> lValues = new HashSet<>();
for (ActivityId anId : ac.getActivityMap().keySet()) {
lValues.add(new PartitionCountExpression(anId.getOperatorDescriptorId()));
}
solver.solve(lValues);
Map<OperatorDescriptorId, Integer> nPartMap = new HashMap<>();
for (LValueConstraintExpression lv : lValues) {
Object value = solver.getValue(lv);
if (value == null) {
throw new HyracksException("No value found for " + lv);
}
if (!(value instanceof Number)) {
throw new HyracksException("Unexpected type of value bound to " + lv + ": " + value.getClass() + "(" + value + ")");
}
int nParts = ((Number) value).intValue();
if (nParts <= 0) {
throw new HyracksException("Unsatisfiable number of partitions for " + lv + ": " + nParts);
}
nPartMap.put(((PartitionCountExpression) lv).getOperatorDescriptorId(), nParts);
}
Map<ActivityId, ActivityPartitionDetails> activityPartsMap = new HashMap<>();
for (ActivityId anId : ac.getActivityMap().keySet()) {
int nParts = nPartMap.get(anId.getOperatorDescriptorId());
int[] nInputPartitions = null;
List<IConnectorDescriptor> inputs = ac.getActivityInputMap().get(anId);
if (inputs != null) {
nInputPartitions = new int[inputs.size()];
for (int i = 0; i < nInputPartitions.length; ++i) {
ConnectorDescriptorId cdId = inputs.get(i).getConnectorId();
ActivityId aid = ac.getProducerActivity(cdId);
Integer nPartInt = nPartMap.get(aid.getOperatorDescriptorId());
nInputPartitions[i] = nPartInt;
}
}
int[] nOutputPartitions = null;
List<IConnectorDescriptor> outputs = ac.getActivityOutputMap().get(anId);
if (outputs != null) {
nOutputPartitions = new int[outputs.size()];
for (int i = 0; i < nOutputPartitions.length; ++i) {
ConnectorDescriptorId cdId = outputs.get(i).getConnectorId();
ActivityId aid = ac.getConsumerActivity(cdId);
Integer nPartInt = nPartMap.get(aid.getOperatorDescriptorId());
nOutputPartitions[i] = nPartInt;
}
}
ActivityPartitionDetails apd = new ActivityPartitionDetails(nParts, nInputPartitions, nOutputPartitions);
activityPartsMap.put(anId, apd);
}
return activityPartsMap;
}
use of org.apache.hyracks.api.constraints.expressions.LValueConstraintExpression 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);
}
Aggregations