/**
* @param sessionConfig session configuration
* @param numContainers number of containers to pre-warm
* @param localResources additional resources to pre-warm with
* @return prewarm context object
*/
public PreWarmContext createPreWarmContext(
TezSessionConfiguration sessionConfig,
int numContainers,
Map<String, LocalResource> localResources)
throws IOException, TezException {
Configuration conf = sessionConfig.getTezConfiguration();
ProcessorDescriptor prewarmProcDescriptor =
new ProcessorDescriptor(HivePreWarmProcessor.class.getName());
prewarmProcDescriptor.setUserPayload(MRHelpers.createUserPayloadFromConf(conf));
PreWarmContext context =
new PreWarmContext(
prewarmProcDescriptor,
getContainerResource(conf),
numContainers,
new VertexLocationHint(null));
Map<String, LocalResource> combinedResources = new HashMap<String, LocalResource>();
combinedResources.putAll(sessionConfig.getSessionResources());
if (localResources != null) {
combinedResources.putAll(localResources);
}
context.setLocalResources(combinedResources);
/* boiler plate task env */
Map<String, String> environment = new HashMap<String, String>();
MRHelpers.updateEnvironmentForMRTasks(conf, environment, true);
context.setEnvironment(environment);
context.setJavaOpts(getContainerJavaOpts(conf));
return context;
}
// TODO remove hacky name lookup
@Override
public boolean needsWaitAfterOutputConsumable() {
Vertex vertex = getVertex();
ProcessorDescriptor processorDescriptor = vertex.getProcessorDescriptor();
if (processorDescriptor != null
&& processorDescriptor.getClassName().contains("InitialTaskWithInMemSort")) {
return true;
} else {
return false;
}
}
@Test(timeout = 10000)
public void testBasicSpeculationPerVertexConf() throws Exception {
DAG dag = DAG.create("test");
String vNameNoSpec = "A";
String vNameSpec = "B";
Vertex vA = Vertex.create(vNameNoSpec, ProcessorDescriptor.create("Proc.class"), 5);
Vertex vB = Vertex.create(vNameSpec, ProcessorDescriptor.create("Proc.class"), 5);
vA.setConf(TezConfiguration.TEZ_AM_SPECULATION_ENABLED, "false");
dag.addVertex(vA);
dag.addVertex(vB);
// min/max src fraction is set to 1. So vertices will run sequentially
dag.addEdge(
Edge.create(
vA,
vB,
EdgeProperty.create(
DataMovementType.SCATTER_GATHER,
DataSourceType.PERSISTED,
SchedulingType.SEQUENTIAL,
OutputDescriptor.create("O"),
InputDescriptor.create("I"))));
MockTezClient tezClient = createTezSession();
DAGClient dagClient = tezClient.submitDAG(dag);
DAGImpl dagImpl = (DAGImpl) mockApp.getContext().getCurrentDAG();
TezVertexID vertexId = dagImpl.getVertex(vNameSpec).getVertexId();
TezVertexID vertexIdNoSpec = dagImpl.getVertex(vNameNoSpec).getVertexId();
// original attempt is killed and speculative one is successful
TezTaskAttemptID killedTaId =
TezTaskAttemptID.getInstance(TezTaskID.getInstance(vertexId, 0), 0);
TezTaskAttemptID noSpecTaId =
TezTaskAttemptID.getInstance(TezTaskID.getInstance(vertexIdNoSpec, 0), 0);
// cause speculation trigger for both
mockLauncher.setStatusUpdatesForTask(killedTaId, 100);
mockLauncher.setStatusUpdatesForTask(noSpecTaId, 100);
mockLauncher.startScheduling(true);
dagClient.waitForCompletion();
Assert.assertEquals(DAGStatus.State.SUCCEEDED, dagClient.getDAGStatus(null).getState());
org.apache.tez.dag.app.dag.Vertex vSpec = dagImpl.getVertex(vertexId);
org.apache.tez.dag.app.dag.Vertex vNoSpec = dagImpl.getVertex(vertexIdNoSpec);
// speculation for vA but not for vB
Assert.assertTrue(
vSpec.getAllCounters().findCounter(TaskCounter.NUM_SPECULATIONS).getValue() > 0);
Assert.assertEquals(
0, vNoSpec.getAllCounters().findCounter(TaskCounter.NUM_SPECULATIONS).getValue());
tezClient.stop();
}
public void testBasicSpeculation(boolean withProgress) throws Exception {
DAG dag = DAG.create("test");
Vertex vA = Vertex.create("A", ProcessorDescriptor.create("Proc.class"), 5);
dag.addVertex(vA);
MockTezClient tezClient = createTezSession();
DAGClient dagClient = tezClient.submitDAG(dag);
DAGImpl dagImpl = (DAGImpl) mockApp.getContext().getCurrentDAG();
TezVertexID vertexId = TezVertexID.getInstance(dagImpl.getID(), 0);
// original attempt is killed and speculative one is successful
TezTaskAttemptID killedTaId =
TezTaskAttemptID.getInstance(TezTaskID.getInstance(vertexId, 0), 0);
TezTaskAttemptID successTaId =
TezTaskAttemptID.getInstance(TezTaskID.getInstance(vertexId, 0), 1);
mockLauncher.updateProgress(withProgress);
// cause speculation trigger
mockLauncher.setStatusUpdatesForTask(killedTaId, 100);
mockLauncher.startScheduling(true);
dagClient.waitForCompletion();
Assert.assertEquals(DAGStatus.State.SUCCEEDED, dagClient.getDAGStatus(null).getState());
Task task = dagImpl.getTask(killedTaId.getTaskID());
Assert.assertEquals(2, task.getAttempts().size());
Assert.assertEquals(successTaId, task.getSuccessfulAttempt().getID());
TaskAttempt killedAttempt = task.getAttempt(killedTaId);
Joiner.on(",").join(killedAttempt.getDiagnostics()).contains("Killed as speculative attempt");
Assert.assertEquals(
TaskAttemptTerminationCause.TERMINATED_EFFECTIVE_SPECULATION,
killedAttempt.getTerminationCause());
if (withProgress) {
// without progress updates occasionally more than 1 task speculates
Assert.assertEquals(
1, task.getCounters().findCounter(TaskCounter.NUM_SPECULATIONS).getValue());
Assert.assertEquals(
1, dagImpl.getAllCounters().findCounter(TaskCounter.NUM_SPECULATIONS).getValue());
org.apache.tez.dag.app.dag.Vertex v = dagImpl.getVertex(killedTaId.getTaskID().getVertexID());
Assert.assertEquals(
1, v.getAllCounters().findCounter(TaskCounter.NUM_SPECULATIONS).getValue());
}
tezClient.stop();
}
