// TODO XXX Does this really need to be a YarnConfiguration ?
protected static void initAndStartAppMaster(
final DAGAppMaster appMaster, final Configuration conf, String jobUserName)
throws IOException, InterruptedException {
Credentials credentials = UserGroupInformation.getCurrentUser().getCredentials();
UserGroupInformation.setConfiguration(conf);
UserGroupInformation appMasterUgi = UserGroupInformation.createRemoteUser(jobUserName);
appMasterUgi.addCredentials(credentials);
appMasterUgi.doAs(
new PrivilegedExceptionAction<Object>() {
@Override
public Object run() throws Exception {
appMaster.init(conf);
appMaster.start();
return null;
}
});
}
@Override
public synchronized void flush() {
user.addCredentials(credentials);
}
/**
* Main run function for the application master
*
* @throws YarnException
* @throws IOException
*/
@SuppressWarnings({"unchecked"})
public void run() throws YarnException, IOException {
LOG.info("Starting ApplicationMaster");
try {
publishApplicationAttemptEvent(
timelineClient, appAttemptID.toString(), DSEvent.DS_APP_ATTEMPT_START);
} catch (Exception e) {
LOG.error("App Attempt start event coud not be pulished for " + appAttemptID.toString(), e);
}
Credentials credentials = UserGroupInformation.getCurrentUser().getCredentials();
DataOutputBuffer dob = new DataOutputBuffer();
credentials.writeTokenStorageToStream(dob);
// Now remove the AM->RM token so that containers cannot access it.
Iterator<Token<?>> iter = credentials.getAllTokens().iterator();
LOG.info("Executing with tokens:");
while (iter.hasNext()) {
Token<?> token = iter.next();
LOG.info(token);
if (token.getKind().equals(AMRMTokenIdentifier.KIND_NAME)) {
iter.remove();
}
}
allTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength());
// Create appSubmitterUgi and add original tokens to it
String appSubmitterUserName = System.getenv(ApplicationConstants.Environment.USER.name());
appSubmitterUgi = UserGroupInformation.createRemoteUser(appSubmitterUserName);
appSubmitterUgi.addCredentials(credentials);
AMRMClientAsync.CallbackHandler allocListener = new RMCallbackHandler();
amRMClient = AMRMClientAsync.createAMRMClientAsync(1000, allocListener);
amRMClient.init(conf);
amRMClient.start();
containerListener = createNMCallbackHandler();
nmClientAsync = new NMClientAsyncImpl(containerListener);
nmClientAsync.init(conf);
nmClientAsync.start();
// Setup local RPC Server to accept status requests directly from clients
// TODO need to setup a protocol for client to be able to communicate to
// the RPC server
// TODO use the rpc port info to register with the RM for the client to
// send requests to this app master
// Register self with ResourceManager
// This will start heartbeating to the RM
appMasterHostname = NetUtils.getHostname();
RegisterApplicationMasterResponse response =
amRMClient.registerApplicationMaster(
appMasterHostname, appMasterRpcPort, appMasterTrackingUrl);
// Dump out information about cluster capability as seen by the
// resource manager
int maxMem = response.getMaximumResourceCapability().getMemory();
LOG.info("Max mem capabililty of resources in this cluster " + maxMem);
int maxVCores = response.getMaximumResourceCapability().getVirtualCores();
LOG.info("Max vcores capabililty of resources in this cluster " + maxVCores);
// A resource ask cannot exceed the max.
if (containerMemory > maxMem) {
LOG.info(
"Container memory specified above max threshold of cluster."
+ " Using max value."
+ ", specified="
+ containerMemory
+ ", max="
+ maxMem);
containerMemory = maxMem;
}
if (containerVirtualCores > maxVCores) {
LOG.info(
"Container virtual cores specified above max threshold of cluster."
+ " Using max value."
+ ", specified="
+ containerVirtualCores
+ ", max="
+ maxVCores);
containerVirtualCores = maxVCores;
}
List<Container> previousAMRunningContainers = response.getContainersFromPreviousAttempts();
LOG.info(
"Received "
+ previousAMRunningContainers.size()
+ " previous AM's running containers on AM registration.");
numAllocatedContainers.addAndGet(previousAMRunningContainers.size());
int numTotalContainersToRequest = numTotalContainers - previousAMRunningContainers.size();
// Setup ask for containers from RM
// Send request for containers to RM
// Until we get our fully allocated quota, we keep on polling RM for
// containers
// Keep looping until all the containers are launched and shell script
// executed on them ( regardless of success/failure).
for (int i = 0; i < numTotalContainersToRequest; ++i) {
ContainerRequest containerAsk = setupContainerAskForRM();
amRMClient.addContainerRequest(containerAsk);
}
numRequestedContainers.set(numTotalContainersToRequest);
try {
publishApplicationAttemptEvent(
timelineClient, appAttemptID.toString(), DSEvent.DS_APP_ATTEMPT_END);
} catch (Exception e) {
LOG.error("App Attempt start event coud not be pulished for " + appAttemptID.toString(), e);
}
}
@Override
protected TaskRunner2Result callInternal() throws Exception {
isStarted.set(true);
this.startTime = System.currentTimeMillis();
this.threadName = Thread.currentThread().getName();
if (LOG.isDebugEnabled()) {
LOG.debug("canFinish: " + taskSpec.getTaskAttemptID() + ": " + canFinish());
}
// Unregister from the AMReporter, since the task is now running.
this.amReporter.unregisterTask(request.getAmHost(), request.getAmPort());
synchronized (this) {
if (!shouldRunTask) {
LOG.info("Not starting task {} since it was killed earlier", taskSpec.getTaskAttemptID());
return new TaskRunner2Result(EndReason.KILL_REQUESTED, null, false);
}
}
// TODO This executor seems unnecessary. Here and TezChild
ExecutorService executorReal =
Executors.newFixedThreadPool(
1, new ThreadFactoryBuilder().setDaemon(true).setNameFormat("TezTaskRunner").build());
executor = MoreExecutors.listeningDecorator(executorReal);
// TODO Consolidate this code with TezChild.
runtimeWatch.start();
UserGroupInformation taskUgi = UserGroupInformation.createRemoteUser(request.getUser());
taskUgi.addCredentials(credentials);
Map<String, ByteBuffer> serviceConsumerMetadata = new HashMap<>();
serviceConsumerMetadata.put(
TezConstants.TEZ_SHUFFLE_HANDLER_SERVICE_ID,
TezCommonUtils.convertJobTokenToBytes(jobToken));
Multimap<String, String> startedInputsMap = createStartedInputMap(request.getFragmentSpec());
UserGroupInformation taskOwner =
UserGroupInformation.createRemoteUser(request.getTokenIdentifier());
final InetSocketAddress address =
NetUtils.createSocketAddrForHost(request.getAmHost(), request.getAmPort());
SecurityUtil.setTokenService(jobToken, address);
taskOwner.addToken(jobToken);
umbilical =
taskOwner.doAs(
new PrivilegedExceptionAction<LlapTaskUmbilicalProtocol>() {
@Override
public LlapTaskUmbilicalProtocol run() throws Exception {
return RPC.getProxy(
LlapTaskUmbilicalProtocol.class,
LlapTaskUmbilicalProtocol.versionID,
address,
conf);
}
});
taskReporter =
new LlapTaskReporter(
umbilical,
confParams.amHeartbeatIntervalMsMax,
confParams.amCounterHeartbeatInterval,
confParams.amMaxEventsPerHeartbeat,
new AtomicLong(0),
request.getContainerIdString());
String attemptId = fragmentInfo.getFragmentIdentifierString();
IOContextMap.setThreadAttemptId(attemptId);
try {
synchronized (this) {
if (shouldRunTask) {
taskRunner =
new TezTaskRunner2(
conf,
taskUgi,
fragmentInfo.getLocalDirs(),
taskSpec,
request.getAppAttemptNumber(),
serviceConsumerMetadata,
envMap,
startedInputsMap,
taskReporter,
executor,
objectRegistry,
pid,
executionContext,
memoryAvailable,
false);
}
}
if (taskRunner == null) {
LOG.info("Not starting task {} since it was killed earlier", taskSpec.getTaskAttemptID());
return new TaskRunner2Result(EndReason.KILL_REQUESTED, null, false);
}
try {
TaskRunner2Result result = taskRunner.run();
if (result.isContainerShutdownRequested()) {
LOG.warn("Unexpected container shutdown requested while running task. Ignoring");
}
isCompleted.set(true);
return result;
} finally {
FileSystem.closeAllForUGI(taskUgi);
LOG.info(
"ExecutionTime for Container: "
+ request.getContainerIdString()
+ "="
+ runtimeWatch.stop().elapsedMillis());
if (LOG.isDebugEnabled()) {
LOG.debug(
"canFinish post completion: " + taskSpec.getTaskAttemptID() + ": " + canFinish());
}
}
} finally {
IOContextMap.clearThreadAttempt(attemptId);
}
}
