Google Git
Sign in
apache / hadoop-mapreduce / bc2d1760a64c5cf642110f4b60ac0a2875e99adf / . / src / java / org / apache / hadoop / mapred / JobInProgress.java
blob: 429d2b015356ba0ac1dc383662f30cb3f91142d4 [file] [log] [blame]
/**
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.hadoop.mapred;
import java.io.File;
import java.io.FileOutputStream;
import java.io.IOException;
import java.security.PrivilegedExceptionAction;
import java.util.ArrayList;
import java.util.Collection;
import java.util.Comparator;
import java.util.EnumMap;
import java.util.HashMap;
import java.util.HashSet;
import java.util.IdentityHashMap;
import java.util.Iterator;
import java.util.LinkedHashSet;
import java.util.LinkedList;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.TreeMap;
import java.util.Vector;
import java.util.concurrent.atomic.AtomicBoolean;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.classification.InterfaceAudience;
import org.apache.hadoop.classification.InterfaceStability;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.LocalFileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapred.CleanupQueue.PathDeletionContext;
import org.apache.hadoop.mapreduce.JobContext;
import org.apache.hadoop.mapreduce.JobCounter;
import org.apache.hadoop.mapreduce.JobSubmissionFiles;
import org.apache.hadoop.mapreduce.MRJobConfig;
import org.apache.hadoop.mapreduce.TaskType;
import org.apache.hadoop.mapreduce.jobhistory.JobFinishedEvent;
import org.apache.hadoop.mapreduce.jobhistory.JobHistory;
import org.apache.hadoop.mapreduce.jobhistory.JobInfoChangeEvent;
import org.apache.hadoop.mapreduce.jobhistory.JobInitedEvent;
import org.apache.hadoop.mapreduce.jobhistory.JobPriorityChangeEvent;
import org.apache.hadoop.mapreduce.jobhistory.JobStatusChangedEvent;
import org.apache.hadoop.mapreduce.jobhistory.JobSubmittedEvent;
import org.apache.hadoop.mapreduce.jobhistory.JobUnsuccessfulCompletionEvent;
import org.apache.hadoop.mapreduce.jobhistory.MapAttemptFinishedEvent;
import org.apache.hadoop.mapreduce.jobhistory.ReduceAttemptFinishedEvent;
import org.apache.hadoop.mapreduce.jobhistory.TaskAttemptStartedEvent;
import org.apache.hadoop.mapreduce.jobhistory.TaskAttemptUnsuccessfulCompletionEvent;
import org.apache.hadoop.mapreduce.jobhistory.TaskFailedEvent;
import org.apache.hadoop.mapreduce.jobhistory.TaskFinishedEvent;
import org.apache.hadoop.mapreduce.jobhistory.TaskStartedEvent;
import org.apache.hadoop.mapreduce.security.TokenCache;
import org.apache.hadoop.security.Credentials;
import org.apache.hadoop.mapreduce.security.token.DelegationTokenRenewal;
import org.apache.hadoop.mapreduce.security.token.JobTokenIdentifier;
import org.apache.hadoop.mapreduce.server.jobtracker.TaskTracker;
import org.apache.hadoop.mapreduce.split.JobSplit;
import org.apache.hadoop.mapreduce.split.SplitMetaInfoReader;
import org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo;
import org.apache.hadoop.mapreduce.task.JobContextImpl;
import org.apache.hadoop.net.NetUtils;
import org.apache.hadoop.net.NetworkTopology;
import org.apache.hadoop.net.Node;
import org.apache.hadoop.security.UserGroupInformation;
import org.apache.hadoop.security.token.Token;
import org.apache.hadoop.security.token.TokenIdentifier;
import org.apache.hadoop.util.StringUtils;
/**
* JobInProgress maintains all the info for keeping a Job on the straight and
* narrow. It keeps its JobProfile and its latest JobStatus, plus a set of
* tables for doing bookkeeping of its Tasks.
*/
@InterfaceAudience.LimitedPrivate({"MapReduce"})
@InterfaceStability.Unstable
public class JobInProgress {
/**
* Used when the a kill is issued to a job which is initializing.
*/
static class KillInterruptedException extends InterruptedException {
private static final long serialVersionUID = 1L;
public KillInterruptedException(String msg) {
super(msg);
}
}
static final Log LOG = LogFactory.getLog(JobInProgress.class);
JobProfile profile;
JobStatus status;
Path jobFile = null;
Path localJobFile = null;
TaskInProgress maps[] = new TaskInProgress[0];
TaskInProgress reduces[] = new TaskInProgress[0];
TaskInProgress cleanup[] = new TaskInProgress[0];
TaskInProgress setup[] = new TaskInProgress[0];
int numMapTasks = 0;
int numReduceTasks = 0;
final long memoryPerMap;
final long memoryPerReduce;
volatile int numSlotsPerMap = 1;
volatile int numSlotsPerReduce = 1;
final int maxTaskFailuresPerTracker;
// Counters to track currently running/finished/failed Map/Reduce task-attempts
int runningMapTasks = 0;
int runningReduceTasks = 0;
int finishedMapTasks = 0;
int finishedReduceTasks = 0;
int failedMapTasks = 0;
int failedReduceTasks = 0;
static final float DEFAULT_COMPLETED_MAPS_PERCENT_FOR_REDUCE_SLOWSTART = 0.05f;
int completedMapsForReduceSlowstart = 0;
// runningMapTasks include speculative tasks, so we need to capture
// speculative tasks separately
int speculativeMapTasks = 0;
int speculativeReduceTasks = 0;
int mapFailuresPercent = 0;
int reduceFailuresPercent = 0;
int failedMapTIPs = 0;
int failedReduceTIPs = 0;
private volatile boolean launchedCleanup = false;
private volatile boolean launchedSetup = false;
private volatile boolean jobKilled = false;
private volatile boolean jobFailed = false;
private final boolean jobSetupCleanupNeeded;
private final boolean taskCleanupNeeded;
JobPriority priority = JobPriority.NORMAL;
protected JobTracker jobtracker;
protected Credentials tokenStorage;
JobHistory jobHistory;
// NetworkTopology Node to the set of TIPs
Map<Node, List<TaskInProgress>> nonRunningMapCache;
// Map of NetworkTopology Node to set of running TIPs
Map<Node, Set<TaskInProgress>> runningMapCache;
// A list of non-local non-running maps
List<TaskInProgress> nonLocalMaps;
// A set of non-local running maps
Set<TaskInProgress> nonLocalRunningMaps;
// A list of non-running reduce TIPs
List<TaskInProgress> nonRunningReduces;
// A set of running reduce TIPs
Set<TaskInProgress> runningReduces;
// A list of cleanup tasks for the map task attempts, to be launched
List<TaskAttemptID> mapCleanupTasks = new LinkedList<TaskAttemptID>();
// A list of cleanup tasks for the reduce task attempts, to be launched
List<TaskAttemptID> reduceCleanupTasks = new LinkedList<TaskAttemptID>();
int maxLevel;
/**
* A special value indicating that
* {@link #findNewMapTask(TaskTrackerStatus, int, int, int, double)} should
* schedule any available map tasks for this job, including speculative tasks.
*/
int anyCacheLevel;
/**
* A special value indicating that
* {@link #findNewMapTask(TaskTrackerStatus, int, int, int, double)} should
* schedule any only off-switch and speculative map tasks for this job.
*/
private static final int NON_LOCAL_CACHE_LEVEL = -1;
private int taskCompletionEventTracker = 0;
List<TaskCompletionEvent> taskCompletionEvents;
// The maximum percentage of trackers in cluster added to the 'blacklist'.
private static final double CLUSTER_BLACKLIST_PERCENT = 0.25;
// The maximum percentage of fetch failures allowed for a map
private static final double MAX_ALLOWED_FETCH_FAILURES_PERCENT = 0.5;
// No. of tasktrackers in the cluster
private volatile int clusterSize = 0;
// The no. of tasktrackers where >= conf.getMaxTaskFailuresPerTracker()
// tasks have failed
private volatile int flakyTaskTrackers = 0;
// Map of trackerHostName -> no. of task failures
private Map<String, Integer> trackerToFailuresMap =
new TreeMap<String, Integer>();
//Confine estimation algorithms to an "oracle" class that JIP queries.
ResourceEstimator resourceEstimator;
long startTime;
long launchTime;
long finishTime;
// First *task launch times
final Map<TaskType, Long> firstTaskLaunchTimes =
new EnumMap<TaskType, Long>(TaskType.class);
// Indicates how many times the job got restarted
private final int restartCount;
JobConf conf;
protected AtomicBoolean tasksInited = new AtomicBoolean(false);
private JobInitKillStatus jobInitKillStatus = new JobInitKillStatus();
LocalFileSystem localFs;
FileSystem fs;
String user;
JobID jobId;
volatile private boolean hasSpeculativeMaps;
volatile private boolean hasSpeculativeReduces;
long inputLength = 0;
Counters jobCounters = new Counters();
// Maximum no. of fetch-failure notifications after which map task is killed
private static final int MAX_FETCH_FAILURES_NOTIFICATIONS = 3;
// Don't lower speculativeCap below one TT's worth (for small clusters)
private static final int MIN_SPEC_CAP = 10;
private static final float MIN_SLOTS_CAP = 0.01f;
// Map of mapTaskId -> no. of fetch failures
private Map<TaskAttemptID, Integer> mapTaskIdToFetchFailuresMap =
new TreeMap<TaskAttemptID, Integer>();
private Object schedulingInfo;
private String submitHostName;
private String submitHostAddress;
//thresholds for speculative execution
float slowTaskThreshold;
float speculativeCap;
float slowNodeThreshold; //standard deviations
//Statistics are maintained for a couple of things
//mapTaskStats is used for maintaining statistics about
//the completion time of map tasks on the trackers. On a per
//tracker basis, the mean time for task completion is maintained
private DataStatistics mapTaskStats = new DataStatistics();
//reduceTaskStats is used for maintaining statistics about
//the completion time of reduce tasks on the trackers. On a per
//tracker basis, the mean time for task completion is maintained
private DataStatistics reduceTaskStats = new DataStatistics();
//trackerMapStats used to maintain a mapping from the tracker to the
//the statistics about completion time of map tasks
private Map<String,DataStatistics> trackerMapStats =
new HashMap<String,DataStatistics>();
//trackerReduceStats used to maintain a mapping from the tracker to the
//the statistics about completion time of reduce tasks
private Map<String,DataStatistics> trackerReduceStats =
new HashMap<String,DataStatistics>();
//runningMapStats used to maintain the RUNNING map tasks' statistics
private DataStatistics runningMapTaskStats = new DataStatistics();
//runningReduceStats used to maintain the RUNNING reduce tasks' statistics
private DataStatistics runningReduceTaskStats = new DataStatistics();
private static class FallowSlotInfo {
long timestamp;
int numSlots;
public FallowSlotInfo(long timestamp, int numSlots) {
this.timestamp = timestamp;
this.numSlots = numSlots;
}
public long getTimestamp() {
return timestamp;
}
public void setTimestamp(long timestamp) {
this.timestamp = timestamp;
}
public int getNumSlots() {
return numSlots;
}
public void setNumSlots(int numSlots) {
this.numSlots = numSlots;
}
}
private Map<TaskTracker, FallowSlotInfo> trackersReservedForMaps =
new HashMap<TaskTracker, FallowSlotInfo>();
private Map<TaskTracker, FallowSlotInfo> trackersReservedForReduces =
new HashMap<TaskTracker, FallowSlotInfo>();
private Path jobSubmitDir = null;
/**
* Create an almost empty JobInProgress, which can be used only for tests
*/
protected JobInProgress(JobID jobid, JobConf conf, JobTracker tracker) {
this.conf = conf;
this.jobId = jobid;
this.numMapTasks = conf.getNumMapTasks();
this.numReduceTasks = conf.getNumReduceTasks();
this.maxLevel = NetworkTopology.DEFAULT_HOST_LEVEL;
this.anyCacheLevel = this.maxLevel+1;
this.jobtracker = tracker;
this.restartCount = 0;
this.profile = new JobProfile(conf.getUser(), jobid, "", "",
conf.getJobName(),conf.getQueueName());
this.memoryPerMap = conf.getMemoryForMapTask();
this.memoryPerReduce = conf.getMemoryForReduceTask();
this.maxTaskFailuresPerTracker = conf.getMaxTaskFailuresPerTracker();
hasSpeculativeMaps = conf.getMapSpeculativeExecution();
hasSpeculativeReduces = conf.getReduceSpeculativeExecution();
this.nonLocalMaps = new LinkedList<TaskInProgress>();
this.nonLocalRunningMaps = new LinkedHashSet<TaskInProgress>();
this.runningMapCache = new IdentityHashMap<Node, Set<TaskInProgress>>();
this.nonRunningReduces = new LinkedList<TaskInProgress>();
this.runningReduces = new LinkedHashSet<TaskInProgress>();
this.resourceEstimator = new ResourceEstimator(this);
this.status = new JobStatus(jobid, 0.0f, 0.0f, JobStatus.PREP,
this.profile.getUser(), this.profile.getJobName(),
this.profile.getJobFile(), "");
this.jobtracker.getInstrumentation().addPrepJob(conf, jobid);
this.taskCompletionEvents = new ArrayList<TaskCompletionEvent>
(numMapTasks + numReduceTasks + 10);
this.slowTaskThreshold = Math.max(0.0f,
conf.getFloat(MRJobConfig.SPECULATIVE_SLOWTASK_THRESHOLD,1.0f));
this.speculativeCap = conf.getFloat(
MRJobConfig.SPECULATIVECAP,0.1f);
this.slowNodeThreshold = conf.getFloat(
MRJobConfig.SPECULATIVE_SLOWNODE_THRESHOLD,1.0f);
this.jobSetupCleanupNeeded = conf.getBoolean(
MRJobConfig.SETUP_CLEANUP_NEEDED, true);
this.taskCleanupNeeded = conf.getBoolean(
MRJobConfig.TASK_CLEANUP_NEEDED, true);
if (tracker != null) { // Some mock tests have null tracker
this.jobHistory = tracker.getJobHistory();
}
this.tokenStorage = null;
}
JobInProgress(JobConf conf) {
restartCount = 0;
jobSetupCleanupNeeded = false;
taskCleanupNeeded = true;
this.memoryPerMap = conf.getMemoryForMapTask();
this.memoryPerReduce = conf.getMemoryForReduceTask();
this.maxTaskFailuresPerTracker = conf.getMaxTaskFailuresPerTracker();
}
/**
* Create a JobInProgress with the given job file, plus a handle
* to the tracker.
*/
public JobInProgress(JobTracker jobtracker,
final JobConf default_conf, int rCount,
JobInfo jobInfo,
Credentials ts
) throws IOException, InterruptedException {
try {
this.restartCount = rCount;
this.jobId = JobID.downgrade(jobInfo.getJobID());
String url = "http://" + jobtracker.getJobTrackerMachine() + ":"
+ jobtracker.getInfoPort() + "/jobdetails.jsp?jobid=" + this.jobId;
this.jobtracker = jobtracker;
this.jobHistory = jobtracker.getJobHistory();
this.startTime = System.currentTimeMillis();
this.localFs = jobtracker.getLocalFileSystem();
this.tokenStorage = ts;
// use the user supplied token to add user credentials to the conf
jobSubmitDir = jobInfo.getJobSubmitDir();
user = jobInfo.getUser().toString();
UserGroupInformation ugi = UserGroupInformation.createRemoteUser(user);
if (ts != null) {
for (Token<? extends TokenIdentifier> token : ts.getAllTokens()) {
ugi.addToken(token);
}
}
fs = ugi.doAs(new PrivilegedExceptionAction<FileSystem>() {
public FileSystem run() throws IOException {
return jobSubmitDir.getFileSystem(default_conf);
}
});
this.localJobFile = default_conf.getLocalPath(JobTracker.SUBDIR + "/"
+ this.jobId + ".xml");
jobFile = JobSubmissionFiles.getJobConfPath(jobSubmitDir);
fs.copyToLocalFile(jobFile, localJobFile);
conf = new JobConf(localJobFile);
if (conf.getUser() == null) {
this.conf.setUser(user);
}
if (!conf.getUser().equals(user)) {
String desc = "The username " + conf.getUser() + " obtained from the "
+ "conf doesn't match the username " + user + " the user "
+ "authenticated as";
AuditLogger.logFailure(user, Operation.SUBMIT_JOB.name(),
conf.getUser(), jobId.toString(), desc);
throw new IOException(desc);
}
String userGroups[] = ugi.getGroupNames();
String primaryGroup = (userGroups.length > 0) ? userGroups[0] : null;
if (primaryGroup != null) {
conf.set("group.name", primaryGroup);
}
this.priority = conf.getJobPriority();
this.profile = new JobProfile(conf.getUser(), this.jobId, jobFile
.toString(), url, conf.getJobName(), conf.getQueueName());
this.status = new JobStatus(this.jobId, 0.0f, 0.0f, JobStatus.PREP,
profile.getUser(), profile.getJobName(), profile.getJobFile(),
profile.getURL().toString());
this.jobtracker.getInstrumentation().addPrepJob(conf, this.jobId);
status.setStartTime(startTime);
this.status.setJobPriority(this.priority);
this.numMapTasks = conf.getNumMapTasks();
this.numReduceTasks = conf.getNumReduceTasks();
this.memoryPerMap = conf.getMemoryForMapTask();
this.memoryPerReduce = conf.getMemoryForReduceTask();
this.taskCompletionEvents = new ArrayList<TaskCompletionEvent>(
numMapTasks + numReduceTasks + 10);
JobContext jobContext = new JobContextImpl(conf, jobId);
this.jobSetupCleanupNeeded = jobContext.getJobSetupCleanupNeeded();
this.taskCleanupNeeded = jobContext.getTaskCleanupNeeded();
// Construct the jobACLs
status.setJobACLs(jobtracker.getJobACLsManager().constructJobACLs(conf));
this.mapFailuresPercent = conf.getMaxMapTaskFailuresPercent();
this.reduceFailuresPercent = conf.getMaxReduceTaskFailuresPercent();
this.maxTaskFailuresPerTracker = conf.getMaxTaskFailuresPerTracker();
hasSpeculativeMaps = conf.getMapSpeculativeExecution();
hasSpeculativeReduces = conf.getReduceSpeculativeExecution();
this.maxLevel = jobtracker.getNumTaskCacheLevels();
this.anyCacheLevel = this.maxLevel + 1;
this.nonLocalMaps = new LinkedList<TaskInProgress>();
this.nonLocalRunningMaps = new LinkedHashSet<TaskInProgress>();
this.runningMapCache = new IdentityHashMap<Node, Set<TaskInProgress>>();
this.nonRunningReduces = new LinkedList<TaskInProgress>();
this.runningReduces = new LinkedHashSet<TaskInProgress>();
this.resourceEstimator = new ResourceEstimator(this);
this.submitHostName = conf.getJobSubmitHostName();
this.submitHostAddress = conf.getJobSubmitHostAddress();
this.slowTaskThreshold = Math.max(0.0f, conf.getFloat(
MRJobConfig.SPECULATIVE_SLOWTASK_THRESHOLD, 1.0f));
this.speculativeCap = conf.getFloat(MRJobConfig.SPECULATIVECAP, 0.1f);
this.slowNodeThreshold = conf.getFloat(
MRJobConfig.SPECULATIVE_SLOWNODE_THRESHOLD, 1.0f);
// register job's tokens for renewal
DelegationTokenRenewal.registerDelegationTokensForRenewal(jobInfo
.getJobID(), ts, jobtracker.getConf());
} finally {
// close all FileSystems that was created above for the current user
// At this point, this constructor is called in the context of an RPC, and
// hence the "current user" is actually referring to the kerberos
// authenticated user (if security is ON).
FileSystem.closeAllForUGI(UserGroupInformation.getCurrentUser());
}
}
private void printCache (Map<Node, List<TaskInProgress>> cache) {
LOG.info("The taskcache info:");
for (Map.Entry<Node, List<TaskInProgress>> n : cache.entrySet()) {
List <TaskInProgress> tips = n.getValue();
LOG.info("Cached TIPs on node: " + n.getKey());
for (TaskInProgress tip : tips) {
LOG.info("tip : " + tip.getTIPId());
}
}
}
Map<Node, List<TaskInProgress>> createCache(
TaskSplitMetaInfo[] splits, int maxLevel) {
Map<Node, List<TaskInProgress>> cache =
new IdentityHashMap<Node, List<TaskInProgress>>(maxLevel);
for (int i = 0; i < splits.length; i++) {
String[] splitLocations = splits[i].getLocations();
if (splitLocations.length == 0) {
nonLocalMaps.add(maps[i]);
continue;
}
for(String host: splitLocations) {
Node node = jobtracker.resolveAndAddToTopology(host);
LOG.info("tip:" + maps[i].getTIPId() + " has split on node:" + node);
for (int j = 0; j < maxLevel; j++) {
List<TaskInProgress> hostMaps = cache.get(node);
if (hostMaps == null) {
hostMaps = new ArrayList<TaskInProgress>();
cache.put(node, hostMaps);
hostMaps.add(maps[i]);
}
//check whether the hostMaps already contains an entry for a TIP
//This will be true for nodes that are racks and multiple nodes in
//the rack contain the input for a tip. Note that if it already
//exists in the hostMaps, it must be the last element there since
//we process one TIP at a time sequentially in the split-size order
if (hostMaps.get(hostMaps.size() - 1) != maps[i]) {
hostMaps.add(maps[i]);
}
node = node.getParent();
}
}
}
return cache;
}
/**
* Check if the job has been initialized.
* @return <code>true</code> if the job has been initialized,
* <code>false</code> otherwise
*/
public boolean inited() {
return tasksInited.get();
}
/**
* Get the user for the job
*/
public String getUser() {
return user;
}
boolean getMapSpeculativeExecution() {
return hasSpeculativeMaps;
}
boolean getReduceSpeculativeExecution() {
return hasSpeculativeReduces;
}
long getMemoryForMapTask() {
return memoryPerMap;
}
long getMemoryForReduceTask() {
return memoryPerReduce;
}
/**
* Get the number of slots required to run a single map task-attempt.
* @return the number of slots required to run a single map task-attempt
*/
int getNumSlotsPerMap() {
return numSlotsPerMap;
}
/**
* Set the number of slots required to run a single map task-attempt.
* This is typically set by schedulers which support high-ram jobs.
* @param slots the number of slots required to run a single map task-attempt
*/
void setNumSlotsPerMap(int numSlotsPerMap) {
this.numSlotsPerMap = numSlotsPerMap;
}
/**
* Get the number of slots required to run a single reduce task-attempt.
* @return the number of slots required to run a single reduce task-attempt
*/
int getNumSlotsPerReduce() {
return numSlotsPerReduce;
}
/**
* Set the number of slots required to run a single reduce task-attempt.
* This is typically set by schedulers which support high-ram jobs.
* @param slots the number of slots required to run a single reduce
* task-attempt
*/
void setNumSlotsPerReduce(int numSlotsPerReduce) {
this.numSlotsPerReduce = numSlotsPerReduce;
}
/**
* Construct the splits, etc. This is invoked from an async
* thread so that split-computation doesn't block anyone. Only the
* {@link JobTracker} should invoke this api. Look
* at {@link JobTracker#initJob(JobInProgress)} for more details.
*/
public synchronized void initTasks()
throws IOException, KillInterruptedException {
if (tasksInited.get() || isComplete()) {
return;
}
synchronized(jobInitKillStatus){
if(jobInitKillStatus.killed || jobInitKillStatus.initStarted) {
return;
}
jobInitKillStatus.initStarted = true;
}
LOG.info("Initializing " + jobId);
logSubmissionToJobHistory();
// log the job priority
setPriority(this.priority);
//
// generate security keys needed by Tasks
//
generateAndStoreTokens();
//
// read input splits and create a map per a split
//
TaskSplitMetaInfo[] taskSplitMetaInfo = createSplits(jobId);
numMapTasks = taskSplitMetaInfo.length;
checkTaskLimits();
jobtracker.getInstrumentation().addWaitingMaps(getJobID(), numMapTasks);
jobtracker.getInstrumentation().addWaitingReduces(getJobID(), numReduceTasks);
createMapTasks(jobFile.toString(), taskSplitMetaInfo);
if (numMapTasks > 0) {
nonRunningMapCache = createCache(taskSplitMetaInfo,
maxLevel);
}
// set the launch time
this.launchTime = JobTracker.getClock().getTime();
createReduceTasks(jobFile.toString());
// Calculate the minimum number of maps to be complete before
// we should start scheduling reduces
completedMapsForReduceSlowstart =
(int)Math.ceil(
(conf.getFloat(MRJobConfig.COMPLETED_MAPS_FOR_REDUCE_SLOWSTART,
DEFAULT_COMPLETED_MAPS_PERCENT_FOR_REDUCE_SLOWSTART) *
numMapTasks));
initSetupCleanupTasks(jobFile.toString());
synchronized(jobInitKillStatus){
jobInitKillStatus.initDone = true;
if(jobInitKillStatus.killed) {
//setup not launched so directly terminate
throw new KillInterruptedException("Job " + jobId + " killed in init");
}
}
tasksInited.set(true);
JobInitedEvent jie = new JobInitedEvent(
profile.getJobID(), this.launchTime,
numMapTasks, numReduceTasks,
JobStatus.getJobRunState(JobStatus.PREP));
jobHistory.logEvent(jie, jobId);
// Log the number of map and reduce tasks
LOG.info("Job " + jobId + " initialized successfully with " + numMapTasks
+ " map tasks and " + numReduceTasks + " reduce tasks.");
}
// Returns true if the job is empty (0 maps, 0 reduces and no setup-cleanup)
// else return false.
synchronized boolean isJobEmpty() {
return maps.length == 0 && reduces.length == 0 && !jobSetupCleanupNeeded;
}
synchronized boolean isSetupCleanupRequired() {
return jobSetupCleanupNeeded;
}
// Should be called once the init is done. This will complete the job
// because the job is empty (0 maps, 0 reduces and no setup-cleanup).
synchronized void completeEmptyJob() {
jobComplete();
}
synchronized void completeSetup() {
setupComplete();
}
void logSubmissionToJobHistory() throws IOException {
// log job info
String username = conf.getUser();
if (username == null) { username = ""; }
String jobname = conf.getJobName();
String jobQueueName = conf.getQueueName();
setUpLocalizedJobConf(conf, jobId);
jobHistory.setupEventWriter(jobId, conf);
JobSubmittedEvent jse =
new JobSubmittedEvent(jobId, jobname, username, this.startTime,
jobFile.toString(), status.getJobACLs(), jobQueueName);
jobHistory.logEvent(jse, jobId);
}
TaskSplitMetaInfo[] createSplits(org.apache.hadoop.mapreduce.JobID jobId)
throws IOException {
TaskSplitMetaInfo[] allTaskSplitMetaInfo =
SplitMetaInfoReader.readSplitMetaInfo(jobId, fs, conf, jobSubmitDir);
return allTaskSplitMetaInfo;
}
/**
* If the number of taks is greater than the configured value
* throw an exception that will fail job initialization
*/
void checkTaskLimits() throws IOException {
int maxTasks = jobtracker.getMaxTasksPerJob();
if (maxTasks > 0 && numMapTasks + numReduceTasks > maxTasks) {
throw new IOException(
"The number of tasks for this job " +
(numMapTasks + numReduceTasks) +
" exceeds the configured limit " + maxTasks);
}
}
synchronized void createMapTasks(String jobFile,
TaskSplitMetaInfo[] splits) {
maps = new TaskInProgress[numMapTasks];
for(int i=0; i < numMapTasks; ++i) {
inputLength += splits[i].getInputDataLength();
maps[i] = new TaskInProgress(jobId, jobFile,
splits[i],
jobtracker, conf, this,
i, numSlotsPerMap);
}
LOG.info("Input size for job " + jobId + " = " + inputLength
+ ". Number of splits = " + splits.length);
}
synchronized void createReduceTasks(String jobFile) {
this.reduces = new TaskInProgress[numReduceTasks];
for (int i = 0; i < numReduceTasks; i++) {
reduces[i] = new TaskInProgress(jobId, jobFile,
numMapTasks, i,
jobtracker, conf,
this, numSlotsPerReduce);
nonRunningReduces.add(reduces[i]);
}
}
synchronized void initSetupCleanupTasks(String jobFile) {
if (!jobSetupCleanupNeeded) {
LOG.info("Setup/Cleanup not needed for job " + jobId);
// nothing to initialize
return;
}
// create cleanup two cleanup tips, one map and one reduce.
cleanup = new TaskInProgress[2];
// cleanup map tip. This map doesn't use any splits. Just assign an empty
// split.
TaskSplitMetaInfo emptySplit = JobSplit.EMPTY_TASK_SPLIT;
cleanup[0] = new TaskInProgress(jobId, jobFile, emptySplit,
jobtracker, conf, this, numMapTasks, 1);
cleanup[0].setJobCleanupTask();
// cleanup reduce tip.
cleanup[1] = new TaskInProgress(jobId, jobFile, numMapTasks,
numReduceTasks, jobtracker, conf, this, 1);
cleanup[1].setJobCleanupTask();
// create two setup tips, one map and one reduce.
setup = new TaskInProgress[2];
// setup map tip. This map doesn't use any split. Just assign an empty
// split.
setup[0] = new TaskInProgress(jobId, jobFile, emptySplit,
jobtracker, conf, this, numMapTasks + 1, 1);
setup[0].setJobSetupTask();
// setup reduce tip.
setup[1] = new TaskInProgress(jobId, jobFile, numMapTasks,
numReduceTasks + 1, jobtracker, conf, this, 1);
setup[1].setJobSetupTask();
}
void setupComplete() {
status.setSetupProgress(1.0f);
if (this.status.getRunState() == JobStatus.PREP) {
changeStateTo(JobStatus.RUNNING);
JobStatusChangedEvent jse =
new JobStatusChangedEvent(profile.getJobID(),
JobStatus.getJobRunState(JobStatus.RUNNING));
jobHistory.logEvent(jse, profile.getJobID());
}
}
/////////////////////////////////////////////////////
// Accessors for the JobInProgress
/////////////////////////////////////////////////////
public JobProfile getProfile() {
return profile;
}
public JobStatus getStatus() {
return status;
}
public synchronized long getLaunchTime() {
return launchTime;
}
Map<TaskType, Long> getFirstTaskLaunchTimes() {
return firstTaskLaunchTimes;
}
public long getStartTime() {
return startTime;
}
public long getFinishTime() {
return finishTime;
}
public int desiredMaps() {
return numMapTasks;
}
public synchronized int finishedMaps() {
return finishedMapTasks;
}
public int desiredReduces() {
return numReduceTasks;
}
public synchronized int runningMaps() {
return runningMapTasks;
}
public synchronized int runningReduces() {
return runningReduceTasks;
}
public synchronized int finishedReduces() {
return finishedReduceTasks;
}
public synchronized int pendingMaps() {
return numMapTasks - runningMapTasks - failedMapTIPs -
finishedMapTasks + speculativeMapTasks;
}
public synchronized int pendingReduces() {
return numReduceTasks - runningReduceTasks - failedReduceTIPs -
finishedReduceTasks + speculativeReduceTasks;
}
public int getNumSlotsPerTask(TaskType taskType) {
if (taskType == TaskType.MAP) {
return numSlotsPerMap;
} else if (taskType == TaskType.REDUCE) {
return numSlotsPerReduce;
} else {
return 1;
}
}
public JobPriority getPriority() {
return this.priority;
}
public void setPriority(JobPriority priority) {
if(priority == null) {
priority = JobPriority.NORMAL;
}
synchronized (this) {
this.priority = priority;
status.setJobPriority(priority);
// log and change to the job's priority
JobPriorityChangeEvent prEvent =
new JobPriorityChangeEvent(jobId, priority);
jobHistory.logEvent(prEvent, jobId);
}
}
// Update the job start/launch time (upon restart) and log to history
synchronized void updateJobInfo(long startTime, long launchTime) {
// log and change to the job's start/launch time
this.startTime = startTime;
this.launchTime = launchTime;
JobInfoChangeEvent event =
new JobInfoChangeEvent(jobId, startTime, launchTime);
jobHistory.logEvent(event, jobId);
}
/**
* Get the number of times the job has restarted
*/
int getNumRestarts() {
return restartCount;
}
long getInputLength() {
return inputLength;
}
boolean isCleanupLaunched() {
return launchedCleanup;
}
boolean isSetupLaunched() {
return launchedSetup;
}
/**
* Get all the tasks of the desired type in this job.
* @param type {@link TaskType} of the tasks required
* @return An array of {@link TaskInProgress} matching the given type.
* Returns an empty array if no tasks are found for the given type.
*/
TaskInProgress[] getTasks(TaskType type) {
TaskInProgress[] tasks = null;
switch (type) {
case MAP:
{
tasks = maps;
}
break;
case REDUCE:
{
tasks = reduces;
}
break;
case JOB_SETUP:
{
tasks = setup;
}
break;
case JOB_CLEANUP:
{
tasks = cleanup;
}
break;
default:
{
tasks = new TaskInProgress[0];
}
break;
}
return tasks;
}
/**
* Return the nonLocalRunningMaps
* @return
*/
Set<TaskInProgress> getNonLocalRunningMaps()
{
return nonLocalRunningMaps;
}
/**
* Return the runningMapCache
* @return
*/
Map<Node, Set<TaskInProgress>> getRunningMapCache()
{
return runningMapCache;
}
/**
* Return runningReduces
* @return
*/
Set<TaskInProgress> getRunningReduces()
{
return runningReduces;
}
/**
* Get the job configuration
* @return the job's configuration
*/
JobConf getJobConf() {
return conf;
}
/**
* Return a vector of completed TaskInProgress objects
*/
public synchronized Vector<TaskInProgress> reportTasksInProgress(boolean shouldBeMap,
boolean shouldBeComplete) {
Vector<TaskInProgress> results = new Vector<TaskInProgress>();
TaskInProgress tips[] = null;
if (shouldBeMap) {
tips = maps;
} else {
tips = reduces;
}
for (int i = 0; i < tips.length; i++) {
if (tips[i].isComplete() == shouldBeComplete) {
results.add(tips[i]);
}
}
return results;
}
/**
* Return a vector of cleanup TaskInProgress objects
*/
public synchronized Vector<TaskInProgress> reportCleanupTIPs(
boolean shouldBeComplete) {
Vector<TaskInProgress> results = new Vector<TaskInProgress>();
for (int i = 0; i < cleanup.length; i++) {
if (cleanup[i].isComplete() == shouldBeComplete) {
results.add(cleanup[i]);
}
}
return results;
}
/**
* Return a vector of setup TaskInProgress objects
*/
public synchronized Vector<TaskInProgress> reportSetupTIPs(
boolean shouldBeComplete) {
Vector<TaskInProgress> results = new Vector<TaskInProgress>();
for (int i = 0; i < setup.length; i++) {
if (setup[i].isComplete() == shouldBeComplete) {
results.add(setup[i]);
}
}
return results;
}
////////////////////////////////////////////////////
// Status update methods
////////////////////////////////////////////////////
/**
* Assuming {@link JobTracker} is locked on entry.
*/
public synchronized void updateTaskStatus(TaskInProgress tip,
TaskStatus status) {
double oldProgress = tip.getProgress(); // save old progress
boolean wasRunning = tip.isRunning();
boolean wasComplete = tip.isComplete();
boolean wasPending = tip.isOnlyCommitPending();
TaskAttemptID taskid = status.getTaskID();
boolean wasAttemptRunning = tip.isAttemptRunning(taskid);
// If the TIP is already completed and the task reports as SUCCEEDED then
// mark the task as KILLED.
// In case of task with no promotion the task tracker will mark the task
// as SUCCEEDED.
// User has requested to kill the task, but TT reported SUCCEEDED,
// mark the task KILLED.
if ((wasComplete || tip.wasKilled(taskid)) &&
(status.getRunState() == TaskStatus.State.SUCCEEDED)) {
status.setRunState(TaskStatus.State.KILLED);
}
// If the job is complete or task-cleanup is switched off
// and a task has just reported its state as FAILED_UNCLEAN/KILLED_UNCLEAN,
// make the task's state FAILED/KILLED without launching cleanup attempt.
// Note that if task is already a cleanup attempt,
// we don't change the state to make sure the task gets a killTaskAction
if ((this.isComplete() || jobFailed || jobKilled || !taskCleanupNeeded) &&
!tip.isCleanupAttempt(taskid)) {
if (status.getRunState() == TaskStatus.State.FAILED_UNCLEAN) {
status.setRunState(TaskStatus.State.FAILED);
} else if (status.getRunState() == TaskStatus.State.KILLED_UNCLEAN) {
status.setRunState(TaskStatus.State.KILLED);
}
}
boolean change = tip.updateStatus(status);
if (change) {
TaskStatus.State state = status.getRunState();
// get the TaskTrackerStatus where the task ran
TaskTracker taskTracker =
this.jobtracker.getTaskTracker(tip.machineWhereTaskRan(taskid));
TaskTrackerStatus ttStatus =
(taskTracker == null) ? null : taskTracker.getStatus();
String taskTrackerHttpLocation = null;
if (null != ttStatus){
String host;
if (NetUtils.getStaticResolution(ttStatus.getHost()) != null) {
host = NetUtils.getStaticResolution(ttStatus.getHost());
} else {
host = ttStatus.getHost();
}
taskTrackerHttpLocation = "http://" + host + ":"
+ ttStatus.getHttpPort();
}
TaskCompletionEvent taskEvent = null;
if (state == TaskStatus.State.SUCCEEDED) {
taskEvent = new TaskCompletionEvent(
taskCompletionEventTracker,
taskid,
tip.idWithinJob(),
status.getIsMap() &&
!tip.isJobCleanupTask() &&
!tip.isJobSetupTask(),
TaskCompletionEvent.Status.SUCCEEDED,
taskTrackerHttpLocation
);
taskEvent.setTaskRunTime((int)(status.getFinishTime()
- status.getStartTime()));
tip.setSuccessEventNumber(taskCompletionEventTracker);
} else if (state == TaskStatus.State.COMMIT_PENDING) {
// If it is the first attempt reporting COMMIT_PENDING
// ask the task to commit.
if (!wasComplete && !wasPending) {
tip.doCommit(taskid);
}
return;
} else if (state == TaskStatus.State.FAILED_UNCLEAN ||
state == TaskStatus.State.KILLED_UNCLEAN) {
tip.incompleteSubTask(taskid, this.status);
// add this task, to be rescheduled as cleanup attempt
if (tip.isMapTask()) {
mapCleanupTasks.add(taskid);
} else {
reduceCleanupTasks.add(taskid);
}
// Remove the task entry from jobtracker
jobtracker.removeTaskEntry(taskid);
}
//For a failed task update the JT datastructures.
else if (state == TaskStatus.State.FAILED ||
state == TaskStatus.State.KILLED) {
// Get the event number for the (possibly) previously successful
// task. If there exists one, then set that status to OBSOLETE
int eventNumber;
if ((eventNumber = tip.getSuccessEventNumber()) != -1) {
TaskCompletionEvent t =
this.taskCompletionEvents.get(eventNumber);
if (t.getTaskAttemptId().equals(taskid))
t.setTaskStatus(TaskCompletionEvent.Status.OBSOLETE);
}
// Tell the job to fail the relevant task
failedTask(tip, taskid, status, taskTracker,
wasRunning, wasComplete, wasAttemptRunning);
// Did the task failure lead to tip failure?
TaskCompletionEvent.Status taskCompletionStatus =
(state == TaskStatus.State.FAILED ) ?
TaskCompletionEvent.Status.FAILED :
TaskCompletionEvent.Status.KILLED;
if (tip.isFailed()) {
taskCompletionStatus = TaskCompletionEvent.Status.TIPFAILED;
}
taskEvent = new TaskCompletionEvent(taskCompletionEventTracker,
taskid,
tip.idWithinJob(),
status.getIsMap() &&
!tip.isJobCleanupTask() &&
!tip.isJobSetupTask(),
taskCompletionStatus,
taskTrackerHttpLocation
);
}
// Add the 'complete' task i.e. successful/failed
// It _is_ safe to add the TaskCompletionEvent.Status.SUCCEEDED
// *before* calling TIP.completedTask since:
// a. One and only one task of a TIP is declared as a SUCCESS, the
// other (speculative tasks) are marked KILLED
// b. TIP.completedTask *does not* throw _any_ exception at all.
if (taskEvent != null) {
this.taskCompletionEvents.add(taskEvent);
taskCompletionEventTracker++;
JobTrackerStatistics.TaskTrackerStat ttStat = jobtracker.
getStatistics().getTaskTrackerStat(tip.machineWhereTaskRan(taskid));
if(ttStat != null) { // ttStat can be null in case of lost tracker
ttStat.incrTotalTasks();
}
if (state == TaskStatus.State.SUCCEEDED) {
completedTask(tip, status);
if(ttStat != null) {
ttStat.incrSucceededTasks();
}
}
}
}
//
// Update JobInProgress status
//
if (LOG.isDebugEnabled()) {
LOG.debug("Taking progress for " + tip.getTIPId() + " from " +
oldProgress + " to " + tip.getProgress());
}
if (!tip.isJobCleanupTask() && !tip.isJobSetupTask()) {
double progressDelta = tip.getProgress() - oldProgress;
if (tip.isMapTask()) {
this.status.setMapProgress((float) (this.status.mapProgress() +
progressDelta / maps.length));
} else {
this.status.setReduceProgress((float) (this.status.reduceProgress() +
(progressDelta / reduces.length)));
}
}
}
/**
* Returns the job-level counters.
*
* @return the job-level counters.
*/
public synchronized Counters getJobCounters() {
return jobCounters;
}
/**
* Returns map phase counters by summing over all map tasks in progress.
*/
public synchronized Counters getMapCounters() {
return incrementTaskCounters(new Counters(), maps);
}
/**
* Returns map phase counters by summing over all map tasks in progress.
*/
public synchronized Counters getReduceCounters() {
return incrementTaskCounters(new Counters(), reduces);
}
/**
* Returns the total job counters, by adding together the job,
* the map and the reduce counters.
*/
public Counters getCounters() {
Counters result = new Counters();
synchronized (this) {
result.incrAllCounters(getJobCounters());
}
// the counters of TIPs are not updated in place.
// hence read-only access is ok without any locks
incrementTaskCounters(result, maps);
return incrementTaskCounters(result, reduces);
}
/**
* Increments the counters with the counters from each task.
* @param counters the counters to increment
* @param tips the tasks to add in to counters
* @return counters the same object passed in as counters
*/
private Counters incrementTaskCounters(Counters counters,
TaskInProgress[] tips) {
for (TaskInProgress tip : tips) {
counters.incrAllCounters(tip.getCounters());
}
return counters;
}
/////////////////////////////////////////////////////
// Create/manage tasks
/////////////////////////////////////////////////////
/**
* Return a MapTask, if appropriate, to run on the given tasktracker
*/
public synchronized Task obtainNewMapTask(TaskTrackerStatus tts,
int clusterSize,
int numUniqueHosts,
int maxCacheLevel
) throws IOException {
if (status.getRunState() != JobStatus.RUNNING) {
LOG.info("Cannot create task split for " + profile.getJobID());
return null;
}
int target = findNewMapTask(tts, clusterSize, numUniqueHosts,
maxCacheLevel);
if (target == -1) {
return null;
}
Task result = maps[target].getTaskToRun(tts.getTrackerName());
if (result != null) {
addRunningTaskToTIP(maps[target], result.getTaskID(), tts, true);
}
return result;
}
/**
* Return a MapTask, if appropriate, to run on the given tasktracker
*/
public synchronized Task obtainNewMapTask(TaskTrackerStatus tts,
int clusterSize,
int numUniqueHosts
) throws IOException {
return obtainNewMapTask(tts, clusterSize, numUniqueHosts, anyCacheLevel);
}
/*
* Return task cleanup attempt if any, to run on a given tracker
*/
public Task obtainTaskCleanupTask(TaskTrackerStatus tts,
boolean isMapSlot)
throws IOException {
if (!tasksInited.get()) {
return null;
}
synchronized (this) {
if (this.status.getRunState() != JobStatus.RUNNING ||
jobFailed || jobKilled) {
return null;
}
String taskTracker = tts.getTrackerName();
if (!shouldRunOnTaskTracker(taskTracker)) {
return null;
}
TaskAttemptID taskid = null;
TaskInProgress tip = null;
if (isMapSlot) {
if (!mapCleanupTasks.isEmpty()) {
taskid = mapCleanupTasks.remove(0);
tip = maps[taskid.getTaskID().getId()];
}
} else {
if (!reduceCleanupTasks.isEmpty()) {
taskid = reduceCleanupTasks.remove(0);
tip = reduces[taskid.getTaskID().getId()];
}
}
if (tip != null) {
return tip.addRunningTask(taskid, taskTracker, true);
}
return null;
}
}
public synchronized Task obtainNewLocalMapTask(TaskTrackerStatus tts,
int clusterSize,
int numUniqueHosts)
throws IOException {
if (!tasksInited.get()) {
LOG.info("Cannot create task split for " + profile.getJobID());
return null;
}
return obtainNewMapTask(tts, clusterSize, numUniqueHosts, maxLevel);
}
public synchronized Task obtainNewNonLocalMapTask(TaskTrackerStatus tts,
int clusterSize,
int numUniqueHosts)
throws IOException {
if (!tasksInited.get()) {
LOG.info("Cannot create task split for " + profile.getJobID());
return null;
}
return obtainNewMapTask(tts, clusterSize, numUniqueHosts,
NON_LOCAL_CACHE_LEVEL);
}
/**
* Return a CleanupTask, if appropriate, to run on the given tasktracker
*
*/
public Task obtainJobCleanupTask(TaskTrackerStatus tts,
int clusterSize,
int numUniqueHosts,
boolean isMapSlot
) throws IOException {
if(!tasksInited.get() || !jobSetupCleanupNeeded) {
return null;
}
synchronized(this) {
if (!canLaunchJobCleanupTask()) {
return null;
}
String taskTracker = tts.getTrackerName();
// Update the last-known clusterSize
this.clusterSize = clusterSize;
if (!shouldRunOnTaskTracker(taskTracker)) {
return null;
}
List<TaskInProgress> cleanupTaskList = new ArrayList<TaskInProgress>();
if (isMapSlot) {
cleanupTaskList.add(cleanup[0]);
} else {
cleanupTaskList.add(cleanup[1]);
}
TaskInProgress tip = findTaskFromList(cleanupTaskList,
tts, numUniqueHosts, false);
if (tip == null) {
return null;
}
// Now launch the cleanupTask
Task result = tip.getTaskToRun(tts.getTrackerName());
if (result != null) {
addRunningTaskToTIP(tip, result.getTaskID(), tts, true);
if (jobFailed) {
result.setJobCleanupTaskState(org.apache.hadoop.mapreduce.JobStatus
.State.FAILED);
} else if (jobKilled) {
result.setJobCleanupTaskState(org.apache.hadoop.mapreduce.JobStatus
.State.KILLED);
} else {
result.setJobCleanupTaskState(org.apache.hadoop.mapreduce
.JobStatus.State.SUCCEEDED);
}
}
return result;
}
}
/**
* Check whether cleanup task can be launched for the job.
*
* Cleanup task can be launched if it is not already launched
* or job is Killed
* or all maps and reduces are complete
* @return true/false
*/
private synchronized boolean canLaunchJobCleanupTask() {
// check if the job is running
if (status.getRunState() != JobStatus.RUNNING &&
status.getRunState() != JobStatus.PREP) {
return false;
}
// check if cleanup task has been launched already or if setup isn't
// launched already. The later check is useful when number of maps is
// zero.
if (launchedCleanup || !isSetupFinished()) {
return false;
}
// check if job has failed or killed
if (jobKilled || jobFailed) {
return true;
}
// Check if all maps and reducers have finished.
boolean launchCleanupTask =
((finishedMapTasks + failedMapTIPs) == (numMapTasks));
if (launchCleanupTask) {
launchCleanupTask =
((finishedReduceTasks + failedReduceTIPs) == numReduceTasks);
}
return launchCleanupTask;
}
/**
* Return a SetupTask, if appropriate, to run on the given tasktracker
*
*/
public Task obtainJobSetupTask(TaskTrackerStatus tts,
int clusterSize,
int numUniqueHosts,
boolean isMapSlot
) throws IOException {
if(!tasksInited.get() || !jobSetupCleanupNeeded) {
return null;
}
synchronized(this) {
if (!canLaunchSetupTask()) {
return null;
}
String taskTracker = tts.getTrackerName();
// Update the last-known clusterSize
this.clusterSize = clusterSize;
if (!shouldRunOnTaskTracker(taskTracker)) {
return null;
}
List<TaskInProgress> setupTaskList = new ArrayList<TaskInProgress>();
if (isMapSlot) {
setupTaskList.add(setup[0]);
} else {
setupTaskList.add(setup[1]);
}
TaskInProgress tip = findTaskFromList(setupTaskList,
tts, numUniqueHosts, false);
if (tip == null) {
return null;
}
// Now launch the setupTask
Task result = tip.getTaskToRun(tts.getTrackerName());
if (result != null) {
addRunningTaskToTIP(tip, result.getTaskID(), tts, true);
}
return result;
}
}
public synchronized boolean scheduleReduces() {
return finishedMapTasks >= completedMapsForReduceSlowstart;
}
/**
* Check whether setup task can be launched for the job.
*
* Setup task can be launched after the tasks are inited
* and Job is in PREP state
* and if it is not already launched
* or job is not Killed/Failed
* @return true/false
*/
private synchronized boolean canLaunchSetupTask() {
return (tasksInited.get() && status.getRunState() == JobStatus.PREP &&
!launchedSetup && !jobKilled && !jobFailed);
}
/**
* Return a ReduceTask, if appropriate, to run on the given tasktracker.
* We don't have cache-sensitivity for reduce tasks, as they
* work on temporary MapRed files.
*/
public synchronized Task obtainNewReduceTask(TaskTrackerStatus tts,
int clusterSize,
int numUniqueHosts
) throws IOException {
if (status.getRunState() != JobStatus.RUNNING) {
LOG.info("Cannot create task split for " + profile.getJobID());
return null;
}
// Ensure we have sufficient map outputs ready to shuffle before
// scheduling reduces
if (!scheduleReduces()) {
return null;
}
int target = findNewReduceTask(tts, clusterSize, numUniqueHosts);
if (target == -1) {
return null;
}
Task result = reduces[target].getTaskToRun(tts.getTrackerName());
if (result != null) {
addRunningTaskToTIP(reduces[target], result.getTaskID(), tts, true);
}
return result;
}
// returns the (cache)level at which the nodes matches
private int getMatchingLevelForNodes(Node n1, Node n2) {
int count = 0;
do {
if (n1.equals(n2)) {
return count;
}
++count;
n1 = n1.getParent();
n2 = n2.getParent();
} while (n1 != null);
return this.maxLevel;
}
/**
* Populate the data structures as a task is scheduled.
*
* Assuming {@link JobTracker} is locked on entry.
*
* @param tip The tip for which the task is added
* @param id The attempt-id for the task
* @param tts task-tracker status
* @param isScheduled Whether this task is scheduled from the JT or has
* joined back upon restart
*/
synchronized void addRunningTaskToTIP(TaskInProgress tip, TaskAttemptID id,
TaskTrackerStatus tts,
boolean isScheduled) {
// Make an entry in the tip if the attempt is not scheduled i.e externally
// added
if (!isScheduled) {
tip.addRunningTask(id, tts.getTrackerName());
}
final JobTrackerInstrumentation metrics = jobtracker.getInstrumentation();
// keeping the earlier ordering intact
TaskType name;
String splits = "";
Enum counter = null;
if (tip.isJobSetupTask()) {
launchedSetup = true;
name = TaskType.JOB_SETUP;
} else if (tip.isJobCleanupTask()) {
launchedCleanup = true;
name = TaskType.JOB_CLEANUP;
} else if (tip.isMapTask()) {
++runningMapTasks;
name = TaskType.MAP;
counter = JobCounter.TOTAL_LAUNCHED_MAPS;
splits = tip.getSplitNodes();
if (tip.isSpeculating()) {
speculativeMapTasks++;
metrics.speculateMap(id);
if (LOG.isDebugEnabled()) {
LOG.debug("Chosen speculative task, current speculativeMap task count: "
+ speculativeMapTasks);
}
}
metrics.launchMap(id);
} else {
++runningReduceTasks;
name = TaskType.REDUCE;
counter = JobCounter.TOTAL_LAUNCHED_REDUCES;
if (tip.isSpeculating()) {
speculativeReduceTasks++;
metrics.speculateReduce(id);
if (LOG.isDebugEnabled()) {
LOG.debug("Chosen speculative task, current speculativeReduce task count: "
+ speculativeReduceTasks);
}
}
metrics.launchReduce(id);
}
// Note that the logs are for the scheduled tasks only. Tasks that join on
// restart has already their logs in place.
if (tip.isFirstAttempt(id)) {
TaskStartedEvent tse = new TaskStartedEvent(tip.getTIPId(),
tip.getExecStartTime(),
name, splits);
jobHistory.logEvent(tse, tip.getJob().jobId);
setFirstTaskLaunchTime(tip);
}
if (!tip.isJobSetupTask() && !tip.isJobCleanupTask()) {
jobCounters.incrCounter(counter, 1);
}
//TODO The only problem with these counters would be on restart.
// The jobtracker updates the counter only when the task that is scheduled
// if from a non-running tip and is local (data, rack ...). But upon restart
// as the reports come from the task tracker, there is no good way to infer
// when exactly to increment the locality counters. The only solution is to
// increment the counters for all the tasks irrespective of
// - whether the tip is running or not
// - whether its a speculative task or not
//
// So to simplify, increment the data locality counter whenever there is
// data locality.
if (tip.isMapTask() && !tip.isJobSetupTask() && !tip.isJobCleanupTask()) {
// increment the data locality counter for maps
int level = getLocalityLevel(tip, tts);
switch (level) {
case 0 :
LOG.info("Choosing data-local task " + tip.getTIPId());
jobCounters.incrCounter(JobCounter.DATA_LOCAL_MAPS, 1);
metrics.launchDataLocalMap(id);
break;
case 1:
LOG.info("Choosing rack-local task " + tip.getTIPId());
jobCounters.incrCounter(JobCounter.RACK_LOCAL_MAPS, 1);
metrics.launchRackLocalMap(id);
break;
default :
// check if there is any locality
if (level != this.maxLevel) {
LOG.info("Choosing cached task at level " + level + tip.getTIPId());
jobCounters.incrCounter(JobCounter.OTHER_LOCAL_MAPS, 1);
}
break;
}
}
}
void setFirstTaskLaunchTime(TaskInProgress tip) {
TaskType key = getTaskType(tip);
synchronized(firstTaskLaunchTimes) {
// Could be optimized to do only one lookup with a little more code
if (!firstTaskLaunchTimes.containsKey(key)) {
firstTaskLaunchTimes.put(key, tip.getExecStartTime());
}
}
}
public static String convertTrackerNameToHostName(String trackerName) {
// Ugly!
// Convert the trackerName to it's host name
int indexOfColon = trackerName.indexOf(":");
String trackerHostName = (indexOfColon == -1) ?
trackerName :
trackerName.substring(0, indexOfColon);
return trackerHostName.substring("tracker_".length());
}
/**
* Note that a task has failed on a given tracker and add the tracker
* to the blacklist iff too many trackers in the cluster i.e.
* (clusterSize * CLUSTER_BLACKLIST_PERCENT) haven't turned 'flaky' already.
*
* @param taskTracker task-tracker on which a task failed
*/
synchronized void addTrackerTaskFailure(String trackerName,
TaskTracker taskTracker) {
if (flakyTaskTrackers < (clusterSize * CLUSTER_BLACKLIST_PERCENT)) {
String trackerHostName = convertTrackerNameToHostName(trackerName);
Integer trackerFailures = trackerToFailuresMap.get(trackerHostName);
if (trackerFailures == null) {
trackerFailures = 0;
}
trackerToFailuresMap.put(trackerHostName, ++trackerFailures);
// Check if this tasktracker has turned 'flaky'
if (trackerFailures.intValue() == maxTaskFailuresPerTracker) {
++flakyTaskTrackers;
// Cancel reservations if appropriate
if (taskTracker != null) {
if (trackersReservedForMaps.containsKey(taskTracker)) {
taskTracker.unreserveSlots(TaskType.MAP, this);
}
if (trackersReservedForReduces.containsKey(taskTracker)) {
taskTracker.unreserveSlots(TaskType.REDUCE, this);
}
}
LOG.info("TaskTracker at '" + trackerHostName + "' turned 'flaky'");
}
}
}
public synchronized void reserveTaskTracker(TaskTracker taskTracker,
TaskType type, int numSlots) {
Map<TaskTracker, FallowSlotInfo> map =
(type == TaskType.MAP) ? trackersReservedForMaps : trackersReservedForReduces;
long now = System.currentTimeMillis();
FallowSlotInfo info = map.get(taskTracker);
int reservedSlots = 0;
if (info == null) {
info = new FallowSlotInfo(now, numSlots);
reservedSlots = numSlots;
} else {
// Increment metering info if the reservation is changing
if (info.getNumSlots() != numSlots) {
Enum<JobCounter> counter =
(type == TaskType.MAP) ?
JobCounter.FALLOW_SLOTS_MILLIS_MAPS :
JobCounter.FALLOW_SLOTS_MILLIS_REDUCES;
long fallowSlotMillis = (now - info.getTimestamp()) * info.getNumSlots();
jobCounters.incrCounter(counter, fallowSlotMillis);
// Update
reservedSlots = numSlots - info.getNumSlots();
info.setTimestamp(now);
info.setNumSlots(numSlots);
}
}
map.put(taskTracker, info);
if (type == TaskType.MAP) {
jobtracker.getInstrumentation().addReservedMapSlots(reservedSlots);
}
else {
jobtracker.getInstrumentation().addReservedReduceSlots(reservedSlots);
}
jobtracker.incrementReservations(type, reservedSlots);
}
public synchronized void unreserveTaskTracker(TaskTracker taskTracker,
TaskType type) {
Map<TaskTracker, FallowSlotInfo> map =
(type == TaskType.MAP) ? trackersReservedForMaps :
trackersReservedForReduces;
FallowSlotInfo info = map.get(taskTracker);
if (info == null) {
LOG.warn("Cannot find information about fallow slots for " +
taskTracker.getTrackerName());
return;
}
long now = System.currentTimeMillis();
Enum<JobCounter> counter =
(type == TaskType.MAP) ?
JobCounter.FALLOW_SLOTS_MILLIS_MAPS :
JobCounter.FALLOW_SLOTS_MILLIS_REDUCES;
long fallowSlotMillis = (now - info.getTimestamp()) * info.getNumSlots();
jobCounters.incrCounter(counter, fallowSlotMillis);
map.remove(taskTracker);
if (type == TaskType.MAP) {
jobtracker.getInstrumentation().decReservedMapSlots(info.getNumSlots());
}
else {
jobtracker.getInstrumentation().decReservedReduceSlots(
info.getNumSlots());
}
jobtracker.decrementReservations(type, info.getNumSlots());
}
public int getNumReservedTaskTrackersForMaps() {
return trackersReservedForMaps.size();
}
public int getNumReservedTaskTrackersForReduces() {
return trackersReservedForReduces.size();
}
private int getTrackerTaskFailures(String trackerName) {
String trackerHostName = convertTrackerNameToHostName(trackerName);
Integer failedTasks = trackerToFailuresMap.get(trackerHostName);
return (failedTasks != null) ? failedTasks.intValue() : 0;
}
/**
* Get the black listed trackers for the job
*
* @return List of blacklisted tracker names
*/
List<String> getBlackListedTrackers() {
List<String> blackListedTrackers = new ArrayList<String>();
for (Map.Entry<String,Integer> e : trackerToFailuresMap.entrySet()) {
if (e.getValue().intValue() >= maxTaskFailuresPerTracker) {
blackListedTrackers.add(e.getKey());
}
}
return blackListedTrackers;
}
/**
* Get the no. of 'flaky' tasktrackers for a given job.
*
* @return the no. of 'flaky' tasktrackers for a given job.
*/
int getNoOfBlackListedTrackers() {
return flakyTaskTrackers;
}
/**
* Get the information on tasktrackers and no. of errors which occurred
* on them for a given job.
*
* @return the map of tasktrackers and no. of errors which occurred
* on them for a given job.
*/
synchronized Map<String, Integer> getTaskTrackerErrors() {
// Clone the 'trackerToFailuresMap' and return the copy
Map<String, Integer> trackerErrors =
new TreeMap<String, Integer>(trackerToFailuresMap);
return trackerErrors;
}
/**
* Remove a map TIP from the lists for running maps.
* Called when a map fails/completes (note if a map is killed,
* it won't be present in the list since it was completed earlier)
* @param tip the tip that needs to be retired
*/
private synchronized void retireMap(TaskInProgress tip) {
if (runningMapCache == null) {
LOG.warn("Running cache for maps missing!! "
+ "Job details are missing.");
return;
}
String[] splitLocations = tip.getSplitLocations();
// Remove the TIP from the list for running non-local maps
if (splitLocations == null || splitLocations.length == 0) {
nonLocalRunningMaps.remove(tip);
return;
}
// Remove from the running map caches
for(String host: splitLocations) {
Node node = jobtracker.getNode(host);
for (int j = 0; j < maxLevel; ++j) {
Set<TaskInProgress> hostMaps = runningMapCache.get(node);
if (hostMaps != null) {
hostMaps.remove(tip);
if (hostMaps.size() == 0) {
runningMapCache.remove(node);
}
}
node = node.getParent();
}
}
}
/**
* Remove a reduce TIP from the list for running-reduces
* Called when a reduce fails/completes
* @param tip the tip that needs to be retired
*/
private synchronized void retireReduce(TaskInProgress tip) {
if (runningReduces == null) {
LOG.warn("Running list for reducers missing!! "
+ "Job details are missing.");
return;
}
runningReduces.remove(tip);
}
/**
* Adds a map tip to the list of running maps.
* @param tip the tip that needs to be scheduled as running
*/
protected synchronized void scheduleMap(TaskInProgress tip) {
runningMapTaskStats.add(0.0f);
if (runningMapCache == null) {
LOG.warn("Running cache for maps is missing!! "
+ "Job details are missing.");
return;
}
String[] splitLocations = tip.getSplitLocations();
// Add the TIP to the list of non-local running TIPs
if (splitLocations == null || splitLocations.length == 0) {
nonLocalRunningMaps.add(tip);
return;
}
for(String host: splitLocations) {
Node node = jobtracker.getNode(host);
for (int j = 0; j < maxLevel; ++j) {
Set<TaskInProgress> hostMaps = runningMapCache.get(node);
if (hostMaps == null) {
// create a cache if needed
hostMaps = new LinkedHashSet<TaskInProgress>();
runningMapCache.put(node, hostMaps);
}
hostMaps.add(tip);
node = node.getParent();
}
}
}
/**
* Adds a reduce tip to the list of running reduces
* @param tip the tip that needs to be scheduled as running
*/
protected synchronized void scheduleReduce(TaskInProgress tip) {
runningReduceTaskStats.add(0.0f);
if (runningReduces == null) {
LOG.warn("Running cache for reducers missing!! "
+ "Job details are missing.");
return;
}
runningReduces.add(tip);
}
/**
* Adds the failed TIP in the front of the list for non-running maps
* @param tip the tip that needs to be failed
*/
private synchronized void failMap(TaskInProgress tip) {
if (nonRunningMapCache == null) {
LOG.warn("Non-running cache for maps missing!! "
+ "Job details are missing.");
return;
}
// 1. Its added everywhere since other nodes (having this split local)
// might have removed this tip from their local cache
// 2. Give high priority to failed tip - fail early
String[] splitLocations = tip.getSplitLocations();
// Add the TIP in the front of the list for non-local non-running maps
if (splitLocations.length == 0) {
nonLocalMaps.add(0, tip);
return;
}
for(String host: splitLocations) {
Node node = jobtracker.getNode(host);
for (int j = 0; j < maxLevel; ++j) {
List<TaskInProgress> hostMaps = nonRunningMapCache.get(node);
if (hostMaps == null) {
hostMaps = new LinkedList<TaskInProgress>();
nonRunningMapCache.put(node, hostMaps);
}
hostMaps.add(0, tip);
node = node.getParent();
}
}
}
/**
* Adds a failed TIP in the front of the list for non-running reduces
* @param tip the tip that needs to be failed
*/
private synchronized void failReduce(TaskInProgress tip) {
if (nonRunningReduces == null) {
LOG.warn("Failed cache for reducers missing!! "
+ "Job details are missing.");
return;
}
nonRunningReduces.add(0, tip);
}
/**
* Find a non-running task in the passed list of TIPs
* @param tips a collection of TIPs
* @param ttStatus the status of tracker that has requested a task to run
* @param numUniqueHosts number of unique hosts that run trask trackers
* @param removeFailedTip whether to remove the failed tips
*/
private synchronized TaskInProgress findTaskFromList(
Collection<TaskInProgress> tips, TaskTrackerStatus ttStatus,
int numUniqueHosts,
boolean removeFailedTip) {
Iterator<TaskInProgress> iter = tips.iterator();
while (iter.hasNext()) {
TaskInProgress tip = iter.next();
// Select a tip if
// 1. runnable : still needs to be run and is not completed
// 2. ~running : no other node is running it
// 3. earlier attempt failed : has not failed on this host
// and has failed on all the other hosts
// A TIP is removed from the list if
// (1) this tip is scheduled
// (2) if the passed list is a level 0 (host) cache
// (3) when the TIP is non-schedulable (running, killed, complete)
if (tip.isRunnable() && !tip.isRunning()) {
// check if the tip has failed on this host
if (!tip.hasFailedOnMachine(ttStatus.getHost()) ||
tip.getNumberOfFailedMachines() >= numUniqueHosts) {
// check if the tip has failed on all the nodes
iter.remove();
return tip;
} else if (removeFailedTip) {
// the case where we want to remove a failed tip from the host cache
// point#3 in the TIP removal logic above
iter.remove();
}
} else {
// see point#3 in the comment above for TIP removal logic
iter.remove();
}
}
return null;
}
public boolean hasSpeculativeMaps() {
return hasSpeculativeMaps;
}
public boolean hasSpeculativeReduces() {
return hasSpeculativeReduces;
}
/**
* Retrieve a task for speculation.
* If a task slot becomes available and there are less than SpeculativeCap
* speculative tasks running:
* 1)Ignore the request if the TT's progressRate is < SlowNodeThreshold
* 2)Choose candidate tasks - those tasks whose progress rate is below
* slowTaskThreshold * mean(progress-rates)
* 3)Speculate task that's expected to complete last
* @param list pool of tasks to choose from
* @param taskTrackerName the name of the TaskTracker asking for a task
* @param taskTrackerHost the hostname of the TaskTracker asking for a task
* @param taskType the type of task (MAP/REDUCE) that we are considering
* @return the TIP to speculatively re-execute
*/
protected synchronized TaskInProgress findSpeculativeTask(
Collection<TaskInProgress> list, String taskTrackerName,
String taskTrackerHost, TaskType taskType) {
if (list.isEmpty()) {
return null;
}
long now = JobTracker.getClock().getTime();
// Don't return anything if either the TaskTracker is slow or we have
// already launched enough speculative tasks in the cluster.
if (isSlowTracker(taskTrackerName) || atSpeculativeCap(list, taskType)) {
return null;
}
TaskInProgress slowestTIP = null;
Comparator<TaskInProgress> LateComparator =
new EstimatedTimeLeftComparator(now);
Iterator<TaskInProgress> iter = list.iterator();
while (iter.hasNext()) {
TaskInProgress tip = iter.next();
// If this tip has already run on this machine once or it doesn't need any
// more speculative attempts, skip it.
if (tip.hasRunOnMachine(taskTrackerHost, taskTrackerName) ||
!tip.canBeSpeculated(now)) {
continue;
}
if (slowestTIP == null) {
slowestTIP = tip;
} else {
slowestTIP =
LateComparator.compare(tip, slowestTIP) < 0 ? tip : slowestTIP;
}
}
if (slowestTIP != null) {
if (LOG.isDebugEnabled()) {
LOG.debug("Chose task " + slowestTIP.getTIPId() + ". Statistics: Task's : " +
slowestTIP.getCurrentProgressRate(now) + " Job's : " +
(slowestTIP.isMapTask() ? runningMapTaskStats : runningReduceTaskStats));
}
}
return slowestTIP;
}
/**
* Find new map task
* @param tts The task tracker that is asking for a task
* @param clusterSize The number of task trackers in the cluster
* @param numUniqueHosts The number of hosts that run task trackers
* @param maxCacheLevel The maximum topology level until which to schedule
* maps.
* A value of {@link #anyCacheLevel} implies any
* available task (node-local, rack-local, off-switch and
* speculative tasks).
* A value of {@link #NON_LOCAL_CACHE_LEVEL} implies only
* off-switch/speculative tasks should be scheduled.
* @return the index in tasks of the selected task (or -1 for no task)
*/
private synchronized int findNewMapTask(final TaskTrackerStatus tts,
final int clusterSize,
final int numUniqueHosts,
final int maxCacheLevel) {
String taskTrackerName = tts.getTrackerName();
String taskTrackerHost = tts.getHost();
if (numMapTasks == 0) {
if(LOG.isDebugEnabled()) {
LOG.debug("No maps to schedule for " + profile.getJobID());
}
return -1;
}
TaskInProgress tip = null;
//
// Update the last-known clusterSize
//
this.clusterSize = clusterSize;
if (!shouldRunOnTaskTracker(taskTrackerName)) {
return -1;
}
// Check to ensure this TaskTracker has enough resources to
// run tasks from this job
long outSize = resourceEstimator.getEstimatedMapOutputSize();
long availSpace = tts.getResourceStatus().getAvailableSpace();
if(availSpace < outSize) {
LOG.warn("No room for map task. Node " + tts.getHost() +
" has " + availSpace +
" bytes free; but we expect map to take " + outSize);
return -1; //see if a different TIP might work better.
}
// For scheduling a map task, we have two caches and a list (optional)
// I) one for non-running task
// II) one for running task (this is for handling speculation)
// III) a list of TIPs that have empty locations (e.g., dummy splits),
// the list is empty if all TIPs have associated locations
// First a look up is done on the non-running cache and on a miss, a look
// up is done on the running cache. The order for lookup within the cache:
// 1. from local node to root [bottom up]
// 2. breadth wise for all the parent nodes at max level
// We fall to linear scan of the list (III above) if we have misses in the
// above caches
Node node = jobtracker.getNode(tts.getHost());
//
// I) Non-running TIP :
//
// 1. check from local node to the root [bottom up cache lookup]
// i.e if the cache is available and the host has been resolved
// (node!=null)
if (node != null) {
Node key = node;
int level = 0;
// maxCacheLevel might be greater than this.maxLevel if findNewMapTask is
// called to schedule any task (local, rack-local, off-switch or speculative)
// tasks or it might be NON_LOCAL_CACHE_LEVEL (i.e. -1) if findNewMapTask is
// (i.e. -1) if findNewMapTask is to only schedule off-switch/speculative
// tasks
int maxLevelToSchedule = Math.min(maxCacheLevel, maxLevel);
for (level = 0;level < maxLevelToSchedule; ++level) {
List <TaskInProgress> cacheForLevel = nonRunningMapCache.get(key);
if (cacheForLevel != null) {
tip = findTaskFromList(cacheForLevel, tts,
numUniqueHosts,level == 0);
if (tip != null) {
// Add to running cache
scheduleMap(tip);
// remove the cache if its empty
if (cacheForLevel.size() == 0) {
nonRunningMapCache.remove(key);
}
return tip.getIdWithinJob();
}
}
key = key.getParent();
}
// Check if we need to only schedule a local task (node-local/rack-local)
if (level == maxCacheLevel) {
return -1;
}
}
//2. Search breadth-wise across parents at max level for non-running
// TIP if
// - cache exists and there is a cache miss
// - node information for the tracker is missing (tracker's topology
// info not obtained yet)
// collection of node at max level in the cache structure
Collection<Node> nodesAtMaxLevel = jobtracker.getNodesAtMaxLevel();
// get the node parent at max level
Node nodeParentAtMaxLevel =
(node == null) ? null : JobTracker.getParentNode(node, maxLevel - 1);
for (Node parent : nodesAtMaxLevel) {
// skip the parent that has already been scanned
if (parent == nodeParentAtMaxLevel) {
continue;
}
List<TaskInProgress> cache = nonRunningMapCache.get(parent);
if (cache != null) {
tip = findTaskFromList(cache, tts, numUniqueHosts, false);
if (tip != null) {
// Add to the running cache
scheduleMap(tip);
// remove the cache if empty
if (cache.size() == 0) {
nonRunningMapCache.remove(parent);
}
LOG.info("Choosing a non-local task " + tip.getTIPId());
return tip.getIdWithinJob();
}
}
}
// 3. Search non-local tips for a new task
tip = findTaskFromList(nonLocalMaps, tts, numUniqueHosts, false);
if (tip != null) {
// Add to the running list
scheduleMap(tip);
LOG.info("Choosing a non-local task " + tip.getTIPId());
return tip.getIdWithinJob();
}
//
// II) Running TIP :
//
if (hasSpeculativeMaps) {
tip = getSpeculativeMap(taskTrackerName, taskTrackerHost);
if (tip != null) {
return tip.getIdWithinJob();
}
}
return -1;
}
private synchronized TaskInProgress getSpeculativeMap(String taskTrackerName,
String taskTrackerHost) {
//////// Populate allTips with all TaskInProgress
Set<TaskInProgress> allTips = new HashSet<TaskInProgress>();
// collection of node at max level in the cache structure
Collection<Node> nodesAtMaxLevel = jobtracker.getNodesAtMaxLevel();
// Add all tasks from max-level nodes breadth-wise
for (Node parent : nodesAtMaxLevel) {
Set<TaskInProgress> cache = runningMapCache.get(parent);
if (cache != null) {
allTips.addAll(cache);
}
}
// Add all non-local TIPs
allTips.addAll(nonLocalRunningMaps);
///////// Select a TIP to run on
TaskInProgress tip = findSpeculativeTask(allTips, taskTrackerName,
taskTrackerHost, TaskType.MAP);
if (tip != null) {
LOG.info("Choosing map task " + tip.getTIPId() +
" for speculative execution");
} else {
if (LOG.isDebugEnabled()) {
LOG.debug("No speculative map task found for tracker " + taskTrackerName);
}
}
return tip;
}
/**
* Find new reduce task
* @param tts The task tracker that is asking for a task
* @param clusterSize The number of task trackers in the cluster
* @param numUniqueHosts The number of hosts that run task trackers
* @return the index in tasks of the selected task (or -1 for no task)
*/
private synchronized int findNewReduceTask(TaskTrackerStatus tts,
int clusterSize,
int numUniqueHosts) {
String taskTrackerName = tts.getTrackerName();
String taskTrackerHost = tts.getHost();
if (numReduceTasks == 0) {
if(LOG.isDebugEnabled()) {
LOG.debug("No reduces to schedule for " + profile.getJobID());
}
return -1;
}
TaskInProgress tip = null;
// Update the last-known clusterSize
this.clusterSize = clusterSize;
if (!shouldRunOnTaskTracker(taskTrackerName)) {
return -1;
}
long outSize = resourceEstimator.getEstimatedReduceInputSize();
long availSpace = tts.getResourceStatus().getAvailableSpace();
if(availSpace < outSize) {
LOG.warn("No room for reduce task. Node " + taskTrackerName + " has " +
availSpace +
" bytes free; but we expect reduce input to take " + outSize);
return -1; //see if a different TIP might work better.
}
// 1. check for a never-executed reduce tip
// reducers don't have a cache and so pass -1 to explicitly call that out
tip = findTaskFromList(nonRunningReduces, tts, numUniqueHosts, false);
if (tip != null) {
scheduleReduce(tip);
return tip.getIdWithinJob();
}
// 2. check for a reduce tip to be speculated
if (hasSpeculativeReduces) {
tip = getSpeculativeReduce(taskTrackerName, taskTrackerHost);
if (tip != null) {
return tip.getIdWithinJob();
}
}
return -1;
}
private synchronized TaskInProgress getSpeculativeReduce(
String taskTrackerName, String taskTrackerHost) {
TaskInProgress tip = findSpeculativeTask(
runningReduces, taskTrackerName, taskTrackerHost, TaskType.REDUCE);
if (tip != null) {
LOG.info("Choosing reduce task " + tip.getTIPId() +
" for speculative execution");
} else {
if (LOG.isDebugEnabled()) {
LOG.debug("No speculative map task found for tracker "
+ taskTrackerHost);
}
}
return tip;
}
/**
* Check to see if the maximum number of speculative tasks are
* already being executed currently.
* @param tasks the set of tasks to test
* @param type the type of task (MAP/REDUCE) that we are considering
* @return has the cap been reached?
*/
private boolean atSpeculativeCap(Collection<TaskInProgress> tasks,
TaskType type) {
float numTasks = tasks.size();
if (numTasks == 0){
return true; // avoid divide by zero
}
int speculativeTaskCount = type == TaskType.MAP ? speculativeMapTasks
: speculativeReduceTasks;
//return true if totalSpecTask < max(10, 0.01 * total-slots,
// 0.1 * total-running-tasks)
if (speculativeTaskCount < MIN_SPEC_CAP) {
return false; // at least one slow tracker's worth of slots(default=10)
}
ClusterStatus c = jobtracker.getClusterStatus(false);
int numSlots = (type == TaskType.MAP ? c.getMaxMapTasks() : c.getMaxReduceTasks());
if ((float)speculativeTaskCount < numSlots * MIN_SLOTS_CAP) {
return false;
}
boolean atCap = (((float)(speculativeTaskCount)/numTasks) >= speculativeCap);
if (LOG.isDebugEnabled()) {
LOG.debug("SpeculativeCap is "+speculativeCap+", specTasks/numTasks is " +
((float)(speculativeTaskCount)/numTasks)+
", so atSpecCap() is returning "+atCap);
}
return atCap;
}
/**
* A class for comparing the estimated time to completion of two tasks
*/
private static class EstimatedTimeLeftComparator
implements Comparator<TaskInProgress> {
private long time;
public EstimatedTimeLeftComparator(long now) {
this.time = now;
}
/**
* Estimated time to completion is measured as:
* % of task left to complete (1 - progress) / progress rate of the task.
*
* This assumes that tasks are linear in their progress, which is
* often wrong, especially since progress for reducers is currently
* calculated by evenly weighting their three stages (shuffle, sort, map)
* which rarely account for 1/3 each. This should be fixed in the future
* by calculating progressRate more intelligently or splitting these
* multi-phase tasks into individual tasks.
*
* The ordering this comparator defines is: task1 < task2 if task1 is
* estimated to finish farther in the future => compare(t1,t2) returns -1
*/
public int compare(TaskInProgress tip1, TaskInProgress tip2) {
//we have to use the Math.max in the denominator to avoid divide by zero
//error because prog and progRate can both be zero (if one is zero,
//the other one will be 0 too).
//We use inverse of time_reminaing=[(1- prog) / progRate]
//so that (1-prog) is in denom. because tasks can have arbitrarily
//low progRates in practice (e.g. a task that is half done after 1000
//seconds will have progRate of 0.0000005) so we would rather
//use Math.maxnon (1-prog) by putting it in the denominator
//which will cause tasks with prog=1 look 99.99% done instead of 100%
//which is okay
double t1 = tip1.getCurrentProgressRate(time) / Math.max(0.0001,
1.0 - tip1.getProgress());
double t2 = tip2.getCurrentProgressRate(time) / Math.max(0.0001,
1.0 - tip2.getProgress());
if (t1 < t2) return -1;
else if (t2 < t1) return 1;
else return 0;
}
}
/**
* Compares the ave progressRate of tasks that have finished on this
* taskTracker to the ave of all succesfull tasks thus far to see if this
* TT one is too slow for speculating.
* slowNodeThreshold is used to determine the number of standard deviations
* @param taskTracker the name of the TaskTracker we are checking
* @return is this TaskTracker slow
*/
protected boolean isSlowTracker(String taskTracker) {
if (trackerMapStats.get(taskTracker) != null &&
trackerMapStats.get(taskTracker).mean() -
mapTaskStats.mean() > mapTaskStats.std()*slowNodeThreshold) {
if (LOG.isDebugEnabled()) {
LOG.debug("Tracker " + taskTracker +
" declared slow. trackerMapStats.get(taskTracker).mean() :" + trackerMapStats.get(taskTracker).mean() +
" mapTaskStats :" + mapTaskStats);
}
return true;
}
if (trackerReduceStats.get(taskTracker) != null &&
trackerReduceStats.get(taskTracker).mean() -
reduceTaskStats.mean() > reduceTaskStats.std()*slowNodeThreshold) {
if (LOG.isDebugEnabled()) {
LOG.debug("Tracker " + taskTracker +
" declared slow. trackerReduceStats.get(taskTracker).mean() :" + trackerReduceStats.get(taskTracker).mean() +
" reduceTaskStats :" + reduceTaskStats);
}
return true;
}
return false;
}
static class DataStatistics{
private int count = 0;
private double sum = 0;
private double sumSquares = 0;
public DataStatistics() {
}
public DataStatistics(double initNum) {
this.count = 1;
this.sum = initNum;
this.sumSquares = initNum * initNum;
}
public void add(double newNum) {
this.count++;
this.sum += newNum;
this.sumSquares += newNum * newNum;
}
public void updateStatistics(double old, double update) {
sub(old);
add(update);
}
private void sub(double oldNum) {
this.count--;
this.sum = Math.max(this.sum -= oldNum, 0.0d);
this.sumSquares = Math.max(this.sumSquares -= oldNum * oldNum, 0.0d);
}
public double mean() {
return sum/count;
}
public double var() {
// E(X^2) - E(X)^2
return Math.max((sumSquares/count) - mean() * mean(), 0.0d);
}
public double std() {
return Math.sqrt(this.var());
}
public String toString() {
return "DataStatistics: count is " + count + ", sum is " + sum +
", sumSquares is " + sumSquares + " mean is " + mean() + " std() is " + std();
}
}
private boolean shouldRunOnTaskTracker(String taskTracker) {
//
// Check if too many tasks of this job have failed on this
// tasktracker prior to assigning it a new one.
//
int taskTrackerFailedTasks = getTrackerTaskFailures(taskTracker);
if ((flakyTaskTrackers < (clusterSize * CLUSTER_BLACKLIST_PERCENT)) &&
taskTrackerFailedTasks >= maxTaskFailuresPerTracker) {
if (LOG.isDebugEnabled()) {
String flakyTracker = convertTrackerNameToHostName(taskTracker);
if (LOG.isDebugEnabled()) {
LOG.debug("Ignoring the black-listed tasktracker: '" + flakyTracker
+ "' for assigning a new task");
}
}
return false;
}
return true;
}
/**
* Metering: Occupied Slots * (Finish - Start)
* @param tip {@link TaskInProgress} to be metered which just completed,
* cannot be <code>null</code>
* @param status {@link TaskStatus} of the completed task, cannot be
* <code>null</code>
*/
private void meterTaskAttempt(TaskInProgress tip, TaskStatus status) {
JobCounter slotCounter =
(tip.isMapTask()) ? JobCounter.SLOTS_MILLIS_MAPS :
JobCounter.SLOTS_MILLIS_REDUCES;
jobCounters.incrCounter(slotCounter,
tip.getNumSlotsRequired() *
(status.getFinishTime() - status.getStartTime()));
}
/**
* A taskid assigned to this JobInProgress has reported in successfully.
*/
public synchronized boolean completedTask(TaskInProgress tip,
TaskStatus status)
{
TaskAttemptID taskid = status.getTaskID();
final JobTrackerInstrumentation metrics = jobtracker.getInstrumentation();
// Metering
meterTaskAttempt(tip, status);
// Sanity check: is the TIP already complete?
// This would not happen,
// because no two tasks are SUCCEEDED at the same time.
if (tip.isComplete()) {
// Mark this task as KILLED
tip.alreadyCompletedTask(taskid);
// Let the JobTracker cleanup this taskid if the job isn't running
if (this.status.getRunState() != JobStatus.RUNNING) {
jobtracker.markCompletedTaskAttempt(status.getTaskTracker(), taskid);
}
return false;
}
boolean wasSpeculating = tip.isSpeculating(); //store this fact
LOG.info("Task '" + taskid + "' has completed " + tip.getTIPId() +
" successfully.");
// Mark the TIP as complete
tip.completed(taskid);
resourceEstimator.updateWithCompletedTask(status, tip);
// Update jobhistory
TaskTrackerStatus ttStatus =
this.jobtracker.getTaskTrackerStatus(status.getTaskTracker());
String trackerHostname = jobtracker.getNode(ttStatus.getHost()).toString();
TaskType taskType = getTaskType(tip);
TaskAttemptStartedEvent tse = new TaskAttemptStartedEvent(
status.getTaskID(), taskType, status.getStartTime(),
status.getTaskTracker(), ttStatus.getHttpPort());
jobHistory.logEvent(tse, status.getTaskID().getJobID());
if (status.getIsMap()){
MapAttemptFinishedEvent mfe = new MapAttemptFinishedEvent(
status.getTaskID(), taskType, TaskStatus.State.SUCCEEDED.toString(),
status.getMapFinishTime(),
status.getFinishTime(), trackerHostname,
status.getStateString(),
new org.apache.hadoop.mapreduce.Counters(status.getCounters()));
jobHistory.logEvent(mfe, status.getTaskID().getJobID());
}else{
ReduceAttemptFinishedEvent rfe = new ReduceAttemptFinishedEvent(
status.getTaskID(), taskType, TaskStatus.State.SUCCEEDED.toString(),
status.getShuffleFinishTime(),
status.getSortFinishTime(), status.getFinishTime(),
trackerHostname, status.getStateString(),
new org.apache.hadoop.mapreduce.Counters(status.getCounters()));
jobHistory.logEvent(rfe, status.getTaskID().getJobID());
}
TaskFinishedEvent tfe = new TaskFinishedEvent(tip.getTIPId(),
tip.getExecFinishTime(), taskType,
TaskStatus.State.SUCCEEDED.toString(),
new org.apache.hadoop.mapreduce.Counters(status.getCounters()));
jobHistory.logEvent(tfe, tip.getJob().getJobID());
if (tip.isJobSetupTask()) {
// setup task has finished. kill the extra setup tip
killSetupTip(!tip.isMapTask());
setupComplete();
} else if (tip.isJobCleanupTask()) {
// cleanup task has finished. Kill the extra cleanup tip
if (tip.isMapTask()) {
// kill the reduce tip
cleanup[1].kill();
} else {
cleanup[0].kill();
}
//
// The Job is done
// if the job is failed, then mark the job failed.
if (jobFailed) {
terminateJob(JobStatus.FAILED);
}
// if the job is killed, then mark the job killed.
if (jobKilled) {
terminateJob(JobStatus.KILLED);
}
else {
jobComplete();
}
// The job has been killed/failed/successful
// JobTracker should cleanup this task
jobtracker.markCompletedTaskAttempt(status.getTaskTracker(), taskid);
} else if (tip.isMapTask()) {
runningMapTasks -= 1;
finishedMapTasks += 1;
metrics.completeMap(taskid);
if (!tip.isJobSetupTask() && hasSpeculativeMaps) {
updateTaskTrackerStats(tip,ttStatus,trackerMapStats,mapTaskStats);
}
// remove the completed map from the resp running caches
retireMap(tip);
if ((finishedMapTasks + failedMapTIPs) == (numMapTasks)) {
this.status.setMapProgress(1.0f);
}
} else {
runningReduceTasks -= 1;
finishedReduceTasks += 1;
metrics.completeReduce(taskid);
if (!tip.isJobSetupTask() && hasSpeculativeReduces) {
updateTaskTrackerStats(tip,ttStatus,trackerReduceStats,reduceTaskStats);
}
// remove the completed reduces from the running reducers set
retireReduce(tip);
if ((finishedReduceTasks + failedReduceTIPs) == (numReduceTasks)) {
this.status.setReduceProgress(1.0f);
}
}
decrementSpeculativeCount(wasSpeculating, tip);
// is job complete?
if (!jobSetupCleanupNeeded && canLaunchJobCleanupTask()) {
jobComplete();
}
return true;
}
private void updateTaskTrackerStats(TaskInProgress tip, TaskTrackerStatus ttStatus,
Map<String,DataStatistics> trackerStats, DataStatistics overallStats) {
float tipDuration = tip.getExecFinishTime()-tip.getDispatchTime(tip.getSuccessfulTaskid());
DataStatistics ttStats =
trackerStats.get(ttStatus.getTrackerName());
double oldMean = 0.0d;
//We maintain the mean of TaskTrackers' means. That way, we get a single
//data-point for every tracker (used in the evaluation in isSlowTracker)
if (ttStats != null) {
oldMean = ttStats.mean();
ttStats.add(tipDuration);
overallStats.updateStatistics(oldMean, ttStats.mean());
} else {
trackerStats.put(ttStatus.getTrackerName(),
(ttStats = new DataStatistics(tipDuration)));
overallStats.add(tipDuration);
}
if (LOG.isDebugEnabled()) {
LOG.debug("Added mean of " +ttStats.mean() + " to trackerStats of type "+
(tip.isMapTask() ? "Map" : "Reduce") +
" on "+ttStatus.getTrackerName()+". DataStatistics is now: " +
trackerStats.get(ttStatus.getTrackerName()));
}
}
public void updateStatistics(double oldProg, double newProg, boolean isMap) {
if (isMap) {
runningMapTaskStats.updateStatistics(oldProg, newProg);
} else {
runningReduceTaskStats.updateStatistics(oldProg, newProg);
}
}
public DataStatistics getRunningTaskStatistics(boolean isMap) {
if (isMap) {
return runningMapTaskStats;
} else {
return runningReduceTaskStats;
}
}
public float getSlowTaskThreshold() {
return slowTaskThreshold;
}
/**
* Job state change must happen thru this call
*/
private void changeStateTo(int newState) {
int oldState = this.status.getRunState();
if (oldState == newState) {
return; //old and new states are same
}
this.status.setRunState(newState);
//update the metrics
if (oldState == JobStatus.PREP) {
this.jobtracker.getInstrumentation().decPrepJob(conf, jobId);
} else if (oldState == JobStatus.RUNNING) {
this.jobtracker.getInstrumentation().decRunningJob(conf, jobId);
}
if (newState == JobStatus.PREP) {
this.jobtracker.getInstrumentation().addPrepJob(conf, jobId);
} else if (newState == JobStatus.RUNNING) {
this.jobtracker.getInstrumentation().addRunningJob(conf, jobId);
}
}
/**
* The job is done since all it's component tasks are either
* successful or have failed.
*/
private void jobComplete() {
final JobTrackerInstrumentation metrics = jobtracker.getInstrumentation();
//
// All tasks are complete, then the job is done!
//
if (this.status.getRunState() == JobStatus.RUNNING ||
this.status.getRunState() == JobStatus.PREP) {
changeStateTo(JobStatus.SUCCEEDED);
this.status.setCleanupProgress(1.0f);
if (maps.length == 0) {
this.status.setMapProgress(1.0f);
}
if (reduces.length == 0) {
this.status.setReduceProgress(1.0f);
}
this.finishTime = JobTracker.getClock().getTime();
this.status.setFinishTime(this.finishTime);
LOG.info("Job " + this.status.getJobID() +
" has completed successfully.");
// Log the job summary (this should be done prior to logging to
// job-history to ensure job-counters are in-sync
JobSummary.logJobSummary(this, jobtracker.getClusterStatus(false));
// Log job-history
JobFinishedEvent jfe =
new JobFinishedEvent(this.status.getJobID(),
this.finishTime,
this.finishedMapTasks,this.finishedReduceTasks, failedMapTasks,
failedReduceTasks,
new org.apache.hadoop.mapreduce.Counters(getMapCounters()),
new org.apache.hadoop.mapreduce.Counters(getReduceCounters()),
new org.apache.hadoop.mapreduce.Counters(getCounters()));
jobHistory.logEvent(jfe, this.status.getJobID());
jobHistory.closeWriter(this.status.getJobID());
// Note that finalize will close the job history handles which garbage collect
// might try to finalize
garbageCollect();
metrics.completeJob(this.conf, this.status.getJobID());
}
}
private synchronized void terminateJob(int jobTerminationState) {
if ((status.getRunState() == JobStatus.RUNNING) ||
(status.getRunState() == JobStatus.PREP)) {
this.finishTime = JobTracker.getClock().getTime();
this.status.setMapProgress(1.0f);
this.status.setReduceProgress(1.0f);
this.status.setCleanupProgress(1.0f);
this.status.setFinishTime(this.finishTime);
if (jobTerminationState == JobStatus.FAILED) {
changeStateTo(JobStatus.FAILED);
} else {
changeStateTo(JobStatus.KILLED);
}
// Log the job summary
JobSummary.logJobSummary(this, jobtracker.getClusterStatus(false));
JobUnsuccessfulCompletionEvent failedEvent =
new JobUnsuccessfulCompletionEvent(this.status.getJobID(),
finishTime,
this.finishedMapTasks,
this.finishedReduceTasks,
JobStatus.getJobRunState(jobTerminationState));
jobHistory.logEvent(failedEvent, this.status.getJobID());
jobHistory.closeWriter(this.status.getJobID());
garbageCollect();
jobtracker.getInstrumentation().terminateJob(
this.conf, this.status.getJobID());
if (jobTerminationState == JobStatus.FAILED) {
jobtracker.getInstrumentation().failedJob(
this.conf, this.status.getJobID());
} else {
jobtracker.getInstrumentation().killedJob(
this.conf, this.status.getJobID());
}
}
}
/**
* Terminate the job and all its component tasks.
* Calling this will lead to marking the job as failed/killed. Cleanup
* tip will be launched. If the job has not inited, it will directly call
* terminateJob as there is no need to launch cleanup tip.
* This method is reentrant.
* @param jobTerminationState job termination state
*/
private synchronized void terminate(int jobTerminationState) {
if(!tasksInited.get()) {
//init could not be done, we just terminate directly.
terminateJob(jobTerminationState);
return;
}
if ((status.getRunState() == JobStatus.RUNNING) ||
(status.getRunState() == JobStatus.PREP)) {
LOG.info("Killing job '" + this.status.getJobID() + "'");
if (jobTerminationState == JobStatus.FAILED) {
if(jobFailed) {//reentrant
return;
}
jobFailed = true;
} else if (jobTerminationState == JobStatus.KILLED) {
if(jobKilled) {//reentrant
return;
}
jobKilled = true;
}
// clear all unclean tasks
clearUncleanTasks();
//
// kill all TIPs.
//
for (int i = 0; i < setup.length; i++) {
setup[i].kill();
}
for (int i = 0; i < maps.length; i++) {
maps[i].kill();
}
for (int i = 0; i < reduces.length; i++) {
reduces[i].kill();
}
if (!jobSetupCleanupNeeded) {
terminateJob(jobTerminationState);
}
}
}
/**
* Cancel all reservations since the job is done
*/
private void cancelReservedSlots() {
// Make a copy of the set of TaskTrackers to prevent a
// ConcurrentModificationException ...
Set<TaskTracker> tm =
new HashSet<TaskTracker>(trackersReservedForMaps.keySet());
for (TaskTracker tt : tm) {
tt.unreserveSlots(TaskType.MAP, this);
}
Set<TaskTracker> tr =
new HashSet<TaskTracker>(trackersReservedForReduces.keySet());
for (TaskTracker tt : tr) {
tt.unreserveSlots(TaskType.REDUCE, this);
}
}
private void clearUncleanTasks() {
TaskAttemptID taskid = null;
TaskInProgress tip = null;
while (!mapCleanupTasks.isEmpty()) {
taskid = mapCleanupTasks.remove(0);
tip = maps[taskid.getTaskID().getId()];
updateTaskStatus(tip, tip.getTaskStatus(taskid));
}
while (!reduceCleanupTasks.isEmpty()) {
taskid = reduceCleanupTasks.remove(0);
tip = reduces[taskid.getTaskID().getId()];
updateTaskStatus(tip, tip.getTaskStatus(taskid));
}
}
/**
* Kill the job and all its component tasks. This method should be called from
* jobtracker and should return fast as it locks the jobtracker.
*/
public void kill() {
boolean killNow = false;
synchronized(jobInitKillStatus) {
jobInitKillStatus.killed = true;
//if not in middle of init, terminate it now
if(!jobInitKillStatus.initStarted || jobInitKillStatus.initDone) {
//avoiding nested locking by setting flag
killNow = true;
}
}
if(killNow) {
terminate(JobStatus.KILLED);
}
}
/**
* Fails the job and all its component tasks. This should be called only from
* {@link JobInProgress} or {@link JobTracker}. Look at
* {@link JobTracker#failJob(JobInProgress)} for more details.
* Note that the job doesnt expect itself to be failed before its inited.
* Only when the init is done (successfully or otherwise), the job can be
* failed.
*/
synchronized void fail() {
terminate(JobStatus.FAILED);
}
private void decrementSpeculativeCount(boolean wasSpeculating,
TaskInProgress tip) {
if (wasSpeculating) {
if (tip.isMapTask()) {
speculativeMapTasks--;
if (LOG.isDebugEnabled()) {
LOG.debug("Decremented count for " +
tip.getTIPId()+"/"+tip.getJob().getJobID() +
". Current speculativeMap task count: "
+ speculativeMapTasks);
}
} else {
speculativeReduceTasks--;
if (LOG.isDebugEnabled()) {
LOG.debug("Decremented count for " +
tip.getTIPId()+"/"+tip.getJob().getJobID() +
". Current speculativeReduce task count: "
+ speculativeReduceTasks);
}
}
}
}
/**
* A task assigned to this JobInProgress has reported in as failed.
* Most of the time, we'll just reschedule execution. However, after
* many repeated failures we may instead decide to allow the entire
* job to fail or succeed if the user doesn't care about a few tasks failing.
*
* Even if a task has reported as completed in the past, it might later
* be reported as failed. That's because the TaskTracker that hosts a map
* task might die before the entire job can complete. If that happens,
* we need to schedule reexecution so that downstream reduce tasks can
* obtain the map task's output.
*/
private void failedTask(TaskInProgress tip, TaskAttemptID taskid,
TaskStatus status,
TaskTracker taskTracker, boolean wasRunning,
boolean wasComplete, boolean wasAttemptRunning) {
// check if the TIP is already failed
boolean wasFailed = tip.isFailed();
boolean wasSpeculating = tip.isSpeculating();
// Mark the taskid as FAILED or KILLED
tip.incompleteSubTask(taskid, this.status);
decrementSpeculativeCount(wasSpeculating, tip);
boolean isRunning = tip.isRunning();
boolean isComplete = tip.isComplete();
if(wasAttemptRunning) {
// We are decrementing counters without looking for isRunning ,
// because we increment the counters when we obtain
// new map task attempt or reduce task attempt.We do not really check
// for tip being running.
// Whenever we obtain new task attempt runningMapTasks incremented.
// hence we are decrementing the same.
if(!tip.isJobCleanupTask() && !tip.isJobSetupTask()) {
if(tip.isMapTask()) {
runningMapTasks -= 1;
} else {
runningReduceTasks -= 1;
}
}
// Metering
meterTaskAttempt(tip, status);
}
//update running count on task failure.
if (wasRunning && !isRunning) {
if (tip.isJobCleanupTask()) {
launchedCleanup = false;
} else if (tip.isJobSetupTask()) {
launchedSetup = false;
} else if (tip.isMapTask()) {
// remove from the running queue and put it in the non-running cache
// if the tip is not complete i.e if the tip still needs to be run
if (!isComplete) {
retireMap(tip);
failMap(tip);
}
} else {
// remove from the running queue and put in the failed queue if the tip
// is not complete
if (!isComplete) {
retireReduce(tip);
failReduce(tip);
}
}
}
// The case when the map was complete but the task tracker went down.
// However, we don't need to do any metering here...
if (wasComplete && !isComplete) {
if (tip.isMapTask()) {
// Put the task back in the cache. This will help locality for cases
// where we have a different TaskTracker from the same rack/switch
// asking for a task.
// We bother about only those TIPs that were successful
// earlier (wasComplete and !isComplete)
// (since they might have been removed from the cache of other
// racks/switches, if the input split blocks were present there too)
failMap(tip);
finishedMapTasks -= 1;
}
}
// update job history
// get taskStatus from tip
TaskStatus taskStatus = tip.getTaskStatus(taskid);
String taskTrackerName = taskStatus.getTaskTracker();
String taskTrackerHostName = convertTrackerNameToHostName(taskTrackerName);
int taskTrackerPort = -1;
TaskTrackerStatus taskTrackerStatus =
(taskTracker == null) ? null : taskTracker.getStatus();
if (taskTrackerStatus != null) {
taskTrackerPort = taskTrackerStatus.getHttpPort();
}
long startTime = taskStatus.getStartTime();
long finishTime = taskStatus.getFinishTime();
List<String> taskDiagnosticInfo = tip.getDiagnosticInfo(taskid);
String diagInfo = taskDiagnosticInfo == null ? "" :
StringUtils.arrayToString(taskDiagnosticInfo.toArray(new String[0]));
TaskType taskType = getTaskType(tip);
TaskAttemptStartedEvent tse = new TaskAttemptStartedEvent(
taskid, taskType, startTime, taskTrackerName, taskTrackerPort);
jobHistory.logEvent(tse, taskid.getJobID());
TaskAttemptUnsuccessfulCompletionEvent tue =
new TaskAttemptUnsuccessfulCompletionEvent(taskid,
taskType, taskStatus.getRunState().toString(),
finishTime,
taskTrackerHostName, diagInfo);
jobHistory.logEvent(tue, taskid.getJobID());
// After this, try to assign tasks with the one after this, so that
// the failed task goes to the end of the list.
if (!tip.isJobCleanupTask() && !tip.isJobSetupTask()) {
if (tip.isMapTask()) {
failedMapTasks++;
} else {
failedReduceTasks++;
}
}
//
// Note down that a task has failed on this tasktracker
//
if (status.getRunState() == TaskStatus.State.FAILED) {
addTrackerTaskFailure(taskTrackerName, taskTracker);
}
//
// Let the JobTracker know that this task has failed
//
jobtracker.markCompletedTaskAttempt(status.getTaskTracker(), taskid);
//
// Check if we need to kill the job because of too many failures or
// if the job is complete since all component tasks have completed
// We do it once per TIP and that too for the task that fails the TIP
if (!wasFailed && tip.isFailed()) {
//
// Allow upto 'mapFailuresPercent' of map tasks to fail or
// 'reduceFailuresPercent' of reduce tasks to fail
//
boolean killJob = tip.isJobCleanupTask() || tip.isJobSetupTask() ? true :
tip.isMapTask() ?
((++failedMapTIPs*100) > (mapFailuresPercent*numMapTasks)) :
((++failedReduceTIPs*100) > (reduceFailuresPercent*numReduceTasks));
if (killJob) {
LOG.info("Aborting job " + profile.getJobID());
TaskFailedEvent tfe =
new TaskFailedEvent(tip.getTIPId(), finishTime, taskType, diagInfo,
TaskStatus.State.FAILED.toString(),
null);
jobHistory.logEvent(tfe, tip.getJob().getJobID());
if (tip.isJobCleanupTask()) {
// kill the other tip
if (tip.isMapTask()) {
cleanup[1].kill();
} else {
cleanup[0].kill();
}
terminateJob(JobStatus.FAILED);
} else {
if (tip.isJobSetupTask()) {
// kill the other tip
killSetupTip(!tip.isMapTask());
}
fail();
}
}
//
// Update the counters
//
if (!tip.isJobCleanupTask() && !tip.isJobSetupTask()) {
if (tip.isMapTask()) {
jobCounters.incrCounter(JobCounter.NUM_FAILED_MAPS, 1);
} else {
jobCounters.incrCounter(JobCounter.NUM_FAILED_REDUCES, 1);
}
}
}
}
void killSetupTip(boolean isMap) {
if (isMap) {
setup[0].kill();
} else {
setup[1].kill();
}
}
boolean isSetupFinished() {
// if there is no setup to be launched, consider setup is finished.
if ((tasksInited.get() && setup.length == 0) ||
setup[0].isComplete() || setup[0].isFailed() || setup[1].isComplete()
|| setup[1].isFailed()) {
return true;
}
return false;
}
/**
* Fail a task with a given reason, but without a status object.
*
* Assuming {@link JobTracker} is locked on entry.
*
* @param tip The task's tip
* @param taskid The task id
* @param reason The reason that the task failed
* @param trackerName The task tracker the task failed on
*/
public synchronized void failedTask(TaskInProgress tip, TaskAttemptID taskid,
String reason, TaskStatus.Phase phase, TaskStatus.State state,
String trackerName) {
TaskStatus status = TaskStatus.createTaskStatus(tip.isMapTask(),
taskid,
0.0f,
tip.isMapTask() ?
numSlotsPerMap :
numSlotsPerReduce,
state,
reason,
reason,
trackerName, phase,
new Counters());
// update the actual start-time of the attempt
TaskStatus oldStatus = tip.getTaskStatus(taskid);
long startTime = oldStatus == null
? JobTracker.getClock().getTime()
: oldStatus.getStartTime();
status.setStartTime(startTime);
status.setFinishTime(JobTracker.getClock().getTime());
boolean wasComplete = tip.isComplete();
updateTaskStatus(tip, status);
boolean isComplete = tip.isComplete();
if (wasComplete && !isComplete) { // mark a successful tip as failed
TaskType taskType = getTaskType(tip);
TaskFailedEvent tfe =
new TaskFailedEvent(tip.getTIPId(), tip.getExecFinishTime(), taskType,
reason, TaskStatus.State.FAILED.toString(),
taskid);
jobHistory.logEvent(tfe, tip.getJob().getJobID());
}
}
/**
* The job is dead. We're now GC'ing it, getting rid of the job
* from all tables. Be sure to remove all of this job's tasks
* from the various tables.
*/
void garbageCollect() {
synchronized(this) {
// Cancel task tracker reservation
cancelReservedSlots();
// Let the JobTracker know that a job is complete
jobtracker.getInstrumentation().decWaitingMaps(getJobID(), pendingMaps());
jobtracker.getInstrumentation().decWaitingReduces(getJobID(), pendingReduces());
jobtracker.storeCompletedJob(this);
jobtracker.finalizeJob(this);
try {
// Definitely remove the local-disk copy of the job file
if (localJobFile != null) {
localFs.delete(localJobFile, true);
localJobFile = null;
}
Path tempDir = jobtracker.getSystemDirectoryForJob(getJobID());
new CleanupQueue().addToQueue(new PathDeletionContext(
jobtracker.getFileSystem(), tempDir.toUri().getPath()));
} catch (IOException e) {
LOG.warn("Error cleaning up "+profile.getJobID()+": "+e);
}
// free up the memory used by the data structures
this.nonRunningMapCache = null;
this.runningMapCache = null;
this.nonRunningReduces = null;
this.runningReduces = null;
}
// remove jobs delegation tokens
if(conf.getBoolean(MRJobConfig.JOB_CANCEL_DELEGATION_TOKEN, true)) {
DelegationTokenRenewal.removeDelegationTokenRenewalForJob(jobId);
} // else don't remove it.May be used by spawned tasks
}
/**
* Return the TaskInProgress that matches the tipid.
*/
public synchronized TaskInProgress getTaskInProgress(TaskID tipid) {
if (tipid.getTaskType() == TaskType.MAP) {
// cleanup map tip
if (cleanup.length > 0 && tipid.equals(cleanup[0].getTIPId())) {
return cleanup[0];
}
// setup map tip
if (setup.length > 0 && tipid.equals(setup[0].getTIPId())) {
return setup[0];
}
for (int i = 0; i < maps.length; i++) {
if (tipid.equals(maps[i].getTIPId())){
return maps[i];
}
}
} else {
// cleanup reduce tip
if (cleanup.length > 0 && tipid.equals(cleanup[1].getTIPId())) {
return cleanup[1];
}
// setup reduce tip
if (setup.length > 0 && tipid.equals(setup[1].getTIPId())) {
return setup[1];
}
for (int i = 0; i < reduces.length; i++) {
if (tipid.equals(reduces[i].getTIPId())){
return reduces[i];
}
}
}
return null;
}
/**
* Find the details of someplace where a map has finished
* @param mapId the id of the map
* @return the task status of the completed task
*/
public synchronized TaskStatus findFinishedMap(int mapId) {
TaskInProgress tip = maps[mapId];
if (tip.isComplete()) {
TaskStatus[] statuses = tip.getTaskStatuses();
for(int i=0; i < statuses.length; i++) {
if (statuses[i].getRunState() == TaskStatus.State.SUCCEEDED) {
return statuses[i];
}
}
}
return null;
}
synchronized int getNumTaskCompletionEvents() {
return taskCompletionEvents.size();
}
synchronized public TaskCompletionEvent[] getTaskCompletionEvents(
int fromEventId, int maxEvents) {
TaskCompletionEvent[] events = TaskCompletionEvent.EMPTY_ARRAY;
if (taskCompletionEvents.size() > fromEventId) {
int actualMax = Math.min(maxEvents,
(taskCompletionEvents.size() - fromEventId));
events = taskCompletionEvents.subList(fromEventId, actualMax + fromEventId).toArray(events);
}
return events;
}
synchronized void fetchFailureNotification(TaskInProgress tip,
TaskAttemptID mapTaskId,
String mapTrackerName,
TaskAttemptID reduceTaskId,
String reduceTrackerName) {
Integer fetchFailures = mapTaskIdToFetchFailuresMap.get(mapTaskId);
fetchFailures = (fetchFailures == null) ? 1 : (fetchFailures+1);
mapTaskIdToFetchFailuresMap.put(mapTaskId, fetchFailures);
LOG.info("Failed fetch notification #" + fetchFailures + " for map task: "
+ mapTaskId + " running on tracker: " + mapTrackerName
+ " and reduce task: " + reduceTaskId + " running on tracker: "
+ reduceTrackerName);
float failureRate = (float)fetchFailures / runningReduceTasks;
// declare faulty if fetch-failures >= max-allowed-failures
boolean isMapFaulty = (failureRate >= MAX_ALLOWED_FETCH_FAILURES_PERCENT)
? true
: false;
if (fetchFailures >= MAX_FETCH_FAILURES_NOTIFICATIONS
&& isMapFaulty) {
LOG.info("Too many fetch-failures for output of task: " + mapTaskId
+ " ... killing it");
failedTask(tip, mapTaskId, "Too many fetch-failures",
(tip.isMapTask() ? TaskStatus.Phase.MAP :
TaskStatus.Phase.REDUCE),
TaskStatus.State.FAILED, mapTrackerName);
mapTaskIdToFetchFailuresMap.remove(mapTaskId);
}
}
/**
* @return The JobID of this JobInProgress.
*/
public JobID getJobID() {
return jobId;
}
public synchronized Object getSchedulingInfo() {
return this.schedulingInfo;
}
public synchronized void setSchedulingInfo(Object schedulingInfo) {
this.schedulingInfo = schedulingInfo;
this.status.setSchedulingInfo(schedulingInfo.toString());
}
/**
* To keep track of kill and initTasks status of this job. initTasks() take
* a lock on JobInProgress object. kill should avoid waiting on
* JobInProgress lock since it may take a while to do initTasks().
*/
private static class JobInitKillStatus {
//flag to be set if kill is called
boolean killed;
boolean initStarted;
boolean initDone;
}
boolean isComplete() {
return status.isJobComplete();
}
/**
* Get the task type for logging it to {@link JobHistory}.
*/
private TaskType getTaskType(TaskInProgress tip) {
if (tip.isJobCleanupTask()) {
return TaskType.JOB_CLEANUP;
} else if (tip.isJobSetupTask()) {
return TaskType.JOB_SETUP;
} else if (tip.isMapTask()) {
return TaskType.MAP;
} else {
return TaskType.REDUCE;
}
}
/**
* Get the level of locality that a given task would have if launched on
* a particular TaskTracker. Returns 0 if the task has data on that machine,
* 1 if it has data on the same rack, etc (depending on number of levels in
* the network hierarchy).
*/
int getLocalityLevel(TaskInProgress tip, TaskTrackerStatus tts) {
Node tracker = jobtracker.getNode(tts.getHost());
int level = this.maxLevel;
// find the right level across split locations
for (String local : maps[tip.getIdWithinJob()].getSplitLocations()) {
Node datanode = jobtracker.getNode(local);
int newLevel = this.maxLevel;
if (tracker != null && datanode != null) {
newLevel = getMatchingLevelForNodes(tracker, datanode);
}
if (newLevel < level) {
level = newLevel;
// an optimization
if (level == 0) {
break;
}
}
}
return level;
}
/**
* Test method to set the cluster sizes
*/
void setClusterSize(int clusterSize) {
this.clusterSize = clusterSize;
}
static class JobSummary {
static final Log LOG = LogFactory.getLog(JobSummary.class);
// Escape sequences
static final char EQUALS = '=';
static final char[] charsToEscape =
{StringUtils.COMMA, EQUALS, StringUtils.ESCAPE_CHAR};
static class SummaryBuilder {
final StringBuilder buffer = new StringBuilder();
// A little optimization for a very common case
SummaryBuilder add(String key, long value) {
return _add(key, Long.toString(value));
}
<T> SummaryBuilder add(String key, T value) {
return _add(key, StringUtils.escapeString(String.valueOf(value),
StringUtils.ESCAPE_CHAR, charsToEscape));
}
SummaryBuilder add(SummaryBuilder summary) {
if (buffer.length() > 0) buffer.append(StringUtils.COMMA);
buffer.append(summary.buffer);
return this;
}
SummaryBuilder _add(String key, String value) {
if (buffer.length() > 0) buffer.append(StringUtils.COMMA);
buffer.append(key).append(EQUALS).append(value);
return this;
}
@Override public String toString() {
return buffer.toString();
}
}
static SummaryBuilder getTaskLaunchTimesSummary(JobInProgress job) {
SummaryBuilder summary = new SummaryBuilder();
Map<TaskType, Long> timeMap = job.getFirstTaskLaunchTimes();
synchronized(timeMap) {
for (Map.Entry<TaskType, Long> e : timeMap.entrySet()) {
summary.add("first"+ StringUtils.camelize(e.getKey().name()) +
"TaskLaunchTime", e.getValue().longValue());
}
}
return summary;
}
/**
* Log a summary of the job's runtime.
*
* @param job {@link JobInProgress} whose summary is to be logged, cannot
* be <code>null</code>.
* @param cluster {@link ClusterStatus} of the cluster on which the job was
* run, cannot be <code>null</code>
*/
public static void logJobSummary(JobInProgress job, ClusterStatus cluster) {
JobStatus status = job.getStatus();
JobProfile profile = job.getProfile();
Counters jobCounters = job.getJobCounters();
long mapSlotSeconds =
(jobCounters.getCounter(JobCounter.SLOTS_MILLIS_MAPS) +
jobCounters.getCounter(JobCounter.FALLOW_SLOTS_MILLIS_MAPS)) / 1000;
long reduceSlotSeconds =
(jobCounters.getCounter(JobCounter.SLOTS_MILLIS_REDUCES) +
jobCounters.getCounter(JobCounter.FALLOW_SLOTS_MILLIS_REDUCES)) / 1000;
SummaryBuilder summary = new SummaryBuilder()
.add("jobId", job.getJobID())
.add("submitTime", job.getStartTime())
.add("launchTime", job.getLaunchTime())
.add(getTaskLaunchTimesSummary(job))
.add("finishTime", job.getFinishTime())
.add("numMaps", job.getTasks(TaskType.MAP).length)
.add("numSlotsPerMap", job.getNumSlotsPerMap())
.add("numReduces", job.getTasks(TaskType.REDUCE).length)
.add("numSlotsPerReduce", job.getNumSlotsPerReduce())
.add("user", profile.getUser())
.add("queue", profile.getQueueName())
.add("status", JobStatus.getJobRunState(status.getRunState()))
.add("mapSlotSeconds", mapSlotSeconds)
.add("reduceSlotsSeconds", reduceSlotSeconds)
.add("clusterMapCapacity", cluster.getMaxMapTasks())
.add("clusterReduceCapacity", cluster.getMaxReduceTasks());
LOG.info(summary);
}
}
/**
* Creates the localized copy of job conf
* @param jobConf
* @param id
*/
void setUpLocalizedJobConf(JobConf jobConf,
org.apache.hadoop.mapreduce.JobID id) {
String localJobFilePath = jobtracker.getLocalJobFilePath(id);
File localJobFile = new File(localJobFilePath);
FileOutputStream jobOut = null;
try {
jobOut = new FileOutputStream(localJobFile);
jobConf.writeXml(jobOut);
if (LOG.isDebugEnabled()) {
LOG.debug("Job conf for " + id + " stored at "
+ localJobFile.getAbsolutePath());
}
} catch (IOException ioe) {
LOG.error("Failed to store job conf on the local filesystem ", ioe);
} finally {
if (jobOut != null) {
try {
jobOut.close();
} catch (IOException ie) {
LOG.info("Failed to close the job configuration file "
+ StringUtils.stringifyException(ie));
}
}
}
}
/**
* Deletes localized copy of job conf
*/
void cleanupLocalizedJobConf(org.apache.hadoop.mapreduce.JobID id) {
String localJobFilePath = jobtracker.getLocalJobFilePath(id);
File f = new File (localJobFilePath);
LOG.info("Deleting localized job conf at " + f);
if (!f.delete()) {
if (LOG.isDebugEnabled()) {
LOG.debug("Failed to delete file " + f);
}
}
}
/**
* generate job token and save it into the file
* @throws IOException
*/
private void generateAndStoreTokens() throws IOException{
Path jobDir = jobtracker.getSystemDirectoryForJob(jobId);
Path keysFile = new Path(jobDir, TokenCache.JOB_TOKEN_HDFS_FILE);
if (tokenStorage == null) {
tokenStorage = new Credentials();
}
//create JobToken file and write token to it
JobTokenIdentifier identifier = new JobTokenIdentifier(new Text(jobId
.toString()));
Token<JobTokenIdentifier> token = new Token<JobTokenIdentifier>(identifier,
jobtracker.getJobTokenSecretManager());
token.setService(identifier.getJobId());
TokenCache.setJobToken(token, tokenStorage);
// write TokenStorage out
tokenStorage.writeTokenStorageFile(keysFile, jobtracker.getConf());
LOG.info("jobToken generated and stored with users keys in "
+ keysFile.toUri().getPath());
}
public String getJobSubmitHostAddress() {
return submitHostAddress;
}
public String getJobSubmitHostName() {
return submitHostName;
}
}