/**
* Unregisters a finished or aborted task.
*
* @param id the ID of the task to be unregistered
*/
private void unregisterTask(final ExecutionVertexID id) {
// Task de-registration must be atomic
synchronized (this) {
final Task task = this.runningTasks.remove(id);
if (task == null) {
LOG.error("Cannot find task with ID " + id + " to unregister");
return;
}
// remove the local tmp file for unregistered tasks.
for (Entry<String, DistributedCacheEntry> e :
DistributedCache.readFileInfoFromConfig(task.getEnvironment().getJobConfiguration())) {
this.fileCache.deleteTmpFile(e.getKey(), e.getValue(), task.getJobID());
}
// Unregister task from the byte buffered channel manager
this.channelManager.unregister(id, task);
// Unregister task from profiling
task.unregisterProfiler(this.profiler);
// Unregister task from memory manager
task.unregisterMemoryManager(this.memoryManager);
// Unregister task from library cache manager
try {
LibraryCacheManager.unregister(task.getJobID());
} catch (IOException e) {
if (LOG.isDebugEnabled()) {
LOG.debug("Unregistering the job vertex ID " + id + " caused an IOException");
}
}
}
}
@SuppressWarnings("deprecation")
public CoronaJobTrackerRunner(
TaskTracker.TaskInProgress tip,
Task task,
TaskTracker tracker,
JobConf ttConf,
CoronaSessionInfo info,
String originalPath,
String releasePath)
throws IOException {
super(tip, task, tracker, ttConf);
this.coronaSessionInfo = info;
this.originalPath = originalPath;
this.releasePath = releasePath;
LocalDirAllocator lDirAlloc = new LocalDirAllocator("mapred.local.dir");
workDir =
new File(
lDirAlloc
.getLocalPathForWrite(
TaskTracker.getLocalTaskDir(
task.getJobID().toString(),
task.getTaskID().toString(),
task.isTaskCleanupTask())
+ Path.SEPARATOR
+ MRConstants.WORKDIR,
conf)
.toString());
if (!workDir.mkdirs()) {
if (!workDir.isDirectory()) {
throw new IOException("Mkdirs failed to create " + workDir.toString());
}
}
localizeTaskConfiguration(tracker, ttConf, workDir.toString(), task, task.getJobID());
}
public TaskRunner(TaskTracker.TaskInProgress tip, TaskTracker tracker, JobConf conf) {
this.tip = tip;
this.t = tip.getTask();
this.tracker = tracker;
this.conf = conf;
this.mapOutputFile = new MapOutputFile(t.getJobID());
this.mapOutputFile.setConf(conf);
this.jvmManager = tracker.getJvmManagerInstance();
}
/**
* Write the child's configuration to the disk and set it in configuration so that the child can
* pick it up from there.
*
* @param lDirAlloc
* @throws IOException
*/
void setupChildTaskConfiguration(LocalDirAllocator lDirAlloc) throws IOException {
Path localTaskFile =
lDirAlloc.getLocalPathForWrite(
TaskTracker.getTaskConfFile(
t.getUser(),
t.getJobID().toString(),
t.getTaskID().toString(),
t.isTaskCleanupTask()),
conf);
// write the child's task configuration file to the local disk
JobLocalizer.writeLocalJobFile(localTaskFile, conf);
// Set the final job file in the task. The child needs to know the correct
// path to job.xml. So set this path accordingly.
t.setJobFile(localTaskFile.toString());
}
/**
* Prepare the mapred.local.dir for the child. The child is sand-boxed now. Whenever it uses
* LocalDirAllocator from now on inside the child, it will only see files inside the
* attempt-directory. This is done in the Child's process space.
*/
static void setupChildMapredLocalDirs(Task t, JobConf conf) {
String[] localDirs = conf.getTrimmedStrings(JobConf.MAPRED_LOCAL_DIR_PROPERTY);
String jobId = t.getJobID().toString();
String taskId = t.getTaskID().toString();
boolean isCleanup = t.isTaskCleanupTask();
String user = t.getUser();
StringBuffer childMapredLocalDir =
new StringBuffer(
localDirs[0]
+ Path.SEPARATOR
+ TaskTracker.getLocalTaskDir(user, jobId, taskId, isCleanup));
for (int i = 1; i < localDirs.length; i++) {
childMapredLocalDir.append(
","
+ localDirs[i]
+ Path.SEPARATOR
+ TaskTracker.getLocalTaskDir(user, jobId, taskId, isCleanup));
}
LOG.debug("mapred.local.dir for child : " + childMapredLocalDir);
conf.set("mapred.local.dir", childMapredLocalDir.toString());
}
@SuppressWarnings("deprecation")
@Override
public void run() {
Task task = getTask();
TaskAttemptID taskid = task.getTaskID();
try {
if (!prepare()) {
return;
}
String sep = System.getProperty("path.separator");
StringBuffer classPath = new StringBuffer();
// The alternate runtime can be used to debug tasks by putting a
// custom version of the mapred libraries. This will get loaded before
// the TT's jars.
String debugRuntime = conf.get("mapred.task.debug.runtime.classpath");
if (debugRuntime != null) {
classPath.append(debugRuntime);
classPath.append(sep);
}
// start with same classpath as parent process
String systemClassPath = System.getProperty("java.class.path");
if (releasePath != null
&& !releasePath.isEmpty()
&& originalPath != null
&& !releasePath.isEmpty()) {
systemClassPath = systemClassPath.replaceAll(originalPath, releasePath);
}
classPath.append(systemClassPath);
classPath.append(sep);
// Build exec child jmv args.
Vector<String> vargs = new Vector<String>(8);
File jvm = // use same jvm as parent
new File(new File(System.getProperty("java.home"), "bin"), "java");
vargs.add(jvm.toString());
// Add child (task) java-vm options.
String javaOpts = getCJTJavaOpts(conf);
javaOpts = javaOpts.replace("@taskid@", taskid.toString());
String[] javaOptsSplit = javaOpts.split(" ");
for (int i = 0; i < javaOptsSplit.length; i++) {
vargs.add(javaOptsSplit[i]);
}
// add java.io.tmpdir given by mapred.child.tmp
String tmp = conf.get("mapred.child.tmp", "./tmp");
Path tmpDir = new Path(tmp);
// if temp directory path is not absolute
// prepend it with workDir.
if (!tmpDir.isAbsolute()) {
tmpDir = new Path(workDir.toString(), tmp);
}
FileSystem localFs = FileSystem.getLocal(conf);
if (!localFs.mkdirs(tmpDir) && !localFs.getFileStatus(tmpDir).isDir()) {
throw new IOException("Mkdirs failed to create " + tmpDir.toString());
}
vargs.add("-Djava.io.tmpdir=" + tmpDir.toString());
// Add classpath.
vargs.add("-classpath");
vargs.add(classPath.toString());
// Setup the log4j prop
long logSize = TaskLog.getTaskLogLength(conf);
vargs.add("-Dhadoop.log.dir=" + CoronaTaskTracker.jobTrackerLogDir());
boolean logToScribe = conf.getBoolean("mapred.task.log.scribe", false);
if (logToScribe) {
vargs.addAll(conf.getStringCollection("mapred.task.log.scribe.conf"));
}
String logger = logToScribe ? "INFO,TLA,scribe" : "INFO,TLA";
vargs.add("-Dhadoop.root.logger=" + logger);
vargs.add("-Dhadoop.tasklog.taskid=" + taskid);
vargs.add("-Dhadoop.tasklog.totalLogFileSize=" + logSize);
Path systemDirectory = tracker.systemDirectory;
if (!systemDirectory.isAbsolute()) {
systemDirectory = new Path(tracker.systemFS.getWorkingDirectory(), systemDirectory);
}
systemDirectory = systemDirectory.makeQualified(tracker.systemFS);
vargs.add("-Dmapred.system.dir=" + systemDirectory);
// Add main class and its arguments
vargs.add(CoronaJobTracker.class.getName()); // main of CJT
vargs.add(task.getJobID().toString()); // Pass job id.
vargs.add(task.getTaskID().toString()); // Pass attempt id.
vargs.add(coronaSessionInfo.getJobTrackerAddr().getHostName());
vargs.add(Integer.toString(coronaSessionInfo.getJobTrackerAddr().getPort()));
tracker.addToMemoryManager(task.getTaskID(), task.isMapTask(), conf);
// set memory limit using ulimit if feasible and necessary ...
String[] ulimitCmd = Shell.getUlimitMemoryCommand(getChildUlimit(conf));
List<String> setup = null;
if (ulimitCmd != null) {
setup = new ArrayList<String>();
for (String arg : ulimitCmd) {
setup.add(arg);
}
}
// Set up the redirection of the task's stdout and stderr streams
File stdout = TaskLog.getTaskLogFile(taskid, TaskLog.LogName.STDOUT);
File stderr = TaskLog.getTaskLogFile(taskid, TaskLog.LogName.STDERR);
stdout.getParentFile().mkdirs();
Map<String, String> env = new HashMap<String, String>();
StringBuffer ldLibraryPath = new StringBuffer();
ldLibraryPath.append(workDir.toString());
String oldLdLibraryPath = null;
oldLdLibraryPath = System.getenv("LD_LIBRARY_PATH");
if (oldLdLibraryPath != null) {
ldLibraryPath.append(sep);
ldLibraryPath.append(oldLdLibraryPath);
}
env.put("LD_LIBRARY_PATH", ldLibraryPath.toString());
LOG.info(
"Launching CJT "
+ taskid
+ " in working directory "
+ workDir
+ " Command line "
+ vargs);
jvmManager.launchJvm(
this,
jvmManager.constructJvmEnv(setup, vargs, stdout, stderr, logSize, workDir, env, conf));
synchronized (lock) {
while (!done) {
lock.wait();
}
}
} catch (IOException e) {
LOG.error("Error while launching CJT ", e);
} catch (InterruptedException e) {
LOG.warn("Error while launching CJT ", e);
}
}
@Override
public final void run() {
try {
// before preparing the job localize
// all the archives
TaskAttemptID taskid = t.getTaskID();
LocalDirAllocator lDirAlloc = new LocalDirAllocator("mapred.local.dir");
File jobCacheDir = null;
if (conf.getJar() != null) {
jobCacheDir = new File(new Path(conf.getJar()).getParent().toString());
}
File workDir =
new File(
lDirAlloc
.getLocalPathToRead(
TaskTracker.getJobCacheSubdir()
+ Path.SEPARATOR
+ t.getJobID()
+ Path.SEPARATOR
+ t.getTaskID()
+ Path.SEPARATOR
+ MRConstants.WORKDIR,
conf)
.toString());
URI[] archives = DistributedCache.getCacheArchives(conf);
URI[] files = DistributedCache.getCacheFiles(conf);
FileStatus fileStatus;
FileSystem fileSystem;
Path localPath;
String baseDir;
if ((archives != null) || (files != null)) {
if (archives != null) {
String[] archivesTimestamps = DistributedCache.getArchiveTimestamps(conf);
Path[] p = new Path[archives.length];
for (int i = 0; i < archives.length; i++) {
fileSystem = FileSystem.get(archives[i], conf);
fileStatus = fileSystem.getFileStatus(new Path(archives[i].getPath()));
String cacheId = DistributedCache.makeRelative(archives[i], conf);
String cachePath = TaskTracker.getCacheSubdir() + Path.SEPARATOR + cacheId;
if (lDirAlloc.ifExists(cachePath, conf)) {
localPath = lDirAlloc.getLocalPathToRead(cachePath, conf);
} else {
localPath = lDirAlloc.getLocalPathForWrite(cachePath, fileStatus.getLen(), conf);
}
baseDir = localPath.toString().replace(cacheId, "");
p[i] =
DistributedCache.getLocalCache(
archives[i],
conf,
new Path(baseDir),
fileStatus,
true,
Long.parseLong(archivesTimestamps[i]),
new Path(workDir.getAbsolutePath()),
false);
}
DistributedCache.setLocalArchives(conf, stringifyPathArray(p));
}
if ((files != null)) {
String[] fileTimestamps = DistributedCache.getFileTimestamps(conf);
Path[] p = new Path[files.length];
for (int i = 0; i < files.length; i++) {
fileSystem = FileSystem.get(files[i], conf);
fileStatus = fileSystem.getFileStatus(new Path(files[i].getPath()));
String cacheId = DistributedCache.makeRelative(files[i], conf);
String cachePath = TaskTracker.getCacheSubdir() + Path.SEPARATOR + cacheId;
if (lDirAlloc.ifExists(cachePath, conf)) {
localPath = lDirAlloc.getLocalPathToRead(cachePath, conf);
} else {
localPath = lDirAlloc.getLocalPathForWrite(cachePath, fileStatus.getLen(), conf);
}
baseDir = localPath.toString().replace(cacheId, "");
p[i] =
DistributedCache.getLocalCache(
files[i],
conf,
new Path(baseDir),
fileStatus,
false,
Long.parseLong(fileTimestamps[i]),
new Path(workDir.getAbsolutePath()),
false);
}
DistributedCache.setLocalFiles(conf, stringifyPathArray(p));
}
Path localTaskFile = new Path(t.getJobFile());
FileSystem localFs = FileSystem.getLocal(conf);
localFs.delete(localTaskFile, true);
OutputStream out = localFs.create(localTaskFile);
try {
conf.writeXml(out);
} finally {
out.close();
}
}
if (!prepare()) {
return;
}
String sep = System.getProperty("path.separator");
StringBuffer classPath = new StringBuffer();
// start with same classpath as parent process
classPath.append(System.getProperty("java.class.path"));
classPath.append(sep);
if (!workDir.mkdirs()) {
if (!workDir.isDirectory()) {
LOG.fatal("Mkdirs failed to create " + workDir.toString());
}
}
String jar = conf.getJar();
if (jar != null) {
// if jar exists, it into workDir
File[] libs = new File(jobCacheDir, "lib").listFiles();
if (libs != null) {
for (int i = 0; i < libs.length; i++) {
classPath.append(sep); // add libs from jar to classpath
classPath.append(libs[i]);
}
}
classPath.append(sep);
classPath.append(new File(jobCacheDir, "classes"));
classPath.append(sep);
classPath.append(jobCacheDir);
}
// include the user specified classpath
// archive paths
Path[] archiveClasspaths = DistributedCache.getArchiveClassPaths(conf);
if (archiveClasspaths != null && archives != null) {
Path[] localArchives = DistributedCache.getLocalCacheArchives(conf);
if (localArchives != null) {
for (int i = 0; i < archives.length; i++) {
for (int j = 0; j < archiveClasspaths.length; j++) {
if (archives[i].getPath().equals(archiveClasspaths[j].toString())) {
classPath.append(sep);
classPath.append(localArchives[i].toString());
}
}
}
}
}
// file paths
Path[] fileClasspaths = DistributedCache.getFileClassPaths(conf);
if (fileClasspaths != null && files != null) {
Path[] localFiles = DistributedCache.getLocalCacheFiles(conf);
if (localFiles != null) {
for (int i = 0; i < files.length; i++) {
for (int j = 0; j < fileClasspaths.length; j++) {
if (files[i].getPath().equals(fileClasspaths[j].toString())) {
classPath.append(sep);
classPath.append(localFiles[i].toString());
}
}
}
}
}
classPath.append(sep);
classPath.append(workDir);
// Build exec child jmv args.
Vector<String> vargs = new Vector<String>(8);
File jvm = // use same jvm as parent
new File(new File(System.getProperty("java.home"), "bin"), "java");
vargs.add(jvm.toString());
// Add child (task) java-vm options.
//
// The following symbols if present in mapred.child.java.opts value are
// replaced:
// + @taskid@ is interpolated with value of TaskID.
// Other occurrences of @ will not be altered.
//
// Example with multiple arguments and substitutions, showing
// jvm GC logging, and start of a passwordless JVM JMX agent so can
// connect with jconsole and the likes to watch child memory, threads
// and get thread dumps.
//
// <property>
// <name>mapred.child.java.opts</name>
// <value>-verbose:gc -Xloggc:/tmp/@[email protected] \
// -Dcom.sun.management.jmxremote.authenticate=false \
// -Dcom.sun.management.jmxremote.ssl=false \
// </value>
// </property>
//
String javaOpts = conf.get("mapred.child.java.opts", "-Xmx200m");
javaOpts = javaOpts.replace("@taskid@", taskid.toString());
String[] javaOptsSplit = javaOpts.split(" ");
// Add java.library.path; necessary for loading native libraries.
//
// 1. To support native-hadoop library i.e. libhadoop.so, we add the
// parent processes' java.library.path to the child.
// 2. We also add the 'cwd' of the task to it's java.library.path to help
// users distribute native libraries via the DistributedCache.
// 3. The user can also specify extra paths to be added to the
// java.library.path via mapred.child.java.opts.
//
String libraryPath = System.getProperty("java.library.path");
if (libraryPath == null) {
libraryPath = workDir.getAbsolutePath();
} else {
libraryPath += sep + workDir;
}
boolean hasUserLDPath = false;
for (int i = 0; i < javaOptsSplit.length; i++) {
if (javaOptsSplit[i].startsWith("-Djava.library.path=")) {
javaOptsSplit[i] += sep + libraryPath;
hasUserLDPath = true;
break;
}
}
if (!hasUserLDPath) {
vargs.add("-Djava.library.path=" + libraryPath);
}
for (int i = 0; i < javaOptsSplit.length; i++) {
vargs.add(javaOptsSplit[i]);
}
// add java.io.tmpdir given by mapred.child.tmp
String tmp = conf.get("mapred.child.tmp", "./tmp");
Path tmpDir = new Path(tmp);
// if temp directory path is not absolute
// prepend it with workDir.
if (!tmpDir.isAbsolute()) {
tmpDir = new Path(workDir.toString(), tmp);
}
FileSystem localFs = FileSystem.getLocal(conf);
if (!localFs.mkdirs(tmpDir) && !localFs.getFileStatus(tmpDir).isDir()) {
throw new IOException("Mkdirs failed to create " + tmpDir.toString());
}
vargs.add("-Djava.io.tmpdir=" + tmpDir.toString());
// Add classpath.
vargs.add("-classpath");
vargs.add(classPath.toString());
// Setup the log4j prop
long logSize = TaskLog.getTaskLogLength(conf);
vargs.add(
"-Dhadoop.log.dir=" + new File(System.getProperty("hadoop.log.dir")).getAbsolutePath());
vargs.add("-Dhadoop.root.logger=INFO,TLA");
vargs.add("-Dhadoop.tasklog.taskid=" + taskid);
vargs.add("-Dhadoop.tasklog.totalLogFileSize=" + logSize);
if (conf.getProfileEnabled()) {
if (conf.getProfileTaskRange(t.isMapTask()).isIncluded(t.getPartition())) {
File prof = TaskLog.getTaskLogFile(taskid, TaskLog.LogName.PROFILE);
vargs.add(String.format(conf.getProfileParams(), prof.toString()));
}
}
// Add main class and its arguments
vargs.add(Child.class.getName()); // main of Child
// pass umbilical address
InetSocketAddress address = tracker.getTaskTrackerReportAddress();
vargs.add(address.getAddress().getHostAddress());
vargs.add(Integer.toString(address.getPort()));
vargs.add(taskid.toString()); // pass task identifier
String pidFile = null;
if (tracker.isTaskMemoryManagerEnabled()) {
pidFile =
lDirAlloc
.getLocalPathForWrite(
(TaskTracker.getPidFilesSubdir() + Path.SEPARATOR + taskid), this.conf)
.toString();
}
// set memory limit using ulimit if feasible and necessary ...
String[] ulimitCmd = Shell.getUlimitMemoryCommand(conf);
List<String> setup = null;
if (ulimitCmd != null) {
setup = new ArrayList<String>();
for (String arg : ulimitCmd) {
setup.add(arg);
}
}
// Set up the redirection of the task's stdout and stderr streams
File stdout = TaskLog.getTaskLogFile(taskid, TaskLog.LogName.STDOUT);
File stderr = TaskLog.getTaskLogFile(taskid, TaskLog.LogName.STDERR);
stdout.getParentFile().mkdirs();
tracker.getTaskTrackerInstrumentation().reportTaskLaunch(taskid, stdout, stderr);
Map<String, String> env = new HashMap<String, String>();
StringBuffer ldLibraryPath = new StringBuffer();
ldLibraryPath.append(workDir.toString());
String oldLdLibraryPath = null;
oldLdLibraryPath = System.getenv("LD_LIBRARY_PATH");
if (oldLdLibraryPath != null) {
ldLibraryPath.append(sep);
ldLibraryPath.append(oldLdLibraryPath);
}
env.put("LD_LIBRARY_PATH", ldLibraryPath.toString());
jvmManager.launchJvm(
this,
jvmManager.constructJvmEnv(
setup, vargs, stdout, stderr, logSize, workDir, env, pidFile, conf));
synchronized (lock) {
while (!done) {
lock.wait();
}
}
tracker.getTaskTrackerInstrumentation().reportTaskEnd(t.getTaskID());
if (exitCodeSet) {
if (!killed && exitCode != 0) {
if (exitCode == 65) {
tracker.getTaskTrackerInstrumentation().taskFailedPing(t.getTaskID());
}
throw new IOException("Task process exit with nonzero status of " + exitCode + ".");
}
}
} catch (FSError e) {
LOG.fatal("FSError", e);
try {
tracker.fsError(t.getTaskID(), e.getMessage());
} catch (IOException ie) {
LOG.fatal(t.getTaskID() + " reporting FSError", ie);
}
} catch (Throwable throwable) {
LOG.warn(t.getTaskID() + " Child Error", throwable);
ByteArrayOutputStream baos = new ByteArrayOutputStream();
throwable.printStackTrace(new PrintStream(baos));
try {
tracker.reportDiagnosticInfo(t.getTaskID(), baos.toString());
} catch (IOException e) {
LOG.warn(t.getTaskID() + " Reporting Diagnostics", e);
}
} finally {
try {
URI[] archives = DistributedCache.getCacheArchives(conf);
URI[] files = DistributedCache.getCacheFiles(conf);
if (archives != null) {
for (int i = 0; i < archives.length; i++) {
DistributedCache.releaseCache(archives[i], conf);
}
}
if (files != null) {
for (int i = 0; i < files.length; i++) {
DistributedCache.releaseCache(files[i], conf);
}
}
} catch (IOException ie) {
LOG.warn("Error releasing caches : Cache files might not have been cleaned up");
}
tracker.reportTaskFinished(t.getTaskID(), false);
if (t.isMapTask()) {
tracker.addFreeMapSlot();
} else {
tracker.addFreeReduceSlot();
}
}
}
