org.apache.hadoop.mapred.TaskController Maven / Gradle / Ivy
The newest version!
/**
* 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.IOException;
import java.util.List;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.conf.Configurable;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.fs.permission.FsPermission;
import org.apache.hadoop.mapred.CleanupQueue.PathDeletionContext;
import org.apache.hadoop.mapred.JvmManager.JvmEnv;
import org.apache.hadoop.mapreduce.MRConfig;
import org.apache.hadoop.util.DiskChecker;
import org.apache.hadoop.util.StringUtils;
import org.apache.hadoop.util.Shell.ShellCommandExecutor;
import org.apache.hadoop.classification.InterfaceAudience;
import org.apache.hadoop.classification.InterfaceStability;
/**
* Controls initialization, finalization and clean up of tasks, and
* also the launching and killing of task JVMs.
*
* This class defines the API for initializing, finalizing and cleaning
* up of tasks, as also the launching and killing task JVMs.
* Subclasses of this class will implement the logic required for
* performing the actual actions.
*
*
*/
@InterfaceAudience.Private
public abstract class TaskController implements Configurable {
private Configuration conf;
public static final Log LOG = LogFactory.getLog(TaskController.class);
public Configuration getConf() {
return conf;
}
// The list of directory paths specified in the variable Configs.LOCAL_DIR
// This is used to determine which among the list of directories is picked up
// for storing data for a particular task.
protected String[] mapredLocalDirs;
public void setConf(Configuration conf) {
this.conf = conf;
mapredLocalDirs = conf.getTrimmedStrings(MRConfig.LOCAL_DIR);
}
/**
* Sets up the permissions of the following directories on all the configured
* disks:
*
* - mapreduce.cluster.local.directories
* - Hadoop log directories
*
*/
public void setup() throws IOException {
FileSystem localFs = FileSystem.getLocal(conf);
for (String localDir : this.mapredLocalDirs) {
// Set up the mapreduce.cluster.local.directories.
File mapredlocalDir = new File(localDir);
if (!mapredlocalDir.isDirectory() && !mapredlocalDir.mkdirs()) {
LOG.warn("Unable to create mapreduce.cluster.local.directory : "
+ mapredlocalDir.getPath());
} else {
localFs.setPermission(new Path(mapredlocalDir.getCanonicalPath()),
new FsPermission((short)0755));
}
}
// Set up the user log directory
File taskLog = TaskLog.getUserLogDir();
if (!taskLog.isDirectory() && !taskLog.mkdirs()) {
LOG.warn("Unable to create taskLog directory : " + taskLog.getPath());
} else {
localFs.setPermission(new Path(taskLog.getCanonicalPath()),
new FsPermission((short)0755));
}
DiskChecker.checkDir(TaskLog.getUserLogDir());
}
/**
* Take task-controller specific actions to initialize job. This involves
* setting appropriate permissions to job-files so as to secure the files to
* be accessible only by the user's tasks.
*
* @throws IOException
*/
abstract void initializeJob(JobInitializationContext context) throws IOException;
/**
* Take task-controller specific actions to initialize the distributed cache
* file. This involves setting appropriate permissions for these files so as
* to secure them to be accessible only their owners.
*
* @param context
* @throws IOException
*/
public abstract void initializeDistributedCacheFile(DistributedCacheFileContext context)
throws IOException;
/**
* Launch a task JVM
*
* This method defines how a JVM will be launched to run a task. Each
* task-controller should also do an
* {@link #initializeTask(TaskControllerContext)} inside this method so as to
* initialize the task before launching it. This is for reasons of
* task-controller specific optimizations w.r.t combining initialization and
* launching of tasks.
*
* @param context the context associated to the task
*/
abstract void launchTaskJVM(TaskControllerContext context)
throws IOException;
/**
* Top level cleanup a task JVM method.
*
* - Sends a graceful termiante signal to task JVM to allow subprocesses
* to cleanup.
* - Sends a forceful kill signal to task JVM, terminating all its
* sub-processes forcefully.
*
*
* @param context the task for which kill signal has to be sent.
*/
final void destroyTaskJVM(TaskControllerContext context) {
// Send SIGTERM to try to ask for a polite exit.
terminateTask(context);
try {
Thread.sleep(context.sleeptimeBeforeSigkill);
} catch (InterruptedException e) {
LOG.warn("Sleep interrupted : " +
StringUtils.stringifyException(e));
}
killTask(context);
}
/** Perform initializing actions required before a task can run.
*
* For instance, this method can be used to setup appropriate
* access permissions for files and directories that will be
* used by tasks. Tasks use the job cache, log, and distributed cache
* directories and files as part of their functioning. Typically,
* these files are shared between the daemon and the tasks
* themselves. So, a TaskController that is launching tasks
* as different users can implement this method to setup
* appropriate ownership and permissions for these directories
* and files.
*/
abstract void initializeTask(TaskControllerContext context)
throws IOException;
static class TaskExecContext {
// task being executed
Task task;
}
/**
* Contains task information required for the task controller.
*/
static class TaskControllerContext extends TaskExecContext {
ShellCommandExecutor shExec; // the Shell executor executing the JVM for this task.
// Information used only when this context is used for launching new tasks.
JvmEnv env; // the JVM environment for the task.
// Information used only when this context is used for destroying a task jvm.
String pid; // process handle of task JVM.
long sleeptimeBeforeSigkill; // waiting time before sending SIGKILL to task JVM after sending SIGTERM
}
/**
* Contains info related to the path of the file/dir to be deleted. This info
* is needed by task-controller to build the full path of the file/dir
*/
static abstract class TaskControllerPathDeletionContext
extends PathDeletionContext {
TaskController taskController;
String user;
/**
* mapredLocalDir is the base dir under which to-be-deleted jobLocalDir,
* taskWorkDir or taskAttemptDir exists. fullPath of jobLocalDir,
* taskAttemptDir or taskWorkDir is built using mapredLocalDir, jobId,
* taskId, etc.
*/
Path mapredLocalDir;
public TaskControllerPathDeletionContext(FileSystem fs, Path mapredLocalDir,
TaskController taskController,
String user) {
super(fs, null);
this.taskController = taskController;
this.mapredLocalDir = mapredLocalDir;
this.user = user;
}
@Override
protected String getPathForCleanup() {
if (fullPath == null) {
fullPath = buildPathForDeletion();
}
return fullPath;
}
/**
* Return the component of the path under the {@link #mapredLocalDir} to be
* cleaned up. Its the responsibility of the class that extends
* {@link TaskControllerPathDeletionContext} to provide the correct
* component. For example
* - For task related cleanups, either the task-work-dir or task-local-dir
* might be returned depending on jvm reuse.
* - For job related cleanup, simply the job-local-dir might be returned.
*/
abstract protected String getPath();
/**
* Builds the path of taskAttemptDir OR taskWorkDir based on
* mapredLocalDir, jobId, taskId, etc
*/
String buildPathForDeletion() {
return mapredLocalDir.toUri().getPath() + Path.SEPARATOR + getPath();
}
}
/** Contains info related to the path of the file/dir to be deleted. This info
* is needed by task-controller to build the full path of the task-work-dir or
* task-local-dir depending on whether the jvm is reused or not.
*/
static class TaskControllerTaskPathDeletionContext
extends TaskControllerPathDeletionContext {
final Task task;
final boolean isWorkDir;
public TaskControllerTaskPathDeletionContext(FileSystem fs,
Path mapredLocalDir, Task task, boolean isWorkDir,
TaskController taskController) {
super(fs, mapredLocalDir, taskController, task.getUser());
this.task = task;
this.isWorkDir = isWorkDir;
}
/**
* Returns the taskWorkDir or taskLocalDir based on whether
* {@link TaskControllerTaskPathDeletionContext} is configured to delete
* the workDir.
*/
@Override
protected String getPath() {
String subDir = (isWorkDir) ? TaskTracker.getTaskWorkDir(task.getUser(),
task.getJobID().toString(), task.getTaskID().toString(),
task.isTaskCleanupTask())
: TaskTracker.getLocalTaskDir(task.getUser(),
task.getJobID().toString(), task.getTaskID().toString(),
task.isTaskCleanupTask());
return subDir;
}
/**
* Makes the path(and its subdirectories recursively) fully deletable by
* setting proper permissions(770) by task-controller
*/
@Override
protected void enablePathForCleanup() throws IOException {
getPathForCleanup();// allow init of fullPath, if not inited already
if (fs.exists(new Path(fullPath))) {
taskController.enableTaskForCleanup(this);
}
}
}
/** Contains info related to the path of the file/dir to be deleted. This info
* is needed by task-controller to build the full path of the job-local-dir.
*/
static class TaskControllerJobPathDeletionContext
extends TaskControllerPathDeletionContext {
final JobID jobId;
public TaskControllerJobPathDeletionContext(FileSystem fs,
Path mapredLocalDir, JobID id, String user,
TaskController taskController) {
super(fs, mapredLocalDir, taskController, user);
this.jobId = id;
}
/**
* Returns the jobLocalDir of the job to be cleaned up.
*/
@Override
protected String getPath() {
return TaskTracker.getLocalJobDir(user, jobId.toString());
}
/**
* Makes the path(and its sub-directories recursively) fully deletable by
* setting proper permissions(770) by task-controller
*/
@Override
protected void enablePathForCleanup() throws IOException {
getPathForCleanup();// allow init of fullPath, if not inited already
if (fs.exists(new Path(fullPath))) {
taskController.enableJobForCleanup(this);
}
}
}
@InterfaceAudience.Private
@InterfaceStability.Unstable
public static class InitializationContext {
public File workDir;
public String user;
public InitializationContext() {
}
public InitializationContext(String user, File workDir) {
this.user = user;
this.workDir = workDir;
}
}
/**
* This is used for initializing the private localized files in distributed
* cache. Initialization would involve changing permission, ownership and etc.
*/
@InterfaceAudience.Private
@InterfaceStability.Unstable
public static class DistributedCacheFileContext extends InitializationContext {
// base directory under which file has been localized
Path localizedBaseDir;
// the unique string used to construct the localized path
String uniqueString;
public DistributedCacheFileContext(String user, File workDir,
Path localizedBaseDir, String uniqueString) {
super(user, workDir);
this.localizedBaseDir = localizedBaseDir;
this.uniqueString = uniqueString;
}
public Path getLocalizedUniqueDir() {
return new Path(localizedBaseDir, new Path(TaskTracker
.getPrivateDistributedCacheDir(user), uniqueString));
}
}
static class JobInitializationContext extends InitializationContext {
JobID jobid;
}
static class DebugScriptContext extends TaskExecContext {
List args;
File workDir;
File stdout;
}
/**
* Sends a graceful terminate signal to taskJVM and it sub-processes.
*
* @param context task context
*/
abstract void terminateTask(TaskControllerContext context);
/**
* Sends a KILL signal to forcefully terminate the taskJVM and its
* sub-processes.
*
* @param context task context
*/
abstract void killTask(TaskControllerContext context);
/**
* Sends a QUIT signal to direct the task JVM (and sub-processes) to
* dump their stack to stdout.
*
* @param context task context.
*/
abstract void dumpTaskStack(TaskControllerContext context);
/**
* Initialize user on this TaskTracer in a TaskController specific manner.
*
* @param context
* @throws IOException
*/
public abstract void initializeUser(InitializationContext context)
throws IOException;
/**
* Launch the task debug script
*
* @param context
* @throws IOException
*/
abstract void runDebugScript(DebugScriptContext context)
throws IOException;
/**
* Enable the task for cleanup by changing permissions of the path
* @param context path deletion context
* @throws IOException
*/
abstract void enableTaskForCleanup(PathDeletionContext context)
throws IOException;
/**
* Enable the job for cleanup by changing permissions of the path
* @param context path deletion context
* @throws IOException
*/
abstract void enableJobForCleanup(PathDeletionContext context)
throws IOException;
/**
* Returns the local unix user that a given job will run as.
*/
String getRunAsUser(JobConf conf) {
return System.getProperty("user.name");
}
}