/**
* 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.text.ParseException;
import java.util.ArrayList;
import java.util.List;
import java.util.HashMap;
import java.util.Map;
import java.util.Iterator;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import junit.framework.TestCase;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.fs.FileStatus;
import org.apache.hadoop.fs.permission.FsPermission;
import org.apache.hadoop.hdfs.MiniDFSCluster;
import org.apache.hadoop.mapred.JobHistory.*;
import org.apache.hadoop.mapreduce.TaskType;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
/**
* Tests the JobHistory files - to catch any changes to JobHistory that can
* cause issues for the execution of JobTracker.RecoveryManager, HistoryViewer.
*
* testJobHistoryFile
* Run a job that will be succeeded and validate its history file format and
* content.
*
* testJobHistoryUserLogLocation
* Run jobs with the given values of hadoop.job.history.user.location as
* (1)null(default case), (2)"none", and (3)some dir like "/tmp".
* Validate user history file location in each case.
*
* testJobHistoryJobStatus
* Run jobs that will be (1) succeeded (2) failed (3) killed.
* Validate job status read from history file in each case.
*
* Future changes to job history are to be reflected here in this file.
*/
public class TestJobHistory extends TestCase {
private static final Log LOG = LogFactory.getLog(TestJobHistory.class);
private static String TEST_ROOT_DIR = new File(System.getProperty(
"test.build.data", "/tmp")).toURI().toString().replace(' ', '+');
private static final Pattern digitsPattern =
Pattern.compile(JobHistory.DIGITS);
// hostname like /default-rack/host1.foo.com OR host1.foo.com
private static final Pattern hostNamePattern = Pattern.compile(
"(/(([\\w\\-\\.]+)/)+)?([\\w\\-\\.]+)");
private static final String IP_ADDR =
"\\d\\d?\\d?\\.\\d\\d?\\d?\\.\\d\\d?\\d?\\.\\d\\d?\\d?";
// hostname like /default-rack/host1.foo.com OR host1.foo.com
private static final Pattern trackerNamePattern = Pattern.compile(
"tracker_" + hostNamePattern + ":([\\w\\-\\.]+)/" +
IP_ADDR + ":" + JobHistory.DIGITS);
private static final Pattern splitsPattern = Pattern.compile(
hostNamePattern + "(," + hostNamePattern + ")*");
private static Map<String, List<String>> taskIDsToAttemptIDs =
new HashMap<String, List<String>>();
//Each Task End seen from history file is added here
private static List<String> taskEnds = new ArrayList<String>();
// List of tasks that appear in history file after JT reatart. This is to
// allow START_TIME=0 for these tasks.
private static List<String> ignoreStartTimeOfTasks = new ArrayList<String>();
// List of potential tasks whose start time can be 0 because of JT restart
private static List<String> tempIgnoreStartTimeOfTasks = new ArrayList<String>();
/**
* Listener for history log file, it populates JobHistory.JobInfo
* object with data from log file and validates the data.
*/
static class TestListener
extends DefaultJobHistoryParser.JobTasksParseListener {
int lineNum;//line number of history log file
boolean isJobLaunched;
boolean isJTRestarted;
TestListener(JobInfo job) {
super(job);
lineNum = 0;
isJobLaunched = false;
isJTRestarted = false;
}
// TestListener implementation
public void handle(RecordTypes recType, Map<Keys, String> values)
throws IOException {
lineNum++;
// Check if the record is of type Meta
if (recType == JobHistory.RecordTypes.Meta) {
long version = Long.parseLong(values.get(Keys.VERSION));
assertTrue("Unexpected job history version ",
(version >= 0 && version <= JobHistory.VERSION));
}
else if (recType.equals(RecordTypes.Job)) {
String jobid = values.get(Keys.JOBID);
assertTrue("record type 'Job' is seen without JOBID key" +
" in history file at line " + lineNum, jobid != null);
JobID id = JobID.forName(jobid);
assertTrue("JobID in history file is in unexpected format " +
"at line " + lineNum, id != null);
String time = values.get(Keys.LAUNCH_TIME);
if (time != null) {
if (isJobLaunched) {
// We assume that if we see LAUNCH_TIME again, it is because of JT restart
isJTRestarted = true;
}
else {// job launched first time
isJobLaunched = true;
}
}
time = values.get(Keys.FINISH_TIME);
if (time != null) {
assertTrue ("Job FINISH_TIME is seen in history file at line " +
lineNum + " before LAUNCH_TIME is seen", isJobLaunched);
}
}
else if (recType.equals(RecordTypes.Task)) {
String taskid = values.get(Keys.TASKID);
assertTrue("record type 'Task' is seen without TASKID key" +
" in history file at line " + lineNum, taskid != null);
TaskID id = TaskID.forName(taskid);
assertTrue("TaskID in history file is in unexpected format " +
"at line " + lineNum, id != null);
String time = values.get(Keys.START_TIME);
if (time != null) {
List<String> attemptIDs = taskIDsToAttemptIDs.get(taskid);
assertTrue("Duplicate START_TIME seen for task " + taskid +
" in history file at line " + lineNum, attemptIDs == null);
attemptIDs = new ArrayList<String>();
taskIDsToAttemptIDs.put(taskid, attemptIDs);
if (isJTRestarted) {
// This maintains a potential ignoreStartTimeTasks list
tempIgnoreStartTimeOfTasks.add(taskid);
}
}
time = values.get(Keys.FINISH_TIME);
if (time != null) {
String s = values.get(Keys.TASK_STATUS);
if (s != null) {
List<String> attemptIDs = taskIDsToAttemptIDs.get(taskid);
assertTrue ("Task FINISH_TIME is seen in history file at line " +
lineNum + " before START_TIME is seen", attemptIDs != null);
// Check if all the attemptIDs of this task are finished
assertTrue("TaskId " + taskid + " is finished at line " +
lineNum + " but its attemptID is not finished.",
(attemptIDs.size() <= 1));
// Check if at least 1 attempt of this task is seen
assertTrue("TaskId " + taskid + " is finished at line " +
lineNum + " but no attemptID is seen before this.",
attemptIDs.size() == 1);
if (s.equals("KILLED") || s.equals("FAILED")) {
// Task End with KILLED/FAILED status in history file is
// considered as TaskEnd, TaskStart. This is useful in checking
// the order of history lines.
attemptIDs = new ArrayList<String>();
taskIDsToAttemptIDs.put(taskid, attemptIDs);
}
else {
taskEnds.add(taskid);
}
}
else {
// This line of history file could be just an update to finish time
}
}
}
else if (recType.equals(RecordTypes.MapAttempt) ||
recType.equals(RecordTypes.ReduceAttempt)) {
String taskid = values.get(Keys.TASKID);
assertTrue("record type " + recType + " is seen without TASKID key" +
" in history file at line " + lineNum, taskid != null);
String attemptId = values.get(Keys.TASK_ATTEMPT_ID);
TaskAttemptID id = TaskAttemptID.forName(attemptId);
assertTrue("AttemptID in history file is in unexpected format " +
"at line " + lineNum, id != null);
String time = values.get(Keys.START_TIME);
if (time != null) {
List<String> attemptIDs = taskIDsToAttemptIDs.get(taskid);
assertTrue ("TaskAttempt is seen in history file at line " + lineNum +
" before Task is seen", attemptIDs != null);
assertFalse ("Duplicate TaskAttempt START_TIME is seen in history " +
"file at line " + lineNum, attemptIDs.remove(attemptId));
if (attemptIDs.isEmpty()) {
//just a boolean whether any attempt is seen or not
attemptIDs.add("firstAttemptIsSeen");
}
attemptIDs.add(attemptId);
if (tempIgnoreStartTimeOfTasks.contains(taskid) &&
(id.getId() < 1000)) {
// If Task line of this attempt is seen in history file after
// JT restart and if this attempt is < 1000(i.e. attempt is noti
// started after JT restart) - assuming single JT restart happened
ignoreStartTimeOfTasks.add(taskid);
}
}
time = values.get(Keys.FINISH_TIME);
if (time != null) {
List<String> attemptIDs = taskIDsToAttemptIDs.get(taskid);
assertTrue ("TaskAttempt FINISH_TIME is seen in history file at line "
+ lineNum + " before Task is seen", attemptIDs != null);
assertTrue ("TaskAttempt FINISH_TIME is seen in history file at line "
+ lineNum + " before TaskAttempt START_TIME is seen",
attemptIDs.remove(attemptId));
}
}
super.handle(recType, values);
}
}
// Check if the time is in the expected format
private static boolean isTimeValid(String time) {
Matcher m = digitsPattern.matcher(time);
return m.matches() && (Long.parseLong(time) > 0);
}
private static boolean areTimesInOrder(String time1, String time2) {
return (Long.parseLong(time1) <= Long.parseLong(time2));
}
// Validate Format of Job Level Keys, Values read from history file
private static void validateJobLevelKeyValuesFormat(Map<Keys, String> values,
String status) {
String time = values.get(Keys.SUBMIT_TIME);
assertTrue("Job SUBMIT_TIME is in unexpected format:" + time +
" in history file", isTimeValid(time));
time = values.get(Keys.LAUNCH_TIME);
assertTrue("Job LAUNCH_TIME is in unexpected format:" + time +
" in history file", isTimeValid(time));
String time1 = values.get(Keys.FINISH_TIME);
assertTrue("Job FINISH_TIME is in unexpected format:" + time1 +
" in history file", isTimeValid(time1));
assertTrue("Job FINISH_TIME is < LAUNCH_TIME in history file",
areTimesInOrder(time, time1));
String stat = values.get(Keys.JOB_STATUS);
assertTrue("Unexpected JOB_STATUS \"" + stat + "\" is seen in" +
" history file", (status.equals(stat)));
String priority = values.get(Keys.JOB_PRIORITY);
assertTrue("Unknown priority for the job in history file",
(priority.equals("HIGH") ||
priority.equals("LOW") || priority.equals("NORMAL") ||
priority.equals("VERY_HIGH") || priority.equals("VERY_LOW")));
}
// Validate Format of Task Level Keys, Values read from history file
private static void validateTaskLevelKeyValuesFormat(JobInfo job,
boolean splitsCanBeEmpty) {
Map<String, JobHistory.Task> tasks = job.getAllTasks();
// validate info of each task
for (JobHistory.Task task : tasks.values()) {
String tid = task.get(Keys.TASKID);
String time = task.get(Keys.START_TIME);
// We allow START_TIME=0 for tasks seen in history after JT restart
if (!ignoreStartTimeOfTasks.contains(tid) || (Long.parseLong(time) != 0)) {
assertTrue("Task START_TIME of " + tid + " is in unexpected format:" +
time + " in history file", isTimeValid(time));
}
String time1 = task.get(Keys.FINISH_TIME);
assertTrue("Task FINISH_TIME of " + tid + " is in unexpected format:" +
time1 + " in history file", isTimeValid(time1));
assertTrue("Task FINISH_TIME is < START_TIME in history file",
areTimesInOrder(time, time1));
// Make sure that the Task type exists and it is valid
String type = task.get(Keys.TASK_TYPE);
assertTrue("Unknown Task type \"" + type + "\" is seen in " +
"history file for task " + tid,
(type.equals("MAP") || type.equals("REDUCE") ||
type.equals("SETUP") || type.equals("CLEANUP")));
if (type.equals("MAP")) {
String splits = task.get(Keys.SPLITS);
//order in the condition OR check is important here
if (!splitsCanBeEmpty || splits.length() != 0) {
Matcher m = splitsPattern.matcher(splits);
assertTrue("Unexpected format of SPLITS \"" + splits + "\" is seen" +
" in history file for task " + tid, m.matches());
}
}
// Validate task status
String status = task.get(Keys.TASK_STATUS);
assertTrue("Unexpected TASK_STATUS \"" + status + "\" is seen in" +
" history file for task " + tid, (status.equals("SUCCESS") ||
status.equals("FAILED") || status.equals("KILLED")));
}
}
// Validate foramt of Task Attempt Level Keys, Values read from history file
private static void validateTaskAttemptLevelKeyValuesFormat(JobInfo job) {
Map<String, JobHistory.Task> tasks = job.getAllTasks();
// For each task
for (JobHistory.Task task : tasks.values()) {
// validate info of each attempt
for (JobHistory.TaskAttempt attempt : task.getTaskAttempts().values()) {
String id = attempt.get(Keys.TASK_ATTEMPT_ID);
String time = attempt.get(Keys.START_TIME);
assertTrue("START_TIME of task attempt " + id +
" is in unexpected format:" + time +
" in history file", isTimeValid(time));
String time1 = attempt.get(Keys.FINISH_TIME);
assertTrue("FINISH_TIME of task attempt " + id +
" is in unexpected format:" + time1 +
" in history file", isTimeValid(time1));
assertTrue("Task FINISH_TIME is < START_TIME in history file",
areTimesInOrder(time, time1));
// Make sure that the Task type exists and it is valid
String type = attempt.get(Keys.TASK_TYPE);
assertTrue("Unknown Task type \"" + type + "\" is seen in " +
"history file for task attempt " + id,
(type.equals("MAP") || type.equals("REDUCE") ||
type.equals("SETUP") || type.equals("CLEANUP")));
// Validate task status
String status = attempt.get(Keys.TASK_STATUS);
assertTrue("Unexpected TASK_STATUS \"" + status + "\" is seen in" +
" history file for task attempt " + id,
(status.equals("SUCCESS") || status.equals("FAILED") ||
status.equals("KILLED")));
// Reduce Task Attempts should have valid SHUFFLE_FINISHED time and
// SORT_FINISHED time
if (type.equals("REDUCE") && status.equals("SUCCESS")) {
time1 = attempt.get(Keys.SHUFFLE_FINISHED);
assertTrue("SHUFFLE_FINISHED time of task attempt " + id +
" is in unexpected format:" + time1 +
" in history file", isTimeValid(time1));
assertTrue("Reduce Task SHUFFLE_FINISHED time is < START_TIME " +
"in history file", areTimesInOrder(time, time1));
time = attempt.get(Keys.SORT_FINISHED);
assertTrue("SORT_FINISHED of task attempt " + id +
" is in unexpected format:" + time +
" in history file", isTimeValid(time));
assertTrue("Reduce Task SORT_FINISHED time is < SORT_FINISHED time" +
" in history file", areTimesInOrder(time1, time));
}
// check if hostname is valid
String hostname = attempt.get(Keys.HOSTNAME);
Matcher m = hostNamePattern.matcher(hostname);
assertTrue("Unexpected Host name of task attempt " + id, m.matches());
// check if trackername is valid
String trackerName = attempt.get(Keys.TRACKER_NAME);
m = trackerNamePattern.matcher(trackerName);
assertTrue("Unexpected tracker name of task attempt " + id,
m.matches());
if (!status.equals("KILLED")) {
// check if http port is valid
String httpPort = attempt.get(Keys.HTTP_PORT);
m = digitsPattern.matcher(httpPort);
assertTrue("Unexpected http port of task attempt " + id, m.matches());
}
// check if counters are parsable
String counters = attempt.get(Keys.COUNTERS);
try {
Counters readCounters = Counters.fromEscapedCompactString(counters);
assertTrue("Counters of task attempt " + id + " are not parsable",
readCounters != null);
} catch (ParseException pe) {
LOG.warn("While trying to parse counters of task attempt " + id +
", " + pe);
}
}
}
}
/**
* Returns the conf file name in the same
* @param path path of the jobhistory file
* @param running whether the job is running or completed
*/
private static Path getPathForConf(Path path, Path dir) {
String parts[] = path.getName().split("_");
//TODO this is a hack :(
// jobtracker-hostname_jobtracker-identifier_
String id = parts[2] + "_" + parts[3] + "_" + parts[4];
String jobUniqueString = parts[0] + "_" + parts[1] + "_" + id;
return new Path(dir, jobUniqueString + "_conf.xml");
}
/**
* Validates the format of contents of history file
* (1) history file exists and in correct location
* (2) Verify if the history file is parsable
* (3) Validate the contents of history file
* (a) Format of all TIMEs are checked against a regex
* (b) validate legality/format of job level key, values
* (c) validate legality/format of task level key, values
* (d) validate legality/format of attempt level key, values
* (e) check if all the TaskAttempts, Tasks started are finished.
* Check finish of each TaskAttemptID against its start to make sure
* that all TaskAttempts, Tasks started are indeed finished and the
* history log lines are in the proper order.
* We want to catch ordering of history lines like
* Task START
* Attempt START
* Task FINISH
* Attempt FINISH
* (speculative execution is turned off for this).
* @param id job id
* @param conf job conf
*/
static void validateJobHistoryFileFormat(JobID id, JobConf conf,
String status, boolean splitsCanBeEmpty) throws IOException {
// Get the history file name
Path dir = JobHistory.getCompletedJobHistoryLocation();
String logFileName = getDoneFile(conf, id, dir);
// Framework history log file location
Path logFile = new Path(dir, logFileName);
FileSystem fileSys = logFile.getFileSystem(conf);
// Check if the history file exists
assertTrue("History file does not exist", fileSys.exists(logFile));
// check if the history file is parsable
String[] jobDetails = JobHistory.JobInfo.decodeJobHistoryFileName(
logFileName).split("_");
String jobId = jobDetails[2] + "_" + jobDetails[3] + "_" + jobDetails[4];
JobHistory.JobInfo jobInfo = new JobHistory.JobInfo(jobId);
TestListener l = new TestListener(jobInfo);
JobHistory.parseHistoryFromFS(logFile.toString().substring(5), l, fileSys);
// validate format of job level key, values
validateJobLevelKeyValuesFormat(jobInfo.getValues(), status);
// validate format of task level key, values
validateTaskLevelKeyValuesFormat(jobInfo, splitsCanBeEmpty);
// validate format of attempt level key, values
validateTaskAttemptLevelKeyValuesFormat(jobInfo);
// check if all the TaskAttempts, Tasks started are finished for
// successful jobs
if (status.equals("SUCCESS")) {
// Make sure that the lists in taskIDsToAttemptIDs are empty.
for(Iterator<String> it = taskIDsToAttemptIDs.keySet().iterator();it.hasNext();) {
String taskid = it.next();
assertTrue("There are some Tasks which are not finished in history " +
"file.", taskEnds.contains(taskid));
List<String> attemptIDs = taskIDsToAttemptIDs.get(taskid);
if(attemptIDs != null) {
assertTrue("Unexpected. TaskID " + taskid + " has task attempt(s)" +
" that are not finished.", (attemptIDs.size() == 1));
}
}
}
}
// Validate Job Level Keys, Values read from history file by
// comparing them with the actual values from JT.
private static void validateJobLevelKeyValues(MiniMRCluster mr,
RunningJob job, JobInfo jobInfo, JobConf conf) throws IOException {
JobTracker jt = mr.getJobTrackerRunner().getJobTracker();
JobInProgress jip = jt.getJob(job.getID());
Map<Keys, String> values = jobInfo.getValues();
assertTrue("SUBMIT_TIME of job obtained from history file did not " +
"match the expected value", jip.getStartTime() ==
Long.parseLong(values.get(Keys.SUBMIT_TIME)));
assertTrue("LAUNCH_TIME of job obtained from history file did not " +
"match the expected value", jip.getLaunchTime() ==
Long.parseLong(values.get(Keys.LAUNCH_TIME)));
assertTrue("FINISH_TIME of job obtained from history file did not " +
"match the expected value", jip.getFinishTime() ==
Long.parseLong(values.get(Keys.FINISH_TIME)));
assertTrue("Job Status of job obtained from history file did not " +
"match the expected value",
values.get(Keys.JOB_STATUS).equals("SUCCESS"));
assertTrue("Job Priority of job obtained from history file did not " +
"match the expected value", jip.getPriority().toString().equals(
values.get(Keys.JOB_PRIORITY)));
assertTrue("Job Name of job obtained from history file did not " +
"match the expected value", JobInfo.getJobName(conf).equals(
values.get(Keys.JOBNAME)));
assertTrue("User Name of job obtained from history file did not " +
"match the expected value", JobInfo.getUserName(conf).equals(
values.get(Keys.USER)));
// Validate job counters
Counters c = jip.getCounters();
assertTrue("Counters of job obtained from history file did not " +
"match the expected value",
c.makeEscapedCompactString().equals(values.get(Keys.COUNTERS)));
Counters m = jip.getMapCounters();
assertTrue("Map Counters of job obtained from history file did not " +
"match the expected value", m.makeEscapedCompactString().
equals(values.get(Keys.MAP_COUNTERS)));
Counters r = jip.getReduceCounters();
assertTrue("Reduce Counters of job obtained from history file did not " +
"match the expected value", r.makeEscapedCompactString().
equals(values.get(Keys.REDUCE_COUNTERS)));
// Validate number of total maps, total reduces, finished maps,
// finished reduces, failed maps, failed recudes
String totalMaps = values.get(Keys.TOTAL_MAPS);
assertTrue("Unexpected number of total maps in history file",
Integer.parseInt(totalMaps) == jip.desiredMaps());
String totalReduces = values.get(Keys.TOTAL_REDUCES);
assertTrue("Unexpected number of total reduces in history file",
Integer.parseInt(totalReduces) == jip.desiredReduces());
String finMaps = values.get(Keys.FINISHED_MAPS);
assertTrue("Unexpected number of finished maps in history file",
Integer.parseInt(finMaps) == jip.finishedMaps());
String finReduces = values.get(Keys.FINISHED_REDUCES);
assertTrue("Unexpected number of finished reduces in history file",
Integer.parseInt(finReduces) == jip.finishedReduces());
String failedMaps = values.get(Keys.FAILED_MAPS);
assertTrue("Unexpected number of failed maps in history file",
Integer.parseInt(failedMaps) == jip.failedMapTasks);
String failedReduces = values.get(Keys.FAILED_REDUCES);
assertTrue("Unexpected number of failed reduces in history file",
Integer.parseInt(failedReduces) == jip.failedReduceTasks);
}
// Validate Task Level Keys, Values read from history file by
// comparing them with the actual values from JT.
private static void validateTaskLevelKeyValues(MiniMRCluster mr,
RunningJob job, JobInfo jobInfo) throws IOException {
JobTracker jt = mr.getJobTrackerRunner().getJobTracker();
JobInProgress jip = jt.getJob(job.getID());
// Get the 1st map, 1st reduce, cleanup & setup taskIDs and
// validate their history info
TaskID mapTaskId = new TaskID(job.getID(), true, 0);
TaskID reduceTaskId = new TaskID(job.getID(), false, 0);
TaskInProgress cleanups[] = jip.getTasks(TaskType.JOB_CLEANUP);
TaskID cleanupTaskId;
if (cleanups[0].isComplete()) {
cleanupTaskId = cleanups[0].getTIPId();
}
else {
cleanupTaskId = cleanups[1].getTIPId();
}
TaskInProgress setups[] = jip.getTasks(TaskType.JOB_SETUP);
TaskID setupTaskId;
if (setups[0].isComplete()) {
setupTaskId = setups[0].getTIPId();
}
else {
setupTaskId = setups[1].getTIPId();
}
Map<String, JobHistory.Task> tasks = jobInfo.getAllTasks();
// validate info of the 4 tasks(cleanup, setup, 1st map, 1st reduce)
for (JobHistory.Task task : tasks.values()) {
String tid = task.get(Keys.TASKID);
if (tid.equals(mapTaskId.toString()) ||
tid.equals(reduceTaskId.toString()) ||
tid.equals(cleanupTaskId.toString()) ||
tid.equals(setupTaskId.toString())) {
TaskID taskId = null;
if (tid.equals(mapTaskId.toString())) {
taskId = mapTaskId;
}
else if (tid.equals(reduceTaskId.toString())) {
taskId = reduceTaskId;
}
else if (tid.equals(cleanupTaskId.toString())) {
taskId = cleanupTaskId;
}
else if (tid.equals(setupTaskId.toString())) {
taskId = setupTaskId;
}
TaskInProgress tip = jip.getTaskInProgress(taskId);
assertTrue("START_TIME of Task " + tid + " obtained from history " +
"file did not match the expected value", tip.getExecStartTime() ==
Long.parseLong(task.get(Keys.START_TIME)));
assertTrue("FINISH_TIME of Task " + tid + " obtained from history " +
"file did not match the expected value", tip.getExecFinishTime() ==
Long.parseLong(task.get(Keys.FINISH_TIME)));
if (taskId == mapTaskId) {//check splits only for map task
assertTrue("Splits of Task " + tid + " obtained from history file " +
" did not match the expected value",
tip.getSplitNodes().equals(task.get(Keys.SPLITS)));
}
TaskAttemptID attemptId = tip.getSuccessfulTaskid();
TaskStatus ts = tip.getTaskStatus(attemptId);
// Validate task counters
Counters c = ts.getCounters();
assertTrue("Counters of Task " + tid + " obtained from history file " +
" did not match the expected value",
c.makeEscapedCompactString().equals(task.get(Keys.COUNTERS)));
}
}
}
// Validate Task Attempt Level Keys, Values read from history file by
// comparing them with the actual values from JT.
private static void validateTaskAttemptLevelKeyValues(MiniMRCluster mr,
RunningJob job, JobInfo jobInfo) throws IOException {
JobTracker jt = mr.getJobTrackerRunner().getJobTracker();
JobInProgress jip = jt.getJob(job.getID());
Map<String, JobHistory.Task> tasks = jobInfo.getAllTasks();
// For each task
for (JobHistory.Task task : tasks.values()) {
// validate info of each attempt
for (JobHistory.TaskAttempt attempt : task.getTaskAttempts().values()) {
String idStr = attempt.get(Keys.TASK_ATTEMPT_ID);
TaskAttemptID attemptId = TaskAttemptID.forName(idStr);
TaskID tid = attemptId.getTaskID();
// Validate task id
assertTrue("Task id of Task Attempt " + idStr + " obtained from " +
"history file did not match the expected value",
tid.toString().equals(attempt.get(Keys.TASKID)));
TaskInProgress tip = jip.getTaskInProgress(tid);
TaskStatus ts = tip.getTaskStatus(attemptId);
// Validate task attempt start time
assertTrue("START_TIME of Task attempt " + idStr + " obtained from " +
"history file did not match the expected value",
ts.getStartTime() == Long.parseLong(attempt.get(Keys.START_TIME)));
// Validate task attempt finish time
assertTrue("FINISH_TIME of Task attempt " + idStr + " obtained from " +
"history file did not match the expected value",
ts.getFinishTime() == Long.parseLong(attempt.get(Keys.FINISH_TIME)));
TaskTrackerStatus ttStatus =
jt.getTaskTrackerStatus(ts.getTaskTracker());
if (ttStatus != null) {
assertTrue("http port of task attempt " + idStr + " obtained from " +
"history file did not match the expected value",
ttStatus.getHttpPort() ==
Integer.parseInt(attempt.get(Keys.HTTP_PORT)));
if (attempt.get(Keys.TASK_STATUS).equals("SUCCESS")) {
String ttHostname = jt.getNode(ttStatus.getHost()).toString();
// check if hostname is valid
assertTrue("Host name of task attempt " + idStr + " obtained from" +
" history file did not match the expected value",
ttHostname.equals(attempt.get(Keys.HOSTNAME)));
}
}
if (attempt.get(Keys.TASK_STATUS).equals("SUCCESS")) {
// Validate SHUFFLE_FINISHED time and SORT_FINISHED time of
// Reduce Task Attempts
if (attempt.get(Keys.TASK_TYPE).equals("REDUCE")) {
assertTrue("SHUFFLE_FINISHED time of task attempt " + idStr +
" obtained from history file did not match the expected" +
" value", ts.getShuffleFinishTime() ==
Long.parseLong(attempt.get(Keys.SHUFFLE_FINISHED)));
assertTrue("SORT_FINISHED time of task attempt " + idStr +
" obtained from history file did not match the expected" +
" value", ts.getSortFinishTime() ==
Long.parseLong(attempt.get(Keys.SORT_FINISHED)));
}
//Validate task counters
Counters c = ts.getCounters();
assertTrue("Counters of Task Attempt " + idStr + " obtained from " +
"history file did not match the expected value",
c.makeEscapedCompactString().equals(attempt.get(Keys.COUNTERS)));
}
// check if tracker name is valid
assertTrue("Tracker name of task attempt " + idStr + " obtained from " +
"history file did not match the expected value",
ts.getTaskTracker().equals(attempt.get(Keys.TRACKER_NAME)));
}
}
}
/**
* Checks if the history file content is as expected comparing with the
* actual values obtained from JT.
* Job Level, Task Level and Task Attempt Level Keys, Values are validated.
* @param job RunningJob object of the job whose history is to be validated
* @param conf job conf
*/
static void validateJobHistoryFileContent(MiniMRCluster mr,
RunningJob job, JobConf conf) throws IOException {
JobID id = job.getID();
Path doneDir = JobHistory.getCompletedJobHistoryLocation();
// Get the history file name
String logFileName = getDoneFile(conf, id, doneDir);
// Framework history log file location
Path logFile = new Path(doneDir, logFileName);
FileSystem fileSys = logFile.getFileSystem(conf);
// Check if the history file exists
assertTrue("History file does not exist", fileSys.exists(logFile));
// check if the history file is parsable
String[] jobDetails = JobHistory.JobInfo.decodeJobHistoryFileName(
logFileName).split("_");
String jobId = jobDetails[2] + "_" + jobDetails[3] + "_" + jobDetails[4];
JobHistory.JobInfo jobInfo = new JobHistory.JobInfo(jobId);
DefaultJobHistoryParser.JobTasksParseListener l =
new DefaultJobHistoryParser.JobTasksParseListener(jobInfo);
JobHistory.parseHistoryFromFS(logFile.toString().substring(5), l, fileSys);
// Now the history file contents are available in jobInfo. Let us compare
// them with the actual values from JT.
validateJobLevelKeyValues(mr, job, jobInfo, conf);
validateTaskLevelKeyValues(mr, job, jobInfo);
validateTaskAttemptLevelKeyValues(mr, job, jobInfo);
}
public void testDoneFolderOnHDFS() throws IOException {
MiniMRCluster mr = null;
try {
JobConf conf = new JobConf();
// keep for less time
conf.setLong("mapred.jobtracker.retirejob.check", 1000);
conf.setLong("mapred.jobtracker.retirejob.interval", 100000);
//set the done folder location
String doneFolder = "history_done";
conf.set("mapred.job.tracker.history.completed.location", doneFolder);
MiniDFSCluster dfsCluster = new MiniDFSCluster(conf, 2, true, null);
mr = new MiniMRCluster(2, dfsCluster.getFileSystem().getUri().toString(),
3, null, null, conf);
// run the TCs
conf = mr.createJobConf();
FileSystem fs = FileSystem.get(conf);
// clean up
fs.delete(new Path("succeed"), true);
Path inDir = new Path("succeed/input");
Path outDir = new Path("succeed/output");
//Disable speculative execution
conf.setSpeculativeExecution(false);
// Make sure that the job is not removed from memory until we do finish
// the validation of history file content
conf.setInt("mapred.jobtracker.completeuserjobs.maximum", 10);
// Run a job that will be succeeded and validate its history file
RunningJob job = UtilsForTests.runJobSucceed(conf, inDir, outDir);
Path doneDir = JobHistory.getCompletedJobHistoryLocation();
assertEquals("History DONE folder not correct",
doneFolder, doneDir.getName());
JobID id = job.getID();
String logFileName = getDoneFile(conf, id, doneDir);
// Framework history log file location
Path logFile = new Path(doneDir, logFileName);
FileSystem fileSys = logFile.getFileSystem(conf);
// Check if the history file exists
assertTrue("History file does not exist", fileSys.exists(logFile));
// check if the corresponding conf file exists
Path confFile = getPathForConf(logFile, doneDir);
assertTrue("Config for completed jobs doesnt exist",
fileSys.exists(confFile));
// check if the file exists in a done folder
assertTrue("Completed job config doesnt exist in the done folder",
doneDir.getName().equals(confFile.getParent().getName()));
// check if the file exists in a done folder
assertTrue("Completed jobs doesnt exist in the done folder",
doneDir.getName().equals(logFile.getParent().getName()));
// check if the job file is removed from the history location
Path runningJobsHistoryFolder = logFile.getParent().getParent();
Path runningJobHistoryFilename =
new Path(runningJobsHistoryFolder, logFile.getName());
Path runningJobConfFilename =
new Path(runningJobsHistoryFolder, confFile.getName());
assertFalse("History file not deleted from the running folder",
fileSys.exists(runningJobHistoryFilename));
assertFalse("Config for completed jobs not deleted from running folder",
fileSys.exists(runningJobConfFilename));
validateJobHistoryFileFormat(job.getID(), conf, "SUCCESS", false);
validateJobHistoryFileContent(mr, job, conf);
// get the job conf filename
} finally {
if (mr != null) {
cleanupLocalFiles(mr);
mr.shutdown();
}
}
}
/** Run a job that will be succeeded and validate its history file format
* and its content.
*/
public void testJobHistoryFile() throws IOException {
MiniMRCluster mr = null;
try {
JobConf conf = new JobConf();
// keep for less time
conf.setLong("mapred.jobtracker.retirejob.check", 1000);
conf.setLong("mapred.jobtracker.retirejob.interval", 100000);
//set the done folder location
String doneFolder = TEST_ROOT_DIR + "history_done";
conf.set("mapred.job.tracker.history.completed.location", doneFolder);
mr = new MiniMRCluster(2, "file:///", 3, null, null, conf);
// run the TCs
conf = mr.createJobConf();
FileSystem fs = FileSystem.get(conf);
// clean up
fs.delete(new Path(TEST_ROOT_DIR + "/succeed"), true);
Path inDir = new Path(TEST_ROOT_DIR + "/succeed/input");
Path outDir = new Path(TEST_ROOT_DIR + "/succeed/output");
//Disable speculative execution
conf.setSpeculativeExecution(false);
// Make sure that the job is not removed from memory until we do finish
// the validation of history file content
conf.setInt("mapred.jobtracker.completeuserjobs.maximum", 10);
// Run a job that will be succeeded and validate its history file
RunningJob job = UtilsForTests.runJobSucceed(conf, inDir, outDir);
Path doneDir = JobHistory.getCompletedJobHistoryLocation();
assertEquals("History DONE folder not correct",
doneFolder, doneDir.toString());
JobID id = job.getID();
String logFileName = getDoneFile(conf, id, doneDir);
// Framework history log file location
Path logFile = new Path(doneDir, logFileName);
FileSystem fileSys = logFile.getFileSystem(conf);
// Check if the history file exists
assertTrue("History file does not exist", fileSys.exists(logFile));
// check if the corresponding conf file exists
Path confFile = getPathForConf(logFile, doneDir);
assertTrue("Config for completed jobs doesnt exist",
fileSys.exists(confFile));
// check if the file exists in a done folder
assertTrue("Completed job config doesnt exist in the done folder",
doneDir.getName().equals(confFile.getParent().getName()));
// check if the file exists in a done folder
assertTrue("Completed jobs doesnt exist in the done folder",
doneDir.getName().equals(logFile.getParent().getName()));
// check if the job file is removed from the history location
Path runningJobsHistoryFolder = logFile.getParent().getParent();
Path runningJobHistoryFilename =
new Path(runningJobsHistoryFolder, logFile.getName());
Path runningJobConfFilename =
new Path(runningJobsHistoryFolder, confFile.getName());
assertFalse("History file not deleted from the running folder",
fileSys.exists(runningJobHistoryFilename));
assertFalse("Config for completed jobs not deleted from running folder",
fileSys.exists(runningJobConfFilename));
validateJobHistoryFileFormat(job.getID(), conf, "SUCCESS", false);
validateJobHistoryFileContent(mr, job, conf);
// get the job conf filename
String name = JobHistory.JobInfo.getLocalJobFilePath(job.getID());
File file = new File(name);
// check if the file get deleted
while (file.exists()) {
LOG.info("Waiting for " + file + " to be deleted");
UtilsForTests.waitFor(100);
}
} finally {
if (mr != null) {
cleanupLocalFiles(mr);
mr.shutdown();
}
}
}
//Returns the file in the done folder
//Waits for sometime to get the file moved to done
private static String getDoneFile(JobConf conf, JobID id,
Path doneDir) throws IOException {
String name = null;
for (int i = 0; name == null && i < 20; i++) {
name = JobHistory.JobInfo.getDoneJobHistoryFileName(conf, id);
UtilsForTests.waitFor(1000);
}
return name;
}
// Returns the output path where user history log file is written to with
// default configuration setting for hadoop.job.history.user.location
private static Path getLogLocationInOutputPath(String logFileName,
JobConf conf) {
JobConf jobConf = new JobConf(true);//default JobConf
FileOutputFormat.setOutputPath(jobConf,
FileOutputFormat.getOutputPath(conf));
return JobHistory.JobInfo.getJobHistoryLogLocationForUser(
logFileName, jobConf);
}
/**
* Checks if the user history file exists in the correct dir
* @param id job id
* @param conf job conf
*/
private static void validateJobHistoryUserLogLocation(JobID id, JobConf conf)
throws IOException {
// Get the history file name
Path doneDir = JobHistory.getCompletedJobHistoryLocation();
String logFileName = getDoneFile(conf, id, doneDir);
// User history log file location
Path logFile = JobHistory.JobInfo.getJobHistoryLogLocationForUser(
logFileName, conf);
if(logFile == null) {
// get the output path where history file is written to when
// hadoop.job.history.user.location is not set
logFile = getLogLocationInOutputPath(logFileName, conf);
}
FileSystem fileSys = null;
fileSys = logFile.getFileSystem(conf);
// Check if the user history file exists in the correct dir
if (conf.get("hadoop.job.history.user.location") == null) {
assertTrue("User log file " + logFile + " does not exist",
fileSys.exists(logFile));
}
else if ("none".equals(conf.get("hadoop.job.history.user.location"))) {
// history file should not exist in the output path
assertFalse("Unexpected. User log file exists in output dir when " +
"hadoop.job.history.user.location is set to \"none\"",
fileSys.exists(logFile));
}
else {
//hadoop.job.history.user.location is set to a specific location.
// User log file should exist in that location
assertTrue("User log file " + logFile + " does not exist",
fileSys.exists(logFile));
// User log file should not exist in output path.
// get the output path where history file is written to when
// hadoop.job.history.user.location is not set
Path logFile1 = getLogLocationInOutputPath(logFileName, conf);
if (logFile != logFile1) {
fileSys = logFile1.getFileSystem(conf);
assertFalse("Unexpected. User log file exists in output dir when " +
"hadoop.job.history.user.location is set to a different location",
fileSys.exists(logFile1));
}
}
}
// Validate user history file location for the given values of
// hadoop.job.history.user.location as
// (1)null(default case), (2)"none", and (3)some dir "/tmp"
public void testJobHistoryUserLogLocation() throws IOException {
MiniMRCluster mr = null;
try {
mr = new MiniMRCluster(2, "file:///", 3);
// run the TCs
JobConf conf = mr.createJobConf();
FileSystem fs = FileSystem.get(conf);
// clean up
fs.delete(new Path(TEST_ROOT_DIR + "/succeed"), true);
Path inDir = new Path(TEST_ROOT_DIR + "/succeed/input1");
Path outDir = new Path(TEST_ROOT_DIR + "/succeed/output1");
// validate for the case of null(default)
RunningJob job = UtilsForTests.runJobSucceed(conf, inDir, outDir);
validateJobHistoryUserLogLocation(job.getID(), conf);
inDir = new Path(TEST_ROOT_DIR + "/succeed/input2");
outDir = new Path(TEST_ROOT_DIR + "/succeed/output2");
// validate for the case of "none"
conf.set("hadoop.job.history.user.location", "none");
job = UtilsForTests.runJobSucceed(conf, inDir, outDir);
validateJobHistoryUserLogLocation(job.getID(), conf);
inDir = new Path(TEST_ROOT_DIR + "/succeed/input3");
outDir = new Path(TEST_ROOT_DIR + "/succeed/output3");
// validate for the case of any dir
conf.set("hadoop.job.history.user.location", "/tmp");
job = UtilsForTests.runJobSucceed(conf, inDir, outDir);
validateJobHistoryUserLogLocation(job.getID(), conf);
} finally {
if (mr != null) {
cleanupLocalFiles(mr);
mr.shutdown();
}
}
}
private void cleanupLocalFiles(MiniMRCluster mr)
throws IOException {
Configuration conf = mr.createJobConf();
JobTracker jt = mr.getJobTrackerRunner().getJobTracker();
Path sysDir = new Path(jt.getSystemDir());
FileSystem fs = sysDir.getFileSystem(conf);
fs.delete(sysDir, true);
Path jobHistoryDir = JobHistory.getJobHistoryLocation();
fs = jobHistoryDir.getFileSystem(conf);
fs.delete(jobHistoryDir, true);
}
/**
* Checks if the history file has expected job status
* @param id job id
* @param conf job conf
*/
private static void validateJobHistoryJobStatus(JobID id, JobConf conf,
String status) throws IOException {
// Get the history file name
Path doneDir = JobHistory.getCompletedJobHistoryLocation();
String logFileName = getDoneFile(conf, id, doneDir);
// Framework history log file location
Path logFile = new Path(doneDir, logFileName);
FileSystem fileSys = logFile.getFileSystem(conf);
// Check if the history file exists
assertTrue("History file does not exist", fileSys.exists(logFile));
// check history file permission
assertTrue("History file permissions does not match",
fileSys.getFileStatus(logFile).getPermission().equals(
new FsPermission(JobHistory.HISTORY_FILE_PERMISSION)));
// check if the history file is parsable
String[] jobDetails = JobHistory.JobInfo.decodeJobHistoryFileName(
logFileName).split("_");
String jobId = jobDetails[2] + "_" + jobDetails[3] + "_" + jobDetails[4];
JobHistory.JobInfo jobInfo = new JobHistory.JobInfo(jobId);
DefaultJobHistoryParser.JobTasksParseListener l =
new DefaultJobHistoryParser.JobTasksParseListener(jobInfo);
JobHistory.parseHistoryFromFS(logFile.toString().substring(5), l, fileSys);
assertTrue("Job Status read from job history file is not the expected" +
" status", status.equals(jobInfo.getValues().get(Keys.JOB_STATUS)));
}
// run jobs that will be (1) succeeded (2) failed (3) killed
// and validate job status read from history file in each case
public void testJobHistoryJobStatus() throws IOException {
MiniMRCluster mr = null;
try {
mr = new MiniMRCluster(2, "file:///", 3);
// run the TCs
JobConf conf = mr.createJobConf();
FileSystem fs = FileSystem.get(conf);
// clean up
fs.delete(new Path(TEST_ROOT_DIR + "/succeedfailkilljob"), true);
Path inDir = new Path(TEST_ROOT_DIR + "/succeedfailkilljob/input");
Path outDir = new Path(TEST_ROOT_DIR + "/succeedfailkilljob/output");
// Run a job that will be succeeded and validate its job status
// existing in history file
RunningJob job = UtilsForTests.runJobSucceed(conf, inDir, outDir);
validateJobHistoryJobStatus(job.getID(), conf, "SUCCESS");
long historyCleanerRanAt = JobHistory.HistoryCleaner.getLastRan();
assertTrue(historyCleanerRanAt != 0);
// Run a job that will be failed and validate its job status
// existing in history file
job = UtilsForTests.runJobFail(conf, inDir, outDir);
validateJobHistoryJobStatus(job.getID(), conf, "FAILED");
assertTrue(historyCleanerRanAt == JobHistory.HistoryCleaner.getLastRan());
// Run a job that will be killed and validate its job status
// existing in history file
job = UtilsForTests.runJobKill(conf, inDir, outDir);
validateJobHistoryJobStatus(job.getID(), conf, "KILLED");
assertTrue(historyCleanerRanAt == JobHistory.HistoryCleaner.getLastRan());
} finally {
if (mr != null) {
cleanupLocalFiles(mr);
mr.shutdown();
}
}
}
}