List of usage examples for org.apache.hadoop.mapreduce Job isSuccessful
public boolean isSuccessful() throws IOException
From source file:com.datasalt.utils.mapred.joiner.TestJoinOneToMany.java
License:Apache License
@Test public void test() throws IOException, InterruptedException, ClassNotFoundException { File input = new File(INPUT1); if (input.exists()) { while (!input.delete()) ;// w w w . j av a 2s . com } Files.write("line", input, Charset.defaultCharset()); input = new File(INPUT2); if (input.exists()) { while (!input.delete()) ; } Files.write("line", input, Charset.defaultCharset()); Configuration conf = getConf(); Job job = getMultiJoiner(conf); job.waitForCompletion(true); assertTrue(job.isSuccessful()); File out = new File(OUTPUT, "part-r-00000"); List<String> lines = Files.readLines(out, Charset.defaultCharset()); System.out.println(lines); assertEquals(6, lines.size()); assertTrue(lines.contains("1 foo")); assertTrue(lines.contains("1 bar")); assertTrue(lines.contains("2 oh la la")); assertTrue(lines.contains("2 blah blah")); assertTrue(lines.contains("-1 bluu")); assertTrue(lines.contains("3 snull")); assertTrue(firstWasSecondClass == 1); assertTrue(noSecondClass == true); cleanUp(); }
From source file:com.datasalt.utils.mapred.joiner.TestMultiJoiner.java
License:Apache License
@Test public void test() throws IOException, InterruptedException, ClassNotFoundException { Configuration conf = getConf(); MultiJoiner multiJoiner = new MultiJoiner("MultiJoiner Test", conf); multiJoiner.setReducer(TestReducer.class); multiJoiner.setOutputKeyClass(Text.class); multiJoiner.setOutputValueClass(Text.class); multiJoiner.setOutputFormat(TextOutputFormat.class); multiJoiner.setOutputPath(new Path(OUTPUT_FOR_TEST)); Job job = multiJoiner .addChanneledInput(2, new Path("src/test/resources/multijoiner.test.a.txt"), A.class, TextInputFormat.class, AMapper.class) .addChanneledInput(4, new Path("src/test/resources/multijoiner.test.b.txt"), B.class, TextInputFormat.class, BMapper.class) .getJob();/*from ww w. j av a 2 s . c o m*/ job.waitForCompletion(true); assertTrue(job.isSuccessful()); HadoopUtils.deleteIfExists(FileSystem.get(conf), new Path(OUTPUT_FOR_TEST)); }
From source file:com.datasalt.utils.mapred.joiner.TestMultiJoinerGlob.java
License:Apache License
@Test public void test() throws IOException, InterruptedException, ClassNotFoundException { Configuration conf = getConf(); MultiJoiner multiJoiner = new MultiJoiner("MultiJoiner Test", conf); multiJoiner.setReducer(TestReducer.class); multiJoiner.setOutputKeyClass(Text.class); multiJoiner.setOutputValueClass(Text.class); multiJoiner.setOutputFormat(TextOutputFormat.class); multiJoiner.setOutputPath(new Path(OUTPUT_FOR_TEST)); Job job = multiJoiner .addChanneledInput(0, new Path("src/test/resources/glob-folder/*"), A.class, TextInputFormat.class, AMapper.class) .addChanneledInput(1, new Path("src/test/resources/multijoiner.test.b.txt"), B.class, TextInputFormat.class, BMapper.class) .getJob();//from ww w . j a va 2 s . com job.waitForCompletion(true); assertTrue(job.isSuccessful()); HadoopUtils.deleteIfExists(FileSystem.get(conf), new Path(OUTPUT_FOR_TEST)); }
From source file:com.datasalt.utils.mapred.joiner.TestMultiJoinerMultiChannel.java
License:Apache License
@Test public void test() throws IOException, InterruptedException, ClassNotFoundException { Configuration conf = getConf(); MultiJoiner multiJoiner = new MultiJoiner("MultiJoiner Test", conf); multiJoiner.setReducer(TestReducer.class); multiJoiner.setOutputKeyClass(Text.class); multiJoiner.setOutputValueClass(Text.class); multiJoiner.setOutputFormat(TextOutputFormat.class); multiJoiner.setOutputPath(new Path(OUTPUT_FOR_TEST)); Job job = multiJoiner.addInput(new Path("src/test/resources/multijoiner.test.a.txt"), TextInputFormat.class, ABMapper.class).setChannelDatumClass(0, A.class).setChannelDatumClass(1, B.class).getJob(); job.waitForCompletion(true);//from www. j a v a 2 s . c om assertTrue(job.isSuccessful()); HadoopUtils.deleteIfExists(FileSystem.get(conf), new Path(OUTPUT_FOR_TEST)); }
From source file:com.datasalt.utils.mapred.joiner.TestMultiJoinerSameClass.java
License:Apache License
@Test public void test() throws IOException, InterruptedException, ClassNotFoundException { Configuration conf = getConf(); MultiJoiner multiJoiner = new MultiJoiner("MultiJoiner Test", conf); multiJoiner.setReducer(TestReducerSameClass.class); multiJoiner.setOutputKeyClass(Text.class); multiJoiner.setOutputValueClass(Text.class); multiJoiner.setOutputFormat(TextOutputFormat.class); multiJoiner.setOutputPath(new Path(OUTPUT_FOR_TEST)); Job job = multiJoiner .addChanneledInput(0, new Path("src/test/resources/multijoiner.test.a.txt"), A.class, TextInputFormat.class, AMapperSameClass.class) .addChanneledInput(1, new Path("src/test/resources/multijoiner.test.same.class.a.txt"), A.class, TextInputFormat.class, AMapperSameClass.class) .getJob();/*w w w.j a va2 s .co m*/ job.waitForCompletion(true); assertTrue(job.isSuccessful()); HadoopUtils.deleteIfExists(FileSystem.get(conf), new Path(OUTPUT_FOR_TEST)); }
From source file:com.datasalt.utils.mapred.joiner.TestMultiJoinerSecondarySort.java
License:Apache License
@Test public void test() throws IOException, InterruptedException, ClassNotFoundException { Configuration conf = getConf(); MultiJoiner multiJoiner = new MultiJoiner("MultiJoiner Test", conf); multiJoiner.setReducer(TestReducerSecondarySort.class); multiJoiner.setOutputKeyClass(Text.class); multiJoiner.setOutputValueClass(Text.class); multiJoiner.setOutputFormat(TextOutputFormat.class); multiJoiner.setOutputPath(new Path(OUTPUT_FOR_TEST)); Job job = multiJoiner.setMultiJoinPairClass(MultiJoinPairText.class) .addChanneledInput(0, new Path("src/test/resources/multijoiner.test.a.2.txt"), A.class, TextInputFormat.class, AMapperSecondarySort.class) .addChanneledInput(1, new Path("src/test/resources/multijoiner.test.b.2.txt"), B.class, TextInputFormat.class, BMapperSecondarySort.class) .getJob();//from w ww. j a v a2s. c o m job.waitForCompletion(true); assertTrue(job.isSuccessful()); HadoopUtils.deleteIfExists(FileSystem.get(conf), new Path(OUTPUT_FOR_TEST)); }
From source file:com.ikanow.aleph2.analytics.hadoop.assets.VerySimpleLocalExample.java
License:Apache License
@SuppressWarnings({ "deprecation", "unchecked", "rawtypes" })
@Test//from w w w . ja va 2s . co m
public void test_localHadoopLaunch()
throws IOException, IllegalStateException, ClassNotFoundException, InterruptedException {
// 0) Setup the temp dir
final String temp_dir = System.getProperty("java.io.tmpdir") + File.separator;
//final Path tmp_path = FileContext.getLocalFSFileContext().makeQualified(new Path(temp_dir));
final Path tmp_path2 = FileContext.getLocalFSFileContext()
.makeQualified(new Path(temp_dir + "/tmp_output"));
try {
FileContext.getLocalFSFileContext().delete(tmp_path2, true);
} catch (Exception e) {
} // (just doesn't exist yet)
// 1) Setup config with local mode
final Configuration config = new Configuration();
config.setBoolean("mapred.used.genericoptionsparser", true); // (just stops an annoying warning from appearing)
config.set("fs.file.impl", "org.apache.hadoop.fs.LocalFileSystem");
config.set("mapred.job.tracker", "local");
config.set("fs.defaultFS", "local");
config.unset("mapreduce.framework.name");
// If running locally, turn "snappy" off - tomcat isn't pointing its native library path in the right place
config.set("mapred.map.output.compression.codec", "org.apache.hadoop.io.compress.DefaultCodec");
// 2) Build job and do more setup using the Job API
//TODO: not sure why this is deprecated, it doesn't seem to be in v1? We do need to move to JobConf at some point, but I ran into some
// issues when trying to do everything I needed to for V1, so seems expedient to start here and migrate away
final Job hj = new Job(config); // (NOTE: from here, changes to config are ignored)
// Input format:
//TOOD: fails because of guava issue, looks like we'll need to move to 2.7 and check it works with 2.5.x server?
//TextInputFormat.addInputPath(hj, tmp_path);
//hj.setInputFormatClass((Class<? extends InputFormat>) Class.forName ("org.apache.hadoop.mapreduce.lib.input.TextInputFormat"));
hj.setInputFormatClass(TestInputFormat.class);
// Output format:
hj.setOutputFormatClass((Class<? extends OutputFormat>) Class
.forName("org.apache.hadoop.mapreduce.lib.output.TextOutputFormat"));
TextOutputFormat.setOutputPath(hj, tmp_path2);
// Mapper etc (combiner/reducer are similar)
hj.setMapperClass(TestMapper.class);
hj.setOutputKeyClass(Text.class);
hj.setOutputValueClass(Text.class);
hj.setNumReduceTasks(0); // (disable reducer for now)
hj.setJar("test");
try {
hj.submit();
} catch (UnsatisfiedLinkError e) {
throw new RuntimeException(
"This is a windows/hadoop compatibility problem - adding the hadoop-commons in the misc_test_assets subdirectory to the top of the classpath should resolve it (and does in V1), though I haven't yet made that work with Aleph2",
e);
}
//hj.getJobID().toString();
while (!hj.isComplete()) {
Thread.sleep(1000);
}
assertTrue("Finished successfully", hj.isSuccessful());
}
From source file:com.ikanow.infinit.e.processing.custom.launcher.CustomHadoopTaskLauncher.java
License:Open Source License
@SuppressWarnings({ "unchecked", "rawtypes" })
public String runHadoopJob(CustomMapReduceJobPojo job, String tempJarLocation)
throws IOException, SAXException, ParserConfigurationException {
StringWriter xml = new StringWriter();
String outputCollection = job.outputCollectionTemp;// (non-append mode)
if ((null != job.appendResults) && job.appendResults)
outputCollection = job.outputCollection; // (append mode, write directly in....)
else if (null != job.incrementalMode)
job.incrementalMode = false; // (not allowed to be in incremental mode and not update mode)
createConfigXML(xml, job.jobtitle, job.inputCollection,
InfiniteHadoopUtils.getQueryOrProcessing(job.query, InfiniteHadoopUtils.QuerySpec.INPUTFIELDS),
job.isCustomTable, job.getOutputDatabase(), job._id.toString(), outputCollection, job.mapper,
job.reducer, job.combiner,/*from ww w . j av a 2 s. com*/
InfiniteHadoopUtils.getQueryOrProcessing(job.query, InfiniteHadoopUtils.QuerySpec.QUERY),
job.communityIds, job.outputKey, job.outputValue, job.arguments, job.incrementalMode,
job.submitterID, job.selfMerge, job.outputCollection, job.appendResults);
ClassLoader savedClassLoader = Thread.currentThread().getContextClassLoader();
URLClassLoader child = new URLClassLoader(new URL[] { new File(tempJarLocation).toURI().toURL() },
savedClassLoader);
Thread.currentThread().setContextClassLoader(child);
// Check version: for now, any infinit.e.data_model with an VersionTest class is acceptable
boolean dataModelLoaded = true;
try {
URLClassLoader versionTest = new URLClassLoader(new URL[] { new File(tempJarLocation).toURI().toURL() },
null);
try {
Class.forName("com.ikanow.infinit.e.data_model.custom.InfiniteMongoInputFormat", true, versionTest);
} catch (ClassNotFoundException e2) {
//(this is fine, will use the cached version)
dataModelLoaded = false;
}
if (dataModelLoaded)
Class.forName("com.ikanow.infinit.e.data_model.custom.InfiniteMongoVersionTest", true, versionTest);
} catch (ClassNotFoundException e1) {
throw new RuntimeException(
"This JAR is compiled with too old a version of the data-model, please recompile with Jan 2014 (rc2) onwards");
}
// Now load the XML into a configuration object:
Configuration config = new Configuration();
// Add the client configuration overrides:
if (!bLocalMode) {
String hadoopConfigPath = props_custom.getHadoopConfigPath() + "/hadoop/";
config.addResource(new Path(hadoopConfigPath + "core-site.xml"));
config.addResource(new Path(hadoopConfigPath + "mapred-site.xml"));
config.addResource(new Path(hadoopConfigPath + "hadoop-site.xml"));
} //TESTED
try {
DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance();
DocumentBuilder dBuilder = dbFactory.newDocumentBuilder();
Document doc = dBuilder.parse(new ByteArrayInputStream(xml.toString().getBytes()));
NodeList nList = doc.getElementsByTagName("property");
for (int temp = 0; temp < nList.getLength(); temp++) {
Node nNode = nList.item(temp);
if (nNode.getNodeType() == Node.ELEMENT_NODE) {
Element eElement = (Element) nNode;
String name = getTagValue("name", eElement);
String value = getTagValue("value", eElement);
if ((null != name) && (null != value)) {
config.set(name, value);
}
}
}
} catch (Exception e) {
throw new IOException(e.getMessage());
}
// Some other config defaults:
// (not sure if these are actually applied, or derived from the defaults - for some reason they don't appear in CDH's client config)
config.set("mapred.map.tasks.speculative.execution", "false");
config.set("mapred.reduce.tasks.speculative.execution", "false");
// (default security is ignored here, have it set via HADOOP_TASKTRACKER_CONF in cloudera)
// Now run the JAR file
try {
BasicDBObject advancedConfigurationDbo = null;
try {
advancedConfigurationDbo = (null != job.query)
? ((BasicDBObject) com.mongodb.util.JSON.parse(job.query))
: (new BasicDBObject());
} catch (Exception e) {
advancedConfigurationDbo = new BasicDBObject();
}
boolean esMode = advancedConfigurationDbo.containsField("qt") && !job.isCustomTable;
if (esMode && !job.inputCollection.equals("doc_metadata.metadata")) {
throw new RuntimeException(
"Infinit.e Queries are only supported on doc_metadata - use MongoDB queries instead.");
}
config.setBoolean("mapred.used.genericoptionsparser", true); // (just stops an annoying warning from appearing)
if (bLocalMode) { // local job tracker and FS mode
config.set("mapred.job.tracker", "local");
config.set("fs.default.name", "local");
} else {
if (bTestMode) { // run job tracker locally but FS mode remotely
config.set("mapred.job.tracker", "local");
} else { // normal job tracker
String trackerUrl = HadoopUtils.getXMLProperty(
props_custom.getHadoopConfigPath() + "/hadoop/mapred-site.xml", "mapred.job.tracker");
config.set("mapred.job.tracker", trackerUrl);
}
String fsUrl = HadoopUtils.getXMLProperty(
props_custom.getHadoopConfigPath() + "/hadoop/core-site.xml", "fs.default.name");
config.set("fs.default.name", fsUrl);
}
if (!dataModelLoaded && !(bTestMode || bLocalMode)) { // If running distributed and no data model loaded then add ourselves
Path jarToCache = InfiniteHadoopUtils.cacheLocalFile("/opt/infinite-home/lib/",
"infinit.e.data_model.jar", config);
DistributedCache.addFileToClassPath(jarToCache, config);
jarToCache = InfiniteHadoopUtils.cacheLocalFile("/opt/infinite-home/lib/",
"infinit.e.processing.custom.library.jar", config);
DistributedCache.addFileToClassPath(jarToCache, config);
} //TESTED
// Debug scripts (only if they exist), and only in non local/test mode
if (!bLocalMode && !bTestMode) {
try {
Path scriptToCache = InfiniteHadoopUtils.cacheLocalFile("/opt/infinite-home/scripts/",
"custom_map_error_handler.sh", config);
config.set("mapred.map.task.debug.script", "custom_map_error_handler.sh " + job.jobtitle);
config.set("mapreduce.map.debug.script", "custom_map_error_handler.sh " + job.jobtitle);
DistributedCache.createSymlink(config);
DistributedCache.addCacheFile(scriptToCache.toUri(), config);
} catch (Exception e) {
} // just carry on
try {
Path scriptToCache = InfiniteHadoopUtils.cacheLocalFile("/opt/infinite-home/scripts/",
"custom_reduce_error_handler.sh", config);
config.set("mapred.reduce.task.debug.script", "custom_reduce_error_handler.sh " + job.jobtitle);
config.set("mapreduce.reduce.debug.script", "custom_reduce_error_handler.sh " + job.jobtitle);
DistributedCache.createSymlink(config);
DistributedCache.addCacheFile(scriptToCache.toUri(), config);
} catch (Exception e) {
} // just carry on
} //TODO (???): TOTEST
// (need to do these 2 things here before the job is created, at which point the config class has been copied across)
//1)
Class<?> mapperClazz = Class.forName(job.mapper, true, child);
if (ICustomInfiniteInternalEngine.class.isAssignableFrom(mapperClazz)) { // Special case: internal custom engine, so gets an additional integration hook
ICustomInfiniteInternalEngine preActivities = (ICustomInfiniteInternalEngine) mapperClazz
.newInstance();
preActivities.preTaskActivities(job._id, job.communityIds, config, !(bTestMode || bLocalMode));
} //TESTED
//2)
if (job.inputCollection.equalsIgnoreCase("file.binary_shares")) {
// Need to download the GridFSZip file
try {
Path jarToCache = InfiniteHadoopUtils.cacheLocalFile("/opt/infinite-home/lib/unbundled/",
"GridFSZipFile.jar", config);
DistributedCache.addFileToClassPath(jarToCache, config);
} catch (Throwable t) {
} // (this is fine, will already be on the classpath .. otherwise lots of other stuff will be failing all over the place!)
}
if (job.inputCollection.equals("records")) {
InfiniteElasticsearchHadoopUtils.handleElasticsearchInput(job, config, advancedConfigurationDbo);
//(won't run under 0.19 so running with "records" should cause all sorts of exceptions)
} //TESTED (by hand)
if (bTestMode || bLocalMode) { // If running locally, turn "snappy" off - tomcat isn't pointing its native library path in the right place
config.set("mapred.map.output.compression.codec", "org.apache.hadoop.io.compress.DefaultCodec");
}
// Manually specified caches
List<URL> localJarCaches = InfiniteHadoopUtils.handleCacheList(advancedConfigurationDbo.get("$caches"),
job, config, props_custom);
Job hj = new Job(config); // (NOTE: from here, changes to config are ignored)
try {
if (null != localJarCaches) {
if (bLocalMode || bTestMode) {
Method method = URLClassLoader.class.getDeclaredMethod("addURL", new Class[] { URL.class });
method.setAccessible(true);
method.invoke(child, localJarCaches.toArray());
} //TOTEST (tested logically)
}
Class<?> classToLoad = Class.forName(job.mapper, true, child);
hj.setJarByClass(classToLoad);
if (job.inputCollection.equalsIgnoreCase("filesystem")) {
String inputPath = null;
try {
inputPath = MongoDbUtil.getProperty(advancedConfigurationDbo, "file.url");
if (!inputPath.endsWith("/")) {
inputPath = inputPath + "/";
}
} catch (Exception e) {
}
if (null == inputPath) {
throw new RuntimeException("Must specify 'file.url' if reading from filesystem.");
}
inputPath = InfiniteHadoopUtils.authenticateInputDirectory(job, inputPath);
InfiniteFileInputFormat.addInputPath(hj, new Path(inputPath + "*/*")); // (that extra bit makes it recursive)
InfiniteFileInputFormat.setMaxInputSplitSize(hj, 33554432); // (32MB)
InfiniteFileInputFormat.setInfiniteInputPathFilter(hj, config);
hj.setInputFormatClass((Class<? extends InputFormat>) Class.forName(
"com.ikanow.infinit.e.data_model.custom.InfiniteFileInputFormat", true, child));
} else if (job.inputCollection.equalsIgnoreCase("file.binary_shares")) {
String[] oidStrs = null;
try {
String inputPath = MongoDbUtil.getProperty(advancedConfigurationDbo, "file.url");
Pattern oidExtractor = Pattern.compile("inf://share/([^/]+)");
Matcher m = oidExtractor.matcher(inputPath);
if (m.find()) {
oidStrs = m.group(1).split("\\s*,\\s*");
} else {
throw new RuntimeException(
"file.url must be in format inf://share/<oid-list>/<string>: " + inputPath);
}
InfiniteHadoopUtils.authenticateShareList(job, oidStrs);
} catch (Exception e) {
throw new RuntimeException(
"Authentication error: " + e.getMessage() + ": " + advancedConfigurationDbo, e);
}
hj.getConfiguration().setStrings("mapred.input.dir", oidStrs);
hj.setInputFormatClass((Class<? extends InputFormat>) Class.forName(
"com.ikanow.infinit.e.data_model.custom.InfiniteShareInputFormat", true, child));
} else if (job.inputCollection.equals("records")) {
hj.setInputFormatClass((Class<? extends InputFormat>) Class
.forName("com.ikanow.infinit.e.data_model.custom.InfiniteEsInputFormat", true, child));
} else {
if (esMode) {
hj.setInputFormatClass((Class<? extends InputFormat>) Class.forName(
"com.ikanow.infinit.e.processing.custom.utils.InfiniteElasticsearchMongoInputFormat",
true, child));
} else {
hj.setInputFormatClass((Class<? extends InputFormat>) Class.forName(
"com.ikanow.infinit.e.data_model.custom.InfiniteMongoInputFormat", true, child));
}
}
if ((null != job.exportToHdfs) && job.exportToHdfs) {
//TODO (INF-2469): Also, if the output key is BSON then also run as text (but output as JSON?)
Path outPath = InfiniteHadoopUtils.ensureOutputDirectory(job, props_custom);
if ((null != job.outputKey) && (null != job.outputValue)
&& job.outputKey.equalsIgnoreCase("org.apache.hadoop.io.text")
&& job.outputValue.equalsIgnoreCase("org.apache.hadoop.io.text")) {
// (slight hack before I sort out the horrendous job class - if key/val both text and exporting to HDFS then output as Text)
hj.setOutputFormatClass((Class<? extends OutputFormat>) Class
.forName("org.apache.hadoop.mapreduce.lib.output.TextOutputFormat", true, child));
TextOutputFormat.setOutputPath(hj, outPath);
} //TESTED
else {
hj.setOutputFormatClass((Class<? extends OutputFormat>) Class.forName(
"org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat", true, child));
SequenceFileOutputFormat.setOutputPath(hj, outPath);
} //TESTED
} else { // normal case, stays in MongoDB
hj.setOutputFormatClass((Class<? extends OutputFormat>) Class.forName(
"com.ikanow.infinit.e.data_model.custom.InfiniteMongoOutputFormat", true, child));
}
hj.setMapperClass((Class<? extends Mapper>) mapperClazz);
String mapperOutputKeyOverride = advancedConfigurationDbo.getString("$mapper_key_class", null);
if (null != mapperOutputKeyOverride) {
hj.setMapOutputKeyClass(Class.forName(mapperOutputKeyOverride));
} //TESTED
String mapperOutputValueOverride = advancedConfigurationDbo.getString("$mapper_value_class", null);
if (null != mapperOutputValueOverride) {
hj.setMapOutputValueClass(Class.forName(mapperOutputValueOverride));
} //TESTED
if ((null != job.reducer) && !job.reducer.startsWith("#") && !job.reducer.equalsIgnoreCase("null")
&& !job.reducer.equalsIgnoreCase("none")) {
hj.setReducerClass((Class<? extends Reducer>) Class.forName(job.reducer, true, child));
// Variable reducers:
if (null != job.query) {
try {
hj.setNumReduceTasks(advancedConfigurationDbo.getInt("$reducers", 1));
} catch (Exception e) {
try {
// (just check it's not a string that is a valid int)
hj.setNumReduceTasks(
Integer.parseInt(advancedConfigurationDbo.getString("$reducers", "1")));
} catch (Exception e2) {
}
}
} //TESTED
} else {
hj.setNumReduceTasks(0);
}
if ((null != job.combiner) && !job.combiner.startsWith("#")
&& !job.combiner.equalsIgnoreCase("null") && !job.combiner.equalsIgnoreCase("none")) {
hj.setCombinerClass((Class<? extends Reducer>) Class.forName(job.combiner, true, child));
}
hj.setOutputKeyClass(Class.forName(job.outputKey, true, child));
hj.setOutputValueClass(Class.forName(job.outputValue, true, child));
hj.setJobName(job.jobtitle);
currJobName = job.jobtitle;
} catch (Error e) { // (messing about with class loaders = lots of chances for errors!)
throw new RuntimeException(e.getMessage(), e);
}
if (bTestMode || bLocalMode) {
hj.submit();
currThreadId = null;
Logger.getRootLogger().addAppender(this);
currLocalJobId = hj.getJobID().toString();
currLocalJobErrs.setLength(0);
while (!hj.isComplete()) {
Thread.sleep(1000);
}
Logger.getRootLogger().removeAppender(this);
if (hj.isSuccessful()) {
if (this.currLocalJobErrs.length() > 0) {
return "local_done: " + this.currLocalJobErrs.toString();
} else {
return "local_done";
}
} else {
return "Error: " + this.currLocalJobErrs.toString();
}
} else {
hj.submit();
String jobId = hj.getJobID().toString();
return jobId;
}
} catch (Exception e) {
e.printStackTrace();
Thread.currentThread().setContextClassLoader(savedClassLoader);
return "Error: " + InfiniteHadoopUtils.createExceptionMessage(e);
} finally {
Thread.currentThread().setContextClassLoader(savedClassLoader);
}
}
From source file:com.inmobi.conduit.local.LocalStreamService.java
License:Apache License
@Override protected void execute() throws Exception { lastProcessedFile.clear();//from w ww . j a v a 2s . c om List<AuditMessage> auditMsgList = new ArrayList<AuditMessage>(); try { FileSystem fs = FileSystem.get(srcCluster.getHadoopConf()); // Cleanup tmpPath before everyRun to avoid // any old data being used in this run if the old run was aborted cleanUpTmp(fs); LOG.info("TmpPath is [" + tmpPath + "]"); long commitTime = srcCluster.getCommitTime(); publishMissingPaths(fs, srcCluster.getLocalFinalDestDirRoot(), commitTime, streamsToProcess); Map<FileStatus, String> fileListing = new TreeMap<FileStatus, String>(); Set<FileStatus> trashSet = new HashSet<FileStatus>(); /* checkpointPaths table contains streamname as rowkey, source(collector) name as column key and checkpoint value as value */ Table<String, String, String> checkpointPaths = HashBasedTable.create(); long totalSize = createMRInput(tmpJobInputPath, fileListing, trashSet, checkpointPaths); if (fileListing.size() == 0) { LOG.info("Nothing to do!"); for (String eachStream : streamsToProcess) { if (lastProcessedFile.get(eachStream) != null) { ConduitMetrics.updateAbsoluteGauge(getServiceType(), LAST_FILE_PROCESSED, eachStream, lastProcessedFile.get(eachStream)); } } return; } Job job = createJob(tmpJobInputPath, totalSize); long jobStartTime = System.nanoTime(); job.waitForCompletion(true); long jobExecutionTimeInSecs = (System.nanoTime() - jobStartTime) / (NANO_SECONDS_IN_SECOND); LOG.info("Time taken to complete " + job.getJobID() + " job : " + jobExecutionTimeInSecs + "secs"); updateJobTimeCounter(jobExecutionTimeInSecs); if (job.isSuccessful()) { commitTime = srcCluster.getCommitTime(); LOG.info("Commiting mvPaths and ConsumerPaths"); commit(prepareForCommit(commitTime), false, auditMsgList, commitTime); updatePathsTobeRegisteredWithLatestDir(commitTime); checkPoint(checkpointPaths); LOG.info("Commiting trashPaths"); commit(populateTrashCommitPaths(trashSet), true, null, commitTime); LOG.info("Committed successfully at " + getLogDateString(commitTime)); for (String eachStream : streamsToProcess) { if (lastProcessedFile.get(eachStream) != null) { ConduitMetrics.updateAbsoluteGauge(getServiceType(), LAST_FILE_PROCESSED, eachStream, lastProcessedFile.get(eachStream)); } } } else { throw new IOException("LocaStreamService job failure: Job " + job.getJobID() + " has failed. "); } } catch (Exception e) { LOG.warn("Error in running LocalStreamService ", e); throw e; } finally { publishAuditMessages(auditMsgList); try { registerPartitions(); } catch (Exception e) { LOG.warn("Got exception while registering partitions. ", e); } } }
From source file:com.inmobi.databus.local.LocalStreamService.java
License:Apache License
@Override protected void execute() throws Exception { try {//from w w w. java 2 s.c o m FileSystem fs = FileSystem.get(cluster.getHadoopConf()); // Cleanup tmpPath before everyRun to avoid // any old data being used in this run if the old run was aborted cleanUpTmp(fs); LOG.info("TmpPath is [" + tmpPath + "]"); publishMissingPaths(fs, cluster.getLocalFinalDestDirRoot()); Map<FileStatus, String> fileListing = new TreeMap<FileStatus, String>(); Set<FileStatus> trashSet = new HashSet<FileStatus>(); // checkpointKey, CheckPointPath Map<String, FileStatus> checkpointPaths = new TreeMap<String, FileStatus>(); createMRInput(tmpJobInputPath, fileListing, trashSet, checkpointPaths); if (fileListing.size() == 0) { LOG.info("Nothing to do!"); return; } Job job = createJob(tmpJobInputPath); job.waitForCompletion(true); if (job.isSuccessful()) { long commitTime = cluster.getCommitTime(); LOG.info("Commiting mvPaths and ConsumerPaths"); commit(prepareForCommit(commitTime, fileListing)); checkPoint(checkpointPaths); LOG.info("Commiting trashPaths"); commit(populateTrashCommitPaths(trashSet)); LOG.info("Committed successfully at " + getLogDateString(commitTime)); } } catch (Exception e) { LOG.warn("Error in running LocalStreamService " + e); throw e; } }