List of usage examples for org.apache.hadoop.mapred JobConf setBoolean
public void setBoolean(String name, boolean value)
name property to a boolean. From source file:com.yolodata.tbana.cascading.csv.CSVLine.java
License:Open Source License
@Override public void sourceConfInit(FlowProcess<JobConf> flowProcess, Tap<JobConf, RecordReader, OutputCollector> tap, JobConf conf) { if (hasZippedFiles(FileInputFormat.getInputPaths(conf))) throw new IllegalStateException( "cannot read zip files: " + Arrays.toString(FileInputFormat.getInputPaths(conf))); conf.set(CSVLineRecordReader.FORMAT_DELIMITER, CSVLineRecordReader.DEFAULT_DELIMITER); conf.set(CSVLineRecordReader.FORMAT_SEPARATOR, CSVLineRecordReader.DEFAULT_SEPARATOR); conf.setBoolean(CSVLineRecordReader.IS_ZIPFILE, false); conf.setInt(CSVNLineInputFormat.LINES_PER_MAP, 40000); conf.setInputFormat(CSVNLineInputFormat.class); }
From source file:de.l3s.streamcorpus.mapreduce.TerrierIndexing.java
License:Mozilla Public License
/** Starts the MapReduce indexing. * @param args//w w w. ja v a 2 s.com * @throws Exception */ public int run(String[] args) throws Exception { long time = System.currentTimeMillis(); // For the moment: Hard-code the terrier home to quick test System.setProperty("terrier.home", "/home/tuan.tran/executable/StreamCorpusIndexer"); boolean docPartitioned = false; int numberOfReducers = Integer .parseInt(ApplicationSetup.getProperty("terrier.hadoop.indexing.reducers", "26")); final HadoopPlugin.JobFactory jf = HadoopPlugin.getJobFactory("HOD-TerrierIndexing"); if (args.length == 2 && args[0].equals("-p")) { logger.debug("Document-partitioned Mode, " + numberOfReducers + " output indices."); numberOfReducers = Integer.parseInt(args[1]); docPartitioned = true; } else if (args.length == 1 && args[0].equals("--merge")) { if (numberOfReducers > 1) mergeLexiconInvertedFiles(ApplicationSetup.TERRIER_INDEX_PATH, numberOfReducers); else logger.error("No point merging 1 reduce task output"); return 0; } else if (args.length == 0) { logger.debug("Term-partitioned Mode, " + numberOfReducers + " reducers creating one inverted index."); docPartitioned = false; if (numberOfReducers > MAX_REDUCE) { logger.warn("Excessive reduce tasks (" + numberOfReducers + ") in use " + "- SplitEmittedTerm.SETPartitionerLowercaseAlphaTerm can use " + MAX_REDUCE + " at most"); } } /*else { logger.fatal(usage()); return 0; }*/ if (!(CompressionFactory.getCompressionConfiguration("inverted", new String[0], false) instanceof BitCompressionConfiguration)) { logger.error("Sorry, only default BitCompressionConfiguration is supported by HadoopIndexing" + " - you can recompress the inverted index later using IndexRecompressor"); return 0; } if (jf == null) throw new Exception("Could not get JobFactory from HadoopPlugin"); final JobConf conf = jf.newJob(); conf.setJarByClass(TerrierIndexing.class); conf.setJobName("StreamCorpusIndexer: Terrier Indexing"); if (Files.exists(ApplicationSetup.TERRIER_INDEX_PATH) && Index.existsIndex(ApplicationSetup.TERRIER_INDEX_PATH, ApplicationSetup.TERRIER_INDEX_PREFIX)) { logger.fatal("Cannot index while index exists at " + ApplicationSetup.TERRIER_INDEX_PATH + "," + ApplicationSetup.TERRIER_INDEX_PREFIX); return 0; } // boolean blockIndexing = ApplicationSetup.BLOCK_INDEXING; boolean blockIndexing = true; if (blockIndexing) { conf.setMapperClass(Hadoop_BlockSinglePassIndexer.class); conf.setReducerClass(Hadoop_BlockSinglePassIndexer.class); } else { conf.setMapperClass(Hadoop_BasicSinglePassIndexer.class); conf.setReducerClass(Hadoop_BasicSinglePassIndexer.class); } FileOutputFormat.setOutputPath(conf, new Path(ApplicationSetup.TERRIER_INDEX_PATH)); conf.set("indexing.hadoop.prefix", ApplicationSetup.TERRIER_INDEX_PREFIX); conf.setMapOutputKeyClass(SplitEmittedTerm.class); conf.setMapOutputValueClass(MapEmittedPostingList.class); conf.setBoolean("indexing.hadoop.multiple.indices", docPartitioned); if (!conf.get("mapred.job.tracker").equals("local")) { conf.setMapOutputCompressorClass(GzipCodec.class); conf.setCompressMapOutput(true); } else { conf.setCompressMapOutput(false); } conf.setInputFormat(MultiFileCollectionInputFormat.class); conf.setOutputFormat(NullOutputFormat.class); conf.setOutputKeyComparatorClass(SplitEmittedTerm.SETRawComparatorTermSplitFlush.class); conf.setOutputValueGroupingComparator(SplitEmittedTerm.SETRawComparatorTerm.class); conf.setReduceSpeculativeExecution(false); //parse the collection.spec BufferedReader specBR = Files.openFileReader(ApplicationSetup.COLLECTION_SPEC); String line = null; List<Path> paths = new ArrayList<Path>(); while ((line = specBR.readLine()) != null) { if (line.startsWith("#")) continue; paths.add(new Path(line)); } specBR.close(); FileInputFormat.setInputPaths(conf, paths.toArray(new Path[paths.size()])); // not sure if this is effective in YARN conf.setNumMapTasks(2000); // increase the heap usage conf.set("mapreduce.map.memory.mb", "6100"); conf.set("mapred.job.map.memory.mb", "6100"); conf.set("mapreduce.reduce.memory.mb", "6144"); conf.set("mapred.job.reduce.memory.mb", "6144"); conf.set("mapreduce.map.java.opts", "-Xmx6100m"); conf.set("mapred.map.child.java.opts", "-Xmx6100m"); conf.set("mapreduce.reduce.java.opts", "-Xmx6144m"); conf.set("mapred.reduce.child.opts", "-Xmx6144m"); //conf.setBoolean("mapred.used.genericoptionsparser", true) ; // This is the nasty thing in MapReduce v2 and YARN: They always prefer their ancient jars first. Set this on to say you don't like it conf.set("mapreduce.job.user.classpath.first", "true"); // increase the yarn memory to 10 GB conf.set("yarn.nodemanager.resource.memory-mb", "12288"); conf.set("yarn.nodemanager.resource.cpu-vcores", "16"); conf.set("yarn.scheduler.minimum-allocation-mb", "4096"); conf.setNumReduceTasks(numberOfReducers); if (numberOfReducers > 1) { if (docPartitioned) conf.setPartitionerClass(SplitEmittedTerm.SETPartitioner.class); else conf.setPartitionerClass(SplitEmittedTerm.SETPartitionerLowercaseAlphaTerm.class); } else { //for JUnit tests, we seem to need to restore the original partitioner class conf.setPartitionerClass(HashPartitioner.class); } /*JobID jobId = null; boolean ranOK = true; try{ RunningJob rj = JobClient.runJob(conf); jobId = rj.getID(); HadoopUtility.finishTerrierJob(conf); } catch (Exception e) { logger.error("Problem running job", e); e.printStackTrace(); ranOK = false; } if (jobId != null) { deleteTaskFiles(ApplicationSetup.TERRIER_INDEX_PATH, jobId); } */ //if (ranOK) //{ System.out.println("Merging indices"); if (!docPartitioned) { if (numberOfReducers > 1) mergeLexiconInvertedFiles(ApplicationSetup.TERRIER_INDEX_PATH, numberOfReducers); } Hadoop_BasicSinglePassIndexer.finish(ApplicationSetup.TERRIER_INDEX_PATH, docPartitioned ? numberOfReducers : 1, jf); //} System.out.println("Time Taken = " + ((System.currentTimeMillis() - time) / 1000) + " seconds"); jf.close(); return 0; }
From source file:de.l3s.streamcorpus.StreamCorpusIndexing.java
License:Mozilla Public License
/** Starts the MapReduce indexing. * @param args/* w w w.j av a2 s .com*/ * @throws Exception */ public int run(String[] args) throws Exception { long time = System.currentTimeMillis(); // For the moment: Hard-code the terrier home to quick test System.setProperty("terrier.home", "/home/tuan.tran/executable/StreamCorpusIndexer"); boolean docPartitioned = false; int numberOfReducers = Integer .parseInt(ApplicationSetup.getProperty("terrier.hadoop.indexing.reducers", "26")); final HadoopPlugin.JobFactory jf = HadoopPlugin.getJobFactory("HOD-TerrierIndexing"); if (args.length == 2 && args[0].equals("-p")) { logger.debug("Document-partitioned Mode, " + numberOfReducers + " output indices."); numberOfReducers = Integer.parseInt(args[1]); docPartitioned = true; } else if (args.length == 1 && args[0].equals("--merge")) { if (numberOfReducers > 1) mergeLexiconInvertedFiles(ApplicationSetup.TERRIER_INDEX_PATH, numberOfReducers); else logger.error("No point merging 1 reduce task output"); return 0; } else if (args.length == 0) { logger.debug("Term-partitioned Mode, " + numberOfReducers + " reducers creating one inverted index."); docPartitioned = false; if (numberOfReducers > MAX_REDUCE) { logger.warn("Excessive reduce tasks (" + numberOfReducers + ") in use " + "- SplitEmittedTerm.SETPartitionerLowercaseAlphaTerm can use " + MAX_REDUCE + " at most"); } } /*else { logger.fatal(usage()); return 0; }*/ if (!(CompressionFactory.getCompressionConfiguration("inverted", new String[0], false) instanceof BitCompressionConfiguration)) { logger.error("Sorry, only default BitCompressionConfiguration is supported by HadoopIndexing" + " - you can recompress the inverted index later using IndexRecompressor"); return 0; } if (jf == null) throw new Exception("Could not get JobFactory from HadoopPlugin"); final JobConf conf = jf.newJob(); conf.setJarByClass(StreamCorpusIndexing.class); conf.setJobName("StreamCorpusIndexer: Terrier Indexing"); if (Files.exists(ApplicationSetup.TERRIER_INDEX_PATH) && Index.existsIndex(ApplicationSetup.TERRIER_INDEX_PATH, ApplicationSetup.TERRIER_INDEX_PREFIX)) { logger.fatal("Cannot index while index exists at " + ApplicationSetup.TERRIER_INDEX_PATH + "," + ApplicationSetup.TERRIER_INDEX_PREFIX); return 0; } // boolean blockIndexing = ApplicationSetup.BLOCK_INDEXING; boolean blockIndexing = true; if (blockIndexing) { conf.setMapperClass(Hadoop_BlockSinglePassIndexer.class); conf.setReducerClass(Hadoop_BlockSinglePassIndexer.class); } else { conf.setMapperClass(Hadoop_BasicSinglePassIndexer.class); conf.setReducerClass(Hadoop_BasicSinglePassIndexer.class); } FileOutputFormat.setOutputPath(conf, new Path(ApplicationSetup.TERRIER_INDEX_PATH)); conf.set("indexing.hadoop.prefix", ApplicationSetup.TERRIER_INDEX_PREFIX); conf.setMapOutputKeyClass(SplitEmittedTerm.class); conf.setMapOutputValueClass(MapEmittedPostingList.class); conf.setBoolean("indexing.hadoop.multiple.indices", docPartitioned); if (!conf.get("mapred.job.tracker").equals("local")) { conf.setMapOutputCompressorClass(GzipCodec.class); conf.setCompressMapOutput(true); } else { conf.setCompressMapOutput(false); } conf.setInputFormat(MultiFileCollectionInputFormat.class); conf.setOutputFormat(NullOutputFormat.class); conf.setOutputKeyComparatorClass(SplitEmittedTerm.SETRawComparatorTermSplitFlush.class); conf.setOutputValueGroupingComparator(SplitEmittedTerm.SETRawComparatorTerm.class); conf.setReduceSpeculativeExecution(false); //parse the collection.spec BufferedReader specBR = Files.openFileReader(ApplicationSetup.COLLECTION_SPEC); String line = null; List<Path> paths = new ArrayList<Path>(); while ((line = specBR.readLine()) != null) { if (line.startsWith("#")) continue; paths.add(new Path(line)); } specBR.close(); FileInputFormat.setInputPaths(conf, paths.toArray(new Path[paths.size()])); // not sure if this is effective in YARN conf.setNumMapTasks(2000); // increase the heap usage conf.set("mapreduce.map.memory.mb", "6100"); conf.set("mapred.job.map.memory.mb", "6100"); conf.set("mapreduce.reduce.memory.mb", "6144"); conf.set("mapred.job.reduce.memory.mb", "6144"); conf.set("mapreduce.map.java.opts", "-Xmx6100m"); conf.set("mapred.map.child.java.opts", "-Xmx6100m"); conf.set("mapreduce.reduce.java.opts", "-Xmx6144m"); conf.set("mapred.reduce.child.opts", "-Xmx6144m"); //conf.setBoolean("mapred.used.genericoptionsparser", true) ; // This is the nasty thing in MapReduce v2 and YARN: They always prefer their ancient jars first. Set this on to say you don't like it conf.set("mapreduce.job.user.classpath.first", "true"); // increase the yarn memory to 10 GB conf.set("yarn.nodemanager.resource.memory-mb", "12288"); conf.set("yarn.nodemanager.resource.cpu-vcores", "16"); conf.set("yarn.scheduler.minimum-allocation-mb", "4096"); conf.setNumReduceTasks(numberOfReducers); if (numberOfReducers > 1) { if (docPartitioned) conf.setPartitionerClass(SplitEmittedTerm.SETPartitioner.class); else conf.setPartitionerClass(SplitEmittedTerm.SETPartitionerLowercaseAlphaTerm.class); } else { //for JUnit tests, we seem to need to restore the original partitioner class conf.setPartitionerClass(HashPartitioner.class); } /*JobID jobId = null; boolean ranOK = true; try{ RunningJob rj = JobClient.runJob(conf); jobId = rj.getID(); HadoopUtility.finishTerrierJob(conf); } catch (Exception e) { logger.error("Problem running job", e); e.printStackTrace(); ranOK = false; } if (jobId != null) { deleteTaskFiles(ApplicationSetup.TERRIER_INDEX_PATH, jobId); } */ //if (ranOK) //{ System.out.println("Merging indices"); if (!docPartitioned) { if (numberOfReducers > 1) mergeLexiconInvertedFiles(ApplicationSetup.TERRIER_INDEX_PATH, numberOfReducers); } Hadoop_BasicSinglePassIndexer.finish(ApplicationSetup.TERRIER_INDEX_PATH, docPartitioned ? numberOfReducers : 1, jf); //} System.out.println("Time Taken = " + ((System.currentTimeMillis() - time) / 1000) + " seconds"); jf.close(); return 0; }
From source file:edu.iu.examples.ExamplesMain.java
License:Apache License
private Job configureBenchmarkJob(String cmd, int bytesPerPartition, int numPartitions, int numMappers, int numIterations, boolean verify, Path inputDirPath, Path outputDirPath) throws IOException { Job job = Job.getInstance(getConf(), "example_job"); FileInputFormat.setInputPaths(job, inputDirPath); FileOutputFormat.setOutputPath(job, outputDirPath); job.setInputFormatClass(SingleFileInputFormat.class); job.setJarByClass(ExamplesMain.class); if (cmd.equals("allreduce")) { job.setMapperClass(AllReduce.class); } else if (cmd.equals("allgather")) { job.setMapperClass(AllGather.class); } else if (cmd.equals("reduce")) { job.setMapperClass(Reduce.class); } else if (cmd.equals("bcast")) { job.setMapperClass(BCast.class); } else if (cmd.equals("rotate")) { job.setMapperClass(Rotate.class); }//ww w . ja va 2s . c o m org.apache.hadoop.mapred.JobConf jobConf = (JobConf) job.getConfiguration(); jobConf.set("mapreduce.framework.name", "map-collective"); jobConf.setNumMapTasks(numMappers); job.setNumReduceTasks(0); jobConf.set(Constants.ARGS_OPERATION, cmd); jobConf.setInt(Constants.ARGS_ELEMENTS, bytesPerPartition); jobConf.setInt(Constants.ARGS_PARTITIONS, numPartitions); jobConf.setInt(Constants.ARGS_MAPPERS, numMappers); jobConf.setInt(Constants.ARGS_ITERATIONS, numIterations); jobConf.setBoolean(Constants.ARGS_VERIFY, verify); return job; }
From source file:edu.stolaf.cs.wmrserver.HadoopEngine.java
License:Apache License
public void submit(JobRequest request, long submissionID, File mapperFile, File reducerFile, File packageDir, Path inputPath) throws ValidationException, NotFoundException, CompilationException, InternalException { // Generate job output path Path outputDir = new Path(_homeDir, "out"); Path outputPath;/*from w ww .jav a 2 s . c o m*/ try { FileSystem fs = outputDir.getFileSystem(new Configuration()); outputPath = JobServiceHandler.getNonexistantPath(outputDir, request.getName(), fs); } catch (IOException ex) { throw JobServiceHandler.wrapException("Could not construct output path.", ex); } JobConf conf = new JobConf(); conf.setJobName(request.getName()); // Set mapper and number of tasks if specified StreamJob.setStreamMapper(conf, mapperFile.toString()); if (request.isSetMapTasks()) conf.setNumMapTasks(request.getMapTasks()); // Set reducer and number of tasks if specified StreamJob.setStreamReducer(conf, reducerFile.toString()); if (request.isSetReduceTasks()) conf.setNumReduceTasks(request.getReduceTasks()); // Create and set job JAR, including necessary files ArrayList<String> jarFiles = new ArrayList<String>(); jarFiles.add(packageDir.toString()); String jarPath; try { jarPath = StreamJob.createJobJar(conf, jarFiles, _tempDir); } catch (IOException ex) { throw JobServiceHandler.wrapException("Could not create job jar.", ex); } if (jarPath != null) conf.setJar(jarPath); // TODO: This is a hack. Rewrite streaming to use DistributedCache. //conf.setPattern("mapreduce.job.jar.unpack.pattern", // Pattern.compile(".*")); // Set I/O formats and paths conf.setInputFormat(KeyValueTextInputFormat.class); conf.setOutputFormat(TextOutputFormat.class); conf.setOutputKeyClass(Text.class); conf.setOutputValueClass(Text.class); FileInputFormat.addInputPath(conf, inputPath); FileOutputFormat.setOutputPath(conf, outputPath); // Use numeric sort if appropriate conf.setBoolean(CONF_NUMERIC, request.isNumericSort()); if (request.isNumericSort()) { conf.setOutputKeyComparatorClass(KeyFieldBasedComparator.class); conf.setPartitionerClass(KeyFieldBasedPartitioner.class); conf.setKeyFieldComparatorOptions("-n"); conf.setKeyFieldPartitionerOptions("-n"); } // Set other job information conf.set(CONF_USER, request.getUser()); conf.set(CONF_LANGUAGE, request.getLanguage()); conf.set(CONF_MAPPER, request.getMapper()); conf.set(CONF_REDUCER, request.getReducer()); // Attempt to submit the job RunningJob job; try { JobClient client = new JobClient(new JobConf()); job = client.submitJob(conf); } catch (IOException ex) { throw JobServiceHandler.wrapException("There was a serious error while attempting to submit the job.", ex); } try { SubmissionDatabase.setSubmitted(submissionID); SubmissionDatabase.setHadoopID(submissionID, job.getID().toString()); } catch (SQLException ex) { throw JobServiceHandler.wrapException("Could not update submission in database.", ex); } }
From source file:edu.ucsb.cs.hybrid.HybridDriver.java
License:Apache License
public static void main(String args[]) throws ParseException, IOException { // job.set("mapred.job.tracker", "local"); // job.set("fs.default.name", "file:///"); JobConf job = new JobConf(); job.setJarByClass(HybridDriver.class); new GenericOptionsParser(job, args); setMapperAndRunner(job);/*from w ww . j a v a2 s. co m*/ job.setMapOutputKeyClass(DocDocWritable.class); job.setMapOutputValueClass(FloatWritable.class); job.setNumReduceTasks(0); job.setOutputKeyClass(DocDocWritable.class); job.setOutputValueClass(FloatWritable.class); Path inputPath = new Path(INPUT_DIR); CustomSequenceFileInputFormat.addInputPath(job, inputPath); Path outputPath = new Path(OUTPUT_DIR); job.setOutputFormat(SequenceFileOutputFormat.class); SequenceFileOutputFormat.setOutputPath(job, outputPath); FileSystem.get(job).delete(outputPath, true); job.setBoolean("fs.hdfs.impl.disable.cache", true); //xun not sure if needed if (job.getBoolean(Config.SPLITABLE_PROPERTY, Config.SPLITABLE_VALUE)) { job.setInputFormat(CustomSequenceFileInputFormat.class); Long splitMB = job.getLong(Config.SPLIT_MB_PROPERTY, Config.SPLIT_MB_VALUE) * 1024 * 1024; job.setLong("mapred.min.split.size", splitMB); job.setLong("mapred.max.split.size", splitMB); job.setLong("dfs.block.size", splitMB); } else { // Comment the following of splitter for www experiments it assumes no splitting // of partitions for load balancing, should be fixed. Splitter.configure(job, inputPath);// remove comment unless for www job.setInputFormat(NonSplitableSequenceInputFormat.class); //remove comment } //SIGIR'14 two-stage balancing //not yet fully incorporated if (job.getInt(Config.LOAD_BALANCE_PROPERTY, Config.LOAD_BALANCE_VALUE) != 0) { TwoStageLoadbalancing.main(job.getInt(Config.LOAD_BALANCE_PROPERTY, Config.LOAD_BALANCE_VALUE), new Path(PartDriver.OUTPUT_DIR), job); } JobSubmitter.run(job, "SIMILARITY", job.getFloat(Config.THRESHOLD_PROPERTY, Config.THRESHOLD_VALUE)); if (job.getBoolean(Config.CONVERT_TEXT_PROPERTY, Config.CONVERT_TEXT_VALUE)) IDMappingJob(args); }
From source file:edu.umd.cloud9.collection.wikipedia.BuildWikipediaDocnoMapping.java
License:Apache License
@SuppressWarnings("static-access") @Override// w w w . j a v a 2 s . c o m public int run(String[] args) throws Exception { Options options = new Options(); options.addOption( OptionBuilder.withArgName("path").hasArg().withDescription("XML dump file").create(INPUT_OPTION)); options.addOption(OptionBuilder.withArgName("path").hasArg().withDescription("tmp output directory") .create(OUTPUT_PATH_OPTION)); options.addOption(OptionBuilder.withArgName("path").hasArg().withDescription("output file") .create(OUTPUT_FILE_OPTION)); options.addOption(OptionBuilder.withArgName("en|sv|de|cs|es|zh|ar|tr").hasArg() .withDescription("two-letter language code").create(LANGUAGE_OPTION)); options.addOption(KEEP_ALL_OPTION, false, "keep all pages"); CommandLine cmdline; CommandLineParser parser = new GnuParser(); try { cmdline = parser.parse(options, args); } catch (ParseException exp) { System.err.println("Error parsing command line: " + exp.getMessage()); return -1; } if (!cmdline.hasOption(INPUT_OPTION) || !cmdline.hasOption(OUTPUT_PATH_OPTION) || !cmdline.hasOption(OUTPUT_FILE_OPTION)) { HelpFormatter formatter = new HelpFormatter(); formatter.printHelp(this.getClass().getName(), options); ToolRunner.printGenericCommandUsage(System.out); return -1; } String language = null; if (cmdline.hasOption(LANGUAGE_OPTION)) { language = cmdline.getOptionValue(LANGUAGE_OPTION); if (language.length() != 2) { System.err.println("Error: \"" + language + "\" unknown language!"); return -1; } } String inputPath = cmdline.getOptionValue(INPUT_OPTION); String outputPath = cmdline.getOptionValue(OUTPUT_PATH_OPTION); String outputFile = cmdline.getOptionValue(OUTPUT_FILE_OPTION); boolean keepAll = cmdline.hasOption(KEEP_ALL_OPTION); LOG.info("Tool name: " + this.getClass().getName()); LOG.info(" - input: " + inputPath); LOG.info(" - output path: " + outputPath); LOG.info(" - output file: " + outputFile); LOG.info(" - keep all pages: " + keepAll); LOG.info(" - language: " + language); JobConf conf = new JobConf(getConf(), BuildWikipediaDocnoMapping.class); conf.setJobName(String.format("BuildWikipediaDocnoMapping[%s: %s, %s: %s, %s: %s]", INPUT_OPTION, inputPath, OUTPUT_FILE_OPTION, outputFile, LANGUAGE_OPTION, language)); conf.setBoolean(KEEP_ALL_OPTION, keepAll); if (language != null) { conf.set("wiki.language", language); } conf.setNumReduceTasks(1); FileInputFormat.setInputPaths(conf, new Path(inputPath)); FileOutputFormat.setOutputPath(conf, new Path(outputPath)); FileOutputFormat.setCompressOutput(conf, false); conf.setInputFormat(WikipediaPageInputFormat.class); conf.setOutputKeyClass(IntWritable.class); conf.setOutputValueClass(IntWritable.class); conf.setOutputFormat(TextOutputFormat.class); conf.setMapperClass(MyMapper.class); conf.setReducerClass(MyReducer.class); // Delete the output directory if it exists already. FileSystem.get(conf).delete(new Path(outputPath), true); RunningJob job = JobClient.runJob(conf); Counters c = job.getCounters(); long cnt = keepAll ? c.getCounter(PageTypes.TOTAL) : c.getCounter(PageTypes.ARTICLE); WikipediaDocnoMapping.writeDocnoMappingData(FileSystem.get(conf), outputPath + "/part-00000", (int) cnt, outputFile); return 0; }
From source file:edu.umn.cs.spatialHadoop.operations.DistributedJoin.java
License:Open Source License
/** * Spatially joins two datasets by repartitioning the smaller dataset based * on the larger one, then apply one-to-one joining for each partition * /* ww w . j av a 2 s .c o m*/ * @author Ibrahim Sabek * @param inputFiles * Input datasets to be spatially joined * @param fileToRepartition * Index of which file will be repartitioned * @param outputFile * Output file contains the joining results * @param params * Job configurations * @return * @throws IOException */ protected static long repartitionJoinStep(final Path[] inputFiles, int fileToRepartition, Path outputFile, OperationsParams params) throws IOException { boolean overwrite = params.getBoolean("overwrite", false); Shape stockShape = params.getShape("shape"); // Do the repartition step long t1 = System.currentTimeMillis(); JobConf repartitionJoinJob = new JobConf(params, DistributedJoin.class); repartitionJoinJob.setJobName("RepartitionJoin"); FileSystem fs = inputFiles[fileToRepartition].getFileSystem(params); Path outputPath = outputFile; if (outputPath == null) { do { outputPath = new Path(inputFiles[0].getName() + ".dj_" + (int) (Math.random() * 1000000)); } while (fs.exists(outputPath)); } LOG.info("Repartition - Joining " + inputFiles[0] + " X " + inputFiles[1]); // Get the cells to use for repartitioning GlobalIndex<Partition> gindex = SpatialSite.getGlobalIndex(fs, inputFiles[1 - fileToRepartition]); OperationsParams.setRepartitionJoinIndexPath(repartitionJoinJob, RepartitionJoinIndexPath, inputFiles[1 - fileToRepartition]); OperationsParams.setInactiveModeFlag(repartitionJoinJob, InactiveMode, isReduceInactive); OperationsParams.setJoiningThresholdPerOnce(repartitionJoinJob, JoiningThresholdPerOnce, joiningThresholdPerOnce); OperationsParams.setFilterOnlyModeFlag(repartitionJoinJob, isFilterOnlyMode, isFilterOnly); CellInfo[] cellsInfo = SpatialSite.cellsOf(fs, inputFiles[1 - fileToRepartition]); // Repartition the file to match the other file boolean isReplicated = gindex.isReplicated(); boolean isCompact = gindex.isCompact(); String sindex; if (isReplicated && !isCompact) sindex = "grid"; else if (isReplicated && isCompact) sindex = "r+tree"; else if (!isReplicated && isCompact) sindex = "rtree"; else throw new RuntimeException("Unknown index at: " + inputFiles[1 - fileToRepartition]); params.set("sindex", sindex); // Decide which map function to use based on the type of global index if (sindex.equals("rtree") || sindex.equals("str")) { // Repartition without replication repartitionJoinJob.setMapperClass(RepartitionMapNoReplication.class); } else { // Repartition with replication (grid and r+tree) repartitionJoinJob.setMapperClass(RepartitionMap.class); } repartitionJoinJob.setMapOutputKeyClass(IntWritable.class); repartitionJoinJob.setMapOutputValueClass(stockShape.getClass()); ShapeInputFormat.setInputPaths(repartitionJoinJob, inputFiles[fileToRepartition]); repartitionJoinJob.setInputFormat(ShapeInputFormat.class); ClusterStatus clusterStatus = new JobClient(repartitionJoinJob).getClusterStatus(); repartitionJoinJob.setNumMapTasks(10 * Math.max(1, clusterStatus.getMaxMapTasks())); SpatialSite.setCells(repartitionJoinJob, cellsInfo); repartitionJoinJob.setBoolean(SpatialSite.OVERWRITE, overwrite); // set reduce function repartitionJoinJob.setReducerClass(RepartitionJoinReduce.class); repartitionJoinJob.setNumReduceTasks( Math.max(1, Math.min(cellsInfo.length, (clusterStatus.getMaxReduceTasks() * 9 + 5) / 10))); repartitionJoinJob.setOutputFormat(TextOutputFormat.class); TextOutputFormat.setOutputPath(repartitionJoinJob, outputPath); RunningJob runningJob = JobClient.runJob(repartitionJoinJob); Counters counters = runningJob.getCounters(); Counter outputRecordCounter = counters.findCounter(Task.Counter.REDUCE_OUTPUT_RECORDS); final long resultCount = outputRecordCounter.getValue(); // Output number of running map tasks Counter mapTaskCountCounter = counters.findCounter(JobInProgress.Counter.TOTAL_LAUNCHED_MAPS); System.out.println("Number of map tasks " + mapTaskCountCounter.getValue()); // Delete output directory if not explicitly set by user if (outputFile == null) fs.delete(outputPath, true); long t2 = System.currentTimeMillis(); System.out.println("Repartitioning and Joining time " + (t2 - t1) + " millis"); return resultCount; }
From source file:edu.umn.cs.spatialHadoop.operations.Plot.java
License:Apache License
public static <S extends Shape> void plotMapReduce(Path inFile, Path outFile, Shape shape, int width, int height, Color color, boolean showBorders, boolean showBlockCount, boolean showRecordCount, boolean background) throws IOException { JobConf job = new JobConf(Plot.class); job.setJobName("Plot"); job.setMapperClass(PlotMap.class); ClusterStatus clusterStatus = new JobClient(job).getClusterStatus(); job.setNumMapTasks(clusterStatus.getMaxMapTasks() * 5); job.setReducerClass(PlotReduce.class); job.setNumReduceTasks(Math.max(1, clusterStatus.getMaxReduceTasks())); job.setMapOutputKeyClass(Rectangle.class); SpatialSite.setShapeClass(job, shape.getClass()); job.setMapOutputValueClass(shape.getClass()); FileSystem inFs = inFile.getFileSystem(job); Rectangle fileMbr = FileMBR.fileMBRMapReduce(inFs, inFile, shape, false); FileStatus inFileStatus = inFs.getFileStatus(inFile); CellInfo[] cellInfos;/*from w ww . j av a2 s . co m*/ GlobalIndex<Partition> gindex = SpatialSite.getGlobalIndex(inFs, inFile); if (gindex == null) { // A heap file. The map function should partition the file GridInfo gridInfo = new GridInfo(fileMbr.x1, fileMbr.y1, fileMbr.x2, fileMbr.y2); gridInfo.calculateCellDimensions(inFileStatus.getLen(), inFileStatus.getBlockSize()); cellInfos = gridInfo.getAllCells(); // Doesn't make sense to show any partition information in a heap file showBorders = showBlockCount = showRecordCount = false; } else { cellInfos = SpatialSite.cellsOf(inFs, inFile); } // Set cell information in the job configuration to be used by the mapper SpatialSite.setCells(job, cellInfos); // Adjust width and height to maintain aspect ratio if ((fileMbr.x2 - fileMbr.x1) / (fileMbr.y2 - fileMbr.y1) > (double) width / height) { // Fix width and change height height = (int) ((fileMbr.y2 - fileMbr.y1) * width / (fileMbr.x2 - fileMbr.x1)); } else { width = (int) ((fileMbr.x2 - fileMbr.x1) * height / (fileMbr.y2 - fileMbr.y1)); } LOG.info("Creating an image of size " + width + "x" + height); ImageOutputFormat.setFileMBR(job, fileMbr); ImageOutputFormat.setImageWidth(job, width); ImageOutputFormat.setImageHeight(job, height); job.setBoolean(ShowBorders, showBorders); job.setBoolean(ShowBlockCount, showBlockCount); job.setBoolean(ShowRecordCount, showRecordCount); job.setInt(StrokeColor, color.getRGB()); // Set input and output job.setInputFormat(ShapeInputFormat.class); ShapeInputFormat.addInputPath(job, inFile); // Set output committer which will stitch images together after all reducers // finish job.setOutputCommitter(PlotOutputCommitter.class); job.setOutputFormat(ImageOutputFormat.class); TextOutputFormat.setOutputPath(job, outFile); if (background) { JobClient jc = new JobClient(job); lastSubmittedJob = jc.submitJob(job); } else { lastSubmittedJob = JobClient.runJob(job); } }
From source file:edu.umn.cs.spatialHadoop.operations.PyramidPlot.java
License:Apache License
/** * Plot a file to a set of images in different zoom levels using a MapReduce * program.//from w w w .ja va 2 s . c o m * @param <S> type of shapes stored in file * @param inFile - Path to the input file(s) * @param outFile - Path to the output file (image) * @param shape - A sample object to be used for parsing input file * @param tileWidth - With of each tile * @param tileHeight - Height of each tile * @param vflip - Set to <code>true</code> to file the whole image vertically * @param color - Color used to draw single shapes * @param numLevels - Number of zoom levels to plot * @throws IOException */ private static <S extends Shape> RunningJob plotMapReduce(Path inFile, Path outFile, OperationsParams params) throws IOException { Color color = params.getColor("color", Color.BLACK); String hdfDataset = (String) params.get("dataset"); Shape shape = hdfDataset != null ? new NASARectangle() : params.getShape("shape"); Shape plotRange = params.getShape("rect"); boolean background = params.is("background"); JobConf job = new JobConf(params, PyramidPlot.class); job.setJobName("PlotPyramid"); String partition = job.get("partition", "space").toLowerCase(); if (partition.equals("space")) { job.setMapperClass(SpacePartitionMap.class); job.setReducerClass(SpacePartitionReduce.class); job.setMapOutputKeyClass(TileIndex.class); job.setMapOutputValueClass(shape.getClass()); job.setInputFormat(ShapeInputFormat.class); } else { job.setMapperClass(DataPartitionMap.class); job.setReducerClass(DataPartitionReduce.class); job.setMapOutputKeyClass(TileIndex.class); job.setMapOutputValueClass(ImageWritable.class); job.setInputFormat(ShapeArrayInputFormat.class); } job.setInt("color", color.getRGB()); ClusterStatus clusterStatus = new JobClient(job).getClusterStatus(); job.setNumMapTasks(clusterStatus.getMaxMapTasks() * 5); job.setNumReduceTasks(Math.max(1, clusterStatus.getMaxReduceTasks())); if (shape instanceof Point && job.getBoolean("sample", false)) { // Enable adaptive sampling int imageWidthRoot = job.getInt("tilewidth", 256); int imageHeightRoot = job.getInt("tileheight", 256); long recordCount = FileMBR.fileMBR(inFile, params).recordCount; float sampleRatio = params.getFloat(GeometricPlot.AdaptiveSampleFactor, 1.0f) * imageWidthRoot * imageHeightRoot / recordCount; job.setFloat(GeometricPlot.AdaptiveSampleRatio, sampleRatio); } Rectangle fileMBR; if (hdfDataset != null) { // Input is HDF job.set(HDFRecordReader.DatasetName, hdfDataset); job.setBoolean(HDFRecordReader.SkipFillValue, true); job.setClass("shape", NASARectangle.class, Shape.class); // Determine the range of values by opening one of the HDF files Aggregate.MinMax minMax = Aggregate.aggregate(new Path[] { inFile }, params); job.setInt(MinValue, minMax.minValue); job.setInt(MaxValue, minMax.maxValue); //fileMBR = new Rectangle(-180, -90, 180, 90); fileMBR = plotRange != null ? plotRange.getMBR() : new Rectangle(-180, -140, 180, 169); // job.setClass(HDFRecordReader.ProjectorClass, MercatorProjector.class, // GeoProjector.class); } else { fileMBR = FileMBR.fileMBR(inFile, params); } boolean keepAspectRatio = params.is("keep-ratio", true); if (keepAspectRatio) { // Expand input file to a rectangle for compatibility with the pyramid // structure if (fileMBR.getWidth() > fileMBR.getHeight()) { fileMBR.y1 -= (fileMBR.getWidth() - fileMBR.getHeight()) / 2; fileMBR.y2 = fileMBR.y1 + fileMBR.getWidth(); } else { fileMBR.x1 -= (fileMBR.getHeight() - fileMBR.getWidth() / 2); fileMBR.x2 = fileMBR.x1 + fileMBR.getHeight(); } } SpatialSite.setRectangle(job, InputMBR, fileMBR); // Set input and output ShapeInputFormat.addInputPath(job, inFile); if (plotRange != null) { job.setClass(SpatialSite.FilterClass, RangeFilter.class, BlockFilter.class); } job.setOutputFormat(PyramidOutputFormat.class); TextOutputFormat.setOutputPath(job, outFile); job.setOutputCommitter(PlotPyramidOutputCommitter.class); if (background) { JobClient jc = new JobClient(job); return lastSubmittedJob = jc.submitJob(job); } else { return lastSubmittedJob = JobClient.runJob(job); } }