List of usage examples for org.apache.hadoop.mapred JobConf setInt
public void setInt(String name, int value)
name property to an int. From source file:voldemort.store.readonly.mr.azkaban.VoldemortBatchIndexJob.java
License:Apache License
/** * Method to allow this process to be a instance call from another Job. * /* w w w. j a va 2 s. c o m*/ * @storeName to dump the value * @inputFile to generate the VFILE * * */ public void execute(String voldemortClusterLocalFile, String storeName, String inputPath, String outputPath, int voldemortCheckDataPercent) throws IOException, URISyntaxException { JobConf conf = createJobConf(VoldemortBatchIndexMapper.class, VoldemortBatchIndexReducer.class); try { // get the voldemort cluster definition // We need to use cluster.xml here where it not yet localized by // TaskRunner _cluster = HadoopUtils.readCluster(voldemortClusterLocalFile, conf); } catch (Exception e) { logger.error("Failed to read Voldemort cluster details", e); throw new RuntimeException("", e); } // set the partitioner conf.setPartitionerClass(VoldemortBatchIndexPartitoner.class); conf.setNumReduceTasks(_cluster.getNumberOfNodes()); // Blow Away the O/p if force.overwirte is available FileInputFormat.setInputPaths(conf, inputPath); FileOutputFormat.setOutputPath(conf, new Path(outputPath)); if (getProps().getBoolean("force.output.overwrite", false)) { FileSystem fs = FileOutputFormat.getOutputPath(conf).getFileSystem(conf); fs.delete(FileOutputFormat.getOutputPath(conf), true); } conf.setInputFormat(SequenceFileInputFormat.class); conf.setOutputFormat(SequenceFileOutputFormat.class); conf.setMapOutputKeyClass(BytesWritable.class); conf.setMapOutputValueClass(BytesWritable.class); conf.setOutputKeyClass(BytesWritable.class); conf.setOutputValueClass(BytesWritable.class); conf.setNumReduceTasks(_cluster.getNumberOfNodes()); // get the store information conf.setStrings("voldemort.index.filename", storeName + ".index"); conf.setStrings("voldemort.data.filename", storeName + ".data"); conf.setInt("input.data.check.percent", voldemortCheckDataPercent); conf.setStrings("voldemort.store.name", storeName); // run(conf); JobClient.runJob(conf); }
From source file:voldemort.store.readonly.mr.HadoopStoreBuilder.java
License:Apache License
/** * Run the job/*from w ww .jav a2 s .com*/ */ public void build() { try { JobConf conf = new JobConf(config); conf.setInt("io.file.buffer.size", DEFAULT_BUFFER_SIZE); conf.set("cluster.xml", new ClusterMapper().writeCluster(cluster)); conf.set("stores.xml", new StoreDefinitionsMapper().writeStoreList(Collections.singletonList(storeDef))); conf.setBoolean("save.keys", saveKeys); conf.setBoolean("reducer.per.bucket", reducerPerBucket); if (!isAvro) { conf.setPartitionerClass(HadoopStoreBuilderPartitioner.class); conf.setMapperClass(mapperClass); conf.setMapOutputKeyClass(BytesWritable.class); conf.setMapOutputValueClass(BytesWritable.class); if (reducerPerBucket) { conf.setReducerClass(HadoopStoreBuilderReducerPerBucket.class); } else { conf.setReducerClass(HadoopStoreBuilderReducer.class); } } conf.setInputFormat(inputFormatClass); conf.setOutputFormat(SequenceFileOutputFormat.class); conf.setOutputKeyClass(BytesWritable.class); conf.setOutputValueClass(BytesWritable.class); conf.setJarByClass(getClass()); conf.setReduceSpeculativeExecution(false); FileInputFormat.setInputPaths(conf, inputPath); conf.set("final.output.dir", outputDir.toString()); conf.set("checksum.type", CheckSum.toString(checkSumType)); FileOutputFormat.setOutputPath(conf, tempDir); FileSystem outputFs = outputDir.getFileSystem(conf); if (outputFs.exists(outputDir)) { throw new IOException("Final output directory already exists."); } // delete output dir if it already exists FileSystem tempFs = tempDir.getFileSystem(conf); tempFs.delete(tempDir, true); long size = sizeOfPath(tempFs, inputPath); logger.info("Data size = " + size + ", replication factor = " + storeDef.getReplicationFactor() + ", numNodes = " + cluster.getNumberOfNodes() + ", chunk size = " + chunkSizeBytes); // Derive "rough" number of chunks and reducers int numReducers; if (saveKeys) { if (this.numChunks == -1) { this.numChunks = Math.max((int) (storeDef.getReplicationFactor() * size / cluster.getNumberOfPartitions() / storeDef.getReplicationFactor() / chunkSizeBytes), 1); } else { logger.info( "Overriding chunk size byte and taking num chunks (" + this.numChunks + ") directly"); } if (reducerPerBucket) { numReducers = cluster.getNumberOfPartitions() * storeDef.getReplicationFactor(); } else { numReducers = cluster.getNumberOfPartitions() * storeDef.getReplicationFactor() * numChunks; } } else { if (this.numChunks == -1) { this.numChunks = Math.max((int) (storeDef.getReplicationFactor() * size / cluster.getNumberOfPartitions() / chunkSizeBytes), 1); } else { logger.info( "Overriding chunk size byte and taking num chunks (" + this.numChunks + ") directly"); } if (reducerPerBucket) { numReducers = cluster.getNumberOfPartitions(); } else { numReducers = cluster.getNumberOfPartitions() * numChunks; } } conf.setInt("num.chunks", numChunks); conf.setNumReduceTasks(numReducers); if (isAvro) { conf.setPartitionerClass(AvroStoreBuilderPartitioner.class); // conf.setMapperClass(mapperClass); conf.setMapOutputKeyClass(ByteBuffer.class); conf.setMapOutputValueClass(ByteBuffer.class); conf.setInputFormat(inputFormatClass); conf.setOutputFormat((Class<? extends OutputFormat>) AvroOutputFormat.class); conf.setOutputKeyClass(ByteBuffer.class); conf.setOutputValueClass(ByteBuffer.class); // AvroJob confs for the avro mapper AvroJob.setInputSchema(conf, Schema.parse(config.get("avro.rec.schema"))); AvroJob.setOutputSchema(conf, Pair.getPairSchema(Schema.create(Schema.Type.BYTES), Schema.create(Schema.Type.BYTES))); AvroJob.setMapperClass(conf, mapperClass); if (reducerPerBucket) { conf.setReducerClass(AvroStoreBuilderReducerPerBucket.class); } else { conf.setReducerClass(AvroStoreBuilderReducer.class); } } logger.info("Number of chunks: " + numChunks + ", number of reducers: " + numReducers + ", save keys: " + saveKeys + ", reducerPerBucket: " + reducerPerBucket); logger.info("Building store..."); RunningJob job = JobClient.runJob(conf); // Once the job has completed log the counter Counters counters = job.getCounters(); if (saveKeys) { if (reducerPerBucket) { logger.info("Number of collisions in the job - " + counters.getCounter(KeyValueWriter.CollisionCounter.NUM_COLLISIONS)); logger.info("Maximum number of collisions for one entry - " + counters.getCounter(KeyValueWriter.CollisionCounter.MAX_COLLISIONS)); } else { logger.info("Number of collisions in the job - " + counters.getCounter(KeyValueWriter.CollisionCounter.NUM_COLLISIONS)); logger.info("Maximum number of collisions for one entry - " + counters.getCounter(KeyValueWriter.CollisionCounter.MAX_COLLISIONS)); } } // Do a CheckSumOfCheckSum - Similar to HDFS CheckSum checkSumGenerator = CheckSum.getInstance(this.checkSumType); if (!this.checkSumType.equals(CheckSumType.NONE) && checkSumGenerator == null) { throw new VoldemortException("Could not generate checksum digest for type " + this.checkSumType); } // Check if all folder exists and with format file for (Node node : cluster.getNodes()) { ReadOnlyStorageMetadata metadata = new ReadOnlyStorageMetadata(); if (saveKeys) { metadata.add(ReadOnlyStorageMetadata.FORMAT, ReadOnlyStorageFormat.READONLY_V2.getCode()); } else { metadata.add(ReadOnlyStorageMetadata.FORMAT, ReadOnlyStorageFormat.READONLY_V1.getCode()); } Path nodePath = new Path(outputDir.toString(), "node-" + node.getId()); if (!outputFs.exists(nodePath)) { logger.info("No data generated for node " + node.getId() + ". Generating empty folder"); outputFs.mkdirs(nodePath); // Create empty folder outputFs.setPermission(nodePath, new FsPermission(HADOOP_FILE_PERMISSION)); logger.info("Setting permission to 755 for " + nodePath); } if (checkSumType != CheckSumType.NONE) { FileStatus[] storeFiles = outputFs.listStatus(nodePath, new PathFilter() { public boolean accept(Path arg0) { if (arg0.getName().endsWith("checksum") && !arg0.getName().startsWith(".")) { return true; } return false; } }); if (storeFiles != null && storeFiles.length > 0) { Arrays.sort(storeFiles, new IndexFileLastComparator()); FSDataInputStream input = null; for (FileStatus file : storeFiles) { try { input = outputFs.open(file.getPath()); byte fileCheckSum[] = new byte[CheckSum.checkSumLength(this.checkSumType)]; input.read(fileCheckSum); logger.debug("Checksum for file " + file.toString() + " - " + new String(Hex.encodeHex(fileCheckSum))); checkSumGenerator.update(fileCheckSum); } catch (Exception e) { logger.error("Error while reading checksum file " + e.getMessage(), e); } finally { if (input != null) input.close(); } outputFs.delete(file.getPath(), false); } metadata.add(ReadOnlyStorageMetadata.CHECKSUM_TYPE, CheckSum.toString(checkSumType)); String checkSum = new String(Hex.encodeHex(checkSumGenerator.getCheckSum())); logger.info("Checksum for node " + node.getId() + " - " + checkSum); metadata.add(ReadOnlyStorageMetadata.CHECKSUM, checkSum); } } // Write metadata Path metadataPath = new Path(nodePath, ".metadata"); FSDataOutputStream metadataStream = outputFs.create(metadataPath); outputFs.setPermission(metadataPath, new FsPermission(HADOOP_FILE_PERMISSION)); logger.info("Setting permission to 755 for " + metadataPath); metadataStream.write(metadata.toJsonString().getBytes()); metadataStream.flush(); metadataStream.close(); } } catch (Exception e) { logger.error("Error in Store builder", e); throw new VoldemortException(e); } }
From source file:voldemort.store.readwrite.mr.HadoopRWStoreBuilder.java
License:Apache License
/** * Run the job//from w ww . j a v a 2 s .c om */ public void build() { JobConf conf = new JobConf(config); conf.setInt("io.file.buffer.size", DEFAULT_BUFFER_SIZE); conf.set("cluster.xml", new ClusterMapper().writeCluster(cluster)); conf.set("stores.xml", new StoreDefinitionsMapper().writeStoreList(Collections.singletonList(storeDef))); conf.setInt("vector.node.id", this.vectorNodeId); conf.setLong("vector.node.version", this.vectorNodeVersion); conf.setLong("job.start.time.ms", System.currentTimeMillis()); conf.setPartitionerClass(HadoopRWStoreBuilderPartitioner.class); conf.setInputFormat(inputFormatClass); conf.setMapperClass(mapperClass); conf.setMapOutputKeyClass(BytesWritable.class); conf.setMapOutputValueClass(BytesWritable.class); conf.setReducerClass(HadoopRWStoreBuilderReducer.class); conf.setOutputFormat(SequenceFileOutputFormat.class); conf.setOutputKeyClass(BytesWritable.class); conf.setOutputValueClass(BytesWritable.class); conf.setReduceSpeculativeExecution(false); conf.setJarByClass(getClass()); FileInputFormat.setInputPaths(conf, inputPath); FileOutputFormat.setOutputPath(conf, tempPath); try { // delete the temp dir if it exists FileSystem tempFs = tempPath.getFileSystem(conf); tempFs.delete(tempPath, true); conf.setInt("num.chunks", reducersPerNode); int numReducers = cluster.getNumberOfNodes() * reducersPerNode; logger.info("Replication factor = " + storeDef.getReplicationFactor() + ", numNodes = " + cluster.getNumberOfNodes() + ", reducers per node = " + reducersPerNode + ", numReducers = " + numReducers); conf.setNumReduceTasks(numReducers); logger.info("Building RW store..."); JobClient.runJob(conf); } catch (Exception e) { throw new VoldemortException(e); } }
From source file:wikiduper.wikipedia.RepackWikipedia.java
License:Apache License
@SuppressWarnings("static-access") @Override//from w ww. j ava2s.co 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("output location") .create(OUTPUT_OPTION)); options.addOption(OptionBuilder.withArgName("path").hasArg().withDescription("mapping file") .create(MAPPING_FILE_OPTION)); options.addOption(OptionBuilder.withArgName("block|record|none").hasArg() .withDescription("compression type").create(COMPRESSION_TYPE_OPTION)); options.addOption(OptionBuilder.withArgName("en|sv|de").hasArg().withDescription("two-letter language code") .create(LANGUAGE_OPTION)); 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_OPTION) || !cmdline.hasOption(MAPPING_FILE_OPTION) || !cmdline.hasOption(COMPRESSION_TYPE_OPTION)) { HelpFormatter formatter = new HelpFormatter(); formatter.printHelp(this.getClass().getName(), options); ToolRunner.printGenericCommandUsage(System.out); return -1; } String inputPath = cmdline.getOptionValue(INPUT_OPTION); String outputPath = cmdline.getOptionValue(OUTPUT_OPTION); String mappingFile = cmdline.getOptionValue(MAPPING_FILE_OPTION); String compressionType = cmdline.getOptionValue(COMPRESSION_TYPE_OPTION); if (!"block".equals(compressionType) && !"record".equals(compressionType) && !"none".equals(compressionType)) { System.err.println("Error: \"" + compressionType + "\" unknown compression type!"); 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; } } // this is the default block size int blocksize = 1000000; JobConf conf = new JobConf(getConf(), RepackWikipedia.class); conf.setJobName(String.format("RepackWikipedia[%s: %s, %s: %s, %s: %s, %s: %s]", INPUT_OPTION, inputPath, OUTPUT_OPTION, outputPath, COMPRESSION_TYPE_OPTION, compressionType, LANGUAGE_OPTION, language)); conf.set(DOCNO_MAPPING_FIELD, mappingFile); LOG.info("Tool name: " + this.getClass().getName()); LOG.info(" - XML dump file: " + inputPath); LOG.info(" - output path: " + outputPath); LOG.info(" - docno mapping data file: " + mappingFile); LOG.info(" - compression type: " + compressionType); LOG.info(" - language: " + language); if ("block".equals(compressionType)) { LOG.info(" - block size: " + blocksize); } int mapTasks = 10; conf.setNumMapTasks(mapTasks); conf.setNumReduceTasks(0); SequenceFileInputFormat.addInputPath(conf, new Path(inputPath)); SequenceFileOutputFormat.setOutputPath(conf, new Path(outputPath)); if ("none".equals(compressionType)) { SequenceFileOutputFormat.setCompressOutput(conf, false); } else { SequenceFileOutputFormat.setCompressOutput(conf, true); if ("record".equals(compressionType)) { SequenceFileOutputFormat.setOutputCompressionType(conf, SequenceFile.CompressionType.RECORD); } else { SequenceFileOutputFormat.setOutputCompressionType(conf, SequenceFile.CompressionType.BLOCK); conf.setInt("io.seqfile.compress.blocksize", blocksize); } } if (language != null) { conf.set("wiki.language", language); } conf.setInputFormat(WikipediaPageInputFormat.class); conf.setOutputFormat(SequenceFileOutputFormat.class); conf.setOutputKeyClass(IntWritable.class); conf.setOutputValueClass(WikipediaPageFactory.getWikipediaPageClass(language)); conf.setMapperClass(MyMapper.class); // Delete the output directory if it exists already. FileSystem.get(conf).delete(new Path(outputPath), true); JobClient.runJob(conf); return 0; }