org.apache.hadoop.hbase.mapreduce.HFileOutputFormat2 Java Examples

@Override
public void init() throws IOException {
  super.init();

  Configuration taskConf = new Configuration();
  Path stagingResultDir = new Path(stagingDir, TajoConstants.RESULT_DIR_NAME);
  taskConf.set(FileOutputFormat.OUTDIR, stagingResultDir.toString());

  ExecutionBlockId ebId = taskAttemptId.getTaskId().getExecutionBlockId();
  writerContext = new TaskAttemptContextImpl(taskConf,
      new TaskAttemptID(ebId.getQueryId().toString(), ebId.getId(), TaskType.MAP,
          taskAttemptId.getTaskId().getId(), taskAttemptId.getId()));

  HFileOutputFormat2 hFileOutputFormat2 = new HFileOutputFormat2();
  try {
    writer = hFileOutputFormat2.getRecordWriter(writerContext);

    committer = new FileOutputCommitter(FileOutputFormat.getOutputPath(writerContext), writerContext);
    workingFilePath = committer.getWorkPath();
  } catch (InterruptedException e) {
    throw new IOException(e.getMessage(), e);
  }

  LOG.info("Created hbase file writer: " + workingFilePath);
}
 

/**
 * Do hfile bulk exporting
 * 
 * @param builder
 * @throws Exception
 */
@SuppressWarnings({ "unchecked", "rawtypes" })
public static void doExporting(CommandLine line) throws Exception {
	// Configuration.
	String tabname = line.getOptionValue("tabname");
	String user = line.getOptionValue("user");
	Configuration conf = new Configuration();
	conf.set("hbase.zookeeper.quorum", line.getOptionValue("zkaddr"));
	conf.set("hbase.fs.tmp.dir", line.getOptionValue("T", DEFAULT_HBASE_MR_TMPDIR));
	conf.set(TableInputFormat.INPUT_TABLE, tabname);
	conf.set(TableInputFormat.SCAN_BATCHSIZE, line.getOptionValue("batchSize", DEFAULT_SCAN_BATCH_SIZE));

	// Check directory.
	String outputDir = line.getOptionValue("output", DEFAULT_HFILE_OUTPUT_DIR) + "/" + tabname;
	FileSystem fs = FileSystem.get(new URI(outputDir), new Configuration(), user);
	state(!fs.exists(new Path(outputDir)), format("HDFS temporary directory already has data, path: '%s'", outputDir));

	// Set scan condition.(if necessary)
	setScanIfNecessary(conf, line);

	// Job.
	Connection conn = ConnectionFactory.createConnection(conf);
	TableName tab = TableName.valueOf(tabname);
	Job job = Job.getInstance(conf);
	job.setJobName(HfileBulkExporter.class.getSimpleName() + "@" + tab.getNameAsString());
	job.setJarByClass(HfileBulkExporter.class);
	job.setMapperClass((Class<Mapper>) ClassUtils.getClass(line.getOptionValue("mapperClass", DEFAULT_MAPPER_CLASS)));
	job.setInputFormatClass(TableInputFormat.class);
	job.setMapOutputKeyClass(ImmutableBytesWritable.class);
	job.setMapOutputValueClass(Put.class);

	HFileOutputFormat2.configureIncrementalLoad(job, conn.getTable(tab), conn.getRegionLocator(tab));
	FileOutputFormat.setOutputPath(job, new Path(outputDir));
	if (job.waitForCompletion(true)) {
		long total = job.getCounters().findCounter(DEFUALT_COUNTER_GROUP, DEFUALT_COUNTER_TOTAL).getValue();
		long processed = job.getCounters().findCounter(DEFUALT_COUNTER_GROUP, DEFUALT_COUNTER_PROCESSED).getValue();
		log.info(String.format("Exported to successfully! with processed:(%d)/total:(%d)", processed, total));
	}

}
 

/**
 * Submits the job and waits for completion.
 * @param job job
 * @param outputPath output path
 * @throws Exception
 */
private void configureRunnableJobUsingBulkLoad(Job job, Path outputPath, TableName outputTableName,
                                               boolean skipDependencyJars) throws Exception {
    job.setMapperClass(getBulkMapperClass());
    job.setMapOutputKeyClass(ImmutableBytesWritable.class);
    job.setMapOutputValueClass(KeyValue.class);
    final Configuration configuration = job.getConfiguration();
    try (Connection conn = ConnectionFactory.createConnection(configuration);
         Admin admin = conn.getAdmin();
         Table table = conn.getTable(outputTableName);
         RegionLocator regionLocator = conn.getRegionLocator(outputTableName)) {
        HFileOutputFormat2.configureIncrementalLoad(job, table, regionLocator);
        if (skipDependencyJars) {
            job.getConfiguration().unset("tmpjars");
        }
        boolean status = job.waitForCompletion(true);
        if (!status) {
            LOG.error("IndexTool job failed!");
            throw new Exception("IndexTool job failed: " + job.toString());
        }

        LOG.info("Loading HFiles from {}", outputPath);
        LoadIncrementalHFiles loader = new LoadIncrementalHFiles(configuration);
        loader.doBulkLoad(outputPath, admin, table, regionLocator);
    }
    FileSystem.get(configuration).delete(outputPath, true);
}
 

/**
 * Sets up the actual job.
 * @param args The command line parameters.
 * @return The newly created job.
 * @throws IOException When setting up the job fails.
 */
public Job createSubmittableJob(String[] args) throws IOException {
  Configuration conf = getConf();
  String inputDirs = args[0];
  String tabName = args[1];
  conf.setStrings(TABLES_KEY, tabName);
  conf.set(FileInputFormat.INPUT_DIR, inputDirs);
  Job job =
      Job.getInstance(conf,
        conf.get(JOB_NAME_CONF_KEY, NAME + "_" + EnvironmentEdgeManager.currentTime()));
  job.setJarByClass(MapReduceHFileSplitterJob.class);
  job.setInputFormatClass(HFileInputFormat.class);
  job.setMapOutputKeyClass(ImmutableBytesWritable.class);
  String hfileOutPath = conf.get(BULK_OUTPUT_CONF_KEY);
  if (hfileOutPath != null) {
    LOG.debug("add incremental job :" + hfileOutPath + " from " + inputDirs);
    TableName tableName = TableName.valueOf(tabName);
    job.setMapperClass(HFileCellMapper.class);
    job.setReducerClass(CellSortReducer.class);
    Path outputDir = new Path(hfileOutPath);
    FileOutputFormat.setOutputPath(job, outputDir);
    job.setMapOutputValueClass(MapReduceExtendedCell.class);
    try (Connection conn = ConnectionFactory.createConnection(conf);
        Table table = conn.getTable(tableName);
        RegionLocator regionLocator = conn.getRegionLocator(tableName)) {
      HFileOutputFormat2.configureIncrementalLoad(job, table.getDescriptor(), regionLocator);
    }
    LOG.debug("success configuring load incremental job");

    TableMapReduceUtil.addDependencyJars(job.getConfiguration(),
      org.apache.hbase.thirdparty.com.google.common.base.Preconditions.class);
  } else {
    throw new IOException("No bulk output directory specified");
  }
  return job;
}
 

/**
 * Do hfile bulk exporting
 * 
 * @param builder
 * @throws Exception
 */
@SuppressWarnings({ "unchecked", "rawtypes" })
public static void doRmdbExporting(CommandLine line) throws Exception {
	// Configuration.
	String tabname = line.getOptionValue("tabname");
	String user = line.getOptionValue("user");
	Configuration conf = new Configuration();
	conf.set("hbase.zookeeper.quorum", line.getOptionValue("zkaddr"));
	conf.set("hbase.fs.tmp.dir", line.getOptionValue("T", DEFAULT_HBASE_MR_TMPDIR));
	conf.set(TableInputFormat.INPUT_TABLE, tabname);
	conf.set(TableInputFormat.SCAN_BATCHSIZE, line.getOptionValue("batchSize", DEFAULT_SCAN_BATCH_SIZE));

	// Check directory.
	String outputDir = line.getOptionValue("output", DEFAULT_HFILE_OUTPUT_DIR) + "/" + tabname;
	FileSystem fs = FileSystem.get(new URI(outputDir), new Configuration(), user);
	if (fs.exists(new Path(outputDir))) {
		fs.delete(new Path(outputDir), true);
	}

	// Set scan condition.(if necessary)
	HfileBulkExporter.setScanIfNecessary(conf, line);

	// Job.
	Connection conn = ConnectionFactory.createConnection(conf);
	TableName tab = TableName.valueOf(tabname);
	Job job = Job.getInstance(conf);
	job.setJobName(HfileBulkExporter.class.getSimpleName() + "@" + tab.getNameAsString());
	job.setJarByClass(HfileBulkExporter.class);
	job.setMapperClass((Class<Mapper>) ClassUtils.getClass(line.getOptionValue("mapperClass", DEFAULT_MAPPER_CLASS)));
	job.setInputFormatClass(TableInputFormat.class);
	job.setMapOutputKeyClass(ImmutableBytesWritable.class);
	job.setMapOutputValueClass(Put.class);

	HFileOutputFormat2.configureIncrementalLoad(job, conn.getTable(tab), conn.getRegionLocator(tab));
	FileOutputFormat.setOutputPath(job, new Path(outputDir));
	if (job.waitForCompletion(true)) {
		long total = job.getCounters().findCounter(DEFUALT_COUNTER_GROUP, DEFUALT_COUNTER_TOTAL).getValue();
		long processed = job.getCounters().findCounter(DEFUALT_COUNTER_GROUP, DEFUALT_COUNTER_PROCESSED).getValue();
		log.info(String.format("Exported to successfully! with processed:(%d)/total:(%d)", processed, total));
	}

}
 

@Override
public int run(CommandLine cmd) throws Exception {
    String source = cmd.getOptionValue('t');
    TableMapReduceUtil.addDependencyJars(getConf(),
        HalyardExport.class,
        NTriplesUtil.class,
        Rio.class,
        AbstractRDFHandler.class,
        RDFFormat.class,
        RDFParser.class,
        HTable.class,
        HBaseConfiguration.class,
        AuthenticationProtos.class,
        Trace.class,
        Gauge.class);
    HBaseConfiguration.addHbaseResources(getConf());
    Job job = Job.getInstance(getConf(), "HalyardDelete " + source);
    if (cmd.hasOption('s')) {
        job.getConfiguration().set(SUBJECT, cmd.getOptionValue('s'));
    }
    if (cmd.hasOption('p')) {
        job.getConfiguration().set(PREDICATE, cmd.getOptionValue('p'));
    }
    if (cmd.hasOption('o')) {
        job.getConfiguration().set(OBJECT, cmd.getOptionValue('o'));
    }
    if (cmd.hasOption('g')) {
        job.getConfiguration().setStrings(CONTEXTS, cmd.getOptionValues('g'));
    }
    job.setJarByClass(HalyardBulkDelete.class);
    TableMapReduceUtil.initCredentials(job);

    Scan scan = HalyardTableUtils.scan(null, null);

    TableMapReduceUtil.initTableMapperJob(source,
        scan,
        DeleteMapper.class,
        ImmutableBytesWritable.class,
        LongWritable.class,
        job);

    job.setMapOutputKeyClass(ImmutableBytesWritable.class);
    job.setMapOutputValueClass(KeyValue.class);
    job.setSpeculativeExecution(false);
    job.setMapSpeculativeExecution(false);
    job.setReduceSpeculativeExecution(false);
    try (HTable hTable = HalyardTableUtils.getTable(getConf(), source, false, 0)) {
        HFileOutputFormat2.configureIncrementalLoad(job, hTable.getTableDescriptor(), hTable.getRegionLocator());
        FileOutputFormat.setOutputPath(job, new Path(cmd.getOptionValue('f')));
        TableMapReduceUtil.addDependencyJars(job);
        if (job.waitForCompletion(true)) {
            new LoadIncrementalHFiles(getConf()).doBulkLoad(new Path(cmd.getOptionValue('f')), hTable);
            LOG.info("Bulk Delete Completed..");
            return 0;
        }
    }
    return -1;
}
 

@Override
protected int run(CommandLine cmd) throws Exception {
    String source = cmd.getOptionValue('s');
    String workdir = cmd.getOptionValue('w');
    String target = cmd.getOptionValue('t');
    getConf().setBoolean(SKIP_INVALID_PROPERTY, cmd.hasOption('i'));
    getConf().setBoolean(VERIFY_DATATYPE_VALUES_PROPERTY, cmd.hasOption('d'));
    getConf().setBoolean(TRUNCATE_PROPERTY, cmd.hasOption('r'));
    getConf().setInt(SPLIT_BITS_PROPERTY, Integer.parseInt(cmd.getOptionValue('b', "3")));
    if (cmd.hasOption('g')) getConf().set(DEFAULT_CONTEXT_PROPERTY, cmd.getOptionValue('g'));
    getConf().setBoolean(OVERRIDE_CONTEXT_PROPERTY, cmd.hasOption('o'));
    getConf().setLong(DEFAULT_TIMESTAMP_PROPERTY, Long.parseLong(cmd.getOptionValue('e', String.valueOf(System.currentTimeMillis()))));
    if (cmd.hasOption('m')) getConf().setLong("mapreduce.input.fileinputformat.split.maxsize", Long.parseLong(cmd.getOptionValue('m')));
    TableMapReduceUtil.addDependencyJars(getConf(),
            NTriplesUtil.class,
            Rio.class,
            AbstractRDFHandler.class,
            RDFFormat.class,
            RDFParser.class);
    HBaseConfiguration.addHbaseResources(getConf());
    Job job = Job.getInstance(getConf(), "HalyardBulkLoad -> " + workdir + " -> " + target);
    job.setJarByClass(HalyardBulkLoad.class);
    job.setMapperClass(RDFMapper.class);
    job.setMapOutputKeyClass(ImmutableBytesWritable.class);
    job.setMapOutputValueClass(KeyValue.class);
    job.setInputFormatClass(RioFileInputFormat.class);
    job.setSpeculativeExecution(false);
    job.setReduceSpeculativeExecution(false);
    try (HTable hTable = HalyardTableUtils.getTable(getConf(), target, true, getConf().getInt(SPLIT_BITS_PROPERTY, 3))) {
        HFileOutputFormat2.configureIncrementalLoad(job, hTable.getTableDescriptor(), hTable.getRegionLocator());
        FileInputFormat.setInputDirRecursive(job, true);
        FileInputFormat.setInputPaths(job, source);
        FileOutputFormat.setOutputPath(job, new Path(workdir));
        TableMapReduceUtil.addDependencyJars(job);
        TableMapReduceUtil.initCredentials(job);
        if (job.waitForCompletion(true)) {
            if (getConf().getBoolean(TRUNCATE_PROPERTY, false)) {
                HalyardTableUtils.truncateTable(hTable).close();
            }
            new LoadIncrementalHFiles(getConf()).doBulkLoad(new Path(workdir), hTable);
            LOG.info("Bulk Load Completed..");
            return 0;
        }
    }
    return -1;
}
 

public int run(CommandLine cmd) throws Exception {
    String source = cmd.getOptionValue('s');
    String queryFiles = cmd.getOptionValue('q');
    String workdir = cmd.getOptionValue('w');
    getConf().setLong(DEFAULT_TIMESTAMP_PROPERTY, Long.parseLong(cmd.getOptionValue('e', String.valueOf(System.currentTimeMillis()))));
    if (cmd.hasOption('i')) getConf().set(ELASTIC_INDEX_URL, cmd.getOptionValue('i'));
    TableMapReduceUtil.addDependencyJars(getConf(),
           HalyardExport.class,
           NTriplesUtil.class,
           Rio.class,
           AbstractRDFHandler.class,
           RDFFormat.class,
           RDFParser.class,
           HTable.class,
           HBaseConfiguration.class,
           AuthenticationProtos.class,
           Trace.class,
           Gauge.class);
    HBaseConfiguration.addHbaseResources(getConf());
    getConf().setStrings(TABLE_NAME_PROPERTY, source);
    getConf().setLong(DEFAULT_TIMESTAMP_PROPERTY, getConf().getLong(DEFAULT_TIMESTAMP_PROPERTY, System.currentTimeMillis()));
    int stages = 1;
    for (int stage = 0; stage < stages; stage++) {
        Job job = Job.getInstance(getConf(), "HalyardBulkUpdate -> " + workdir + " -> " + source + " stage #" + stage);
        job.getConfiguration().setInt(STAGE_PROPERTY, stage);
        job.setJarByClass(HalyardBulkUpdate.class);
        job.setMapperClass(SPARQLUpdateMapper.class);
        job.setMapOutputKeyClass(ImmutableBytesWritable.class);
        job.setMapOutputValueClass(KeyValue.class);
        job.setInputFormatClass(QueryInputFormat.class);
        job.setSpeculativeExecution(false);
        job.setReduceSpeculativeExecution(false);
        try (HTable hTable = HalyardTableUtils.getTable(getConf(), source, false, 0)) {
            HFileOutputFormat2.configureIncrementalLoad(job, hTable.getTableDescriptor(), hTable.getRegionLocator());
            QueryInputFormat.setQueriesFromDirRecursive(job.getConfiguration(), queryFiles, true, stage);
            Path outPath = new Path(workdir, "stage"+stage);
            FileOutputFormat.setOutputPath(job, outPath);
            TableMapReduceUtil.addDependencyJars(job);
            TableMapReduceUtil.initCredentials(job);
            if (stage == 0) { //count real number of stages
                for (InputSplit is : new QueryInputFormat().getSplits(job)) {
                    QueryInputFormat.QueryInputSplit qis = (QueryInputFormat.QueryInputSplit)is;
                    int updates = QueryParserUtil.parseUpdate(QueryLanguage.SPARQL, qis.getQuery(), null).getUpdateExprs().size();
                    if (updates > stages) {
                        stages = updates;
                    }
                    LOG.log(Level.INFO, "{0} contains {1} stages of the update sequence.", new Object[]{qis.getQueryName(), updates});
                }
                LOG.log(Level.INFO, "Bulk Update will process {0} MapReduce stages.", stages);
            }
            if (job.waitForCompletion(true)) {
                new LoadIncrementalHFiles(getConf()).doBulkLoad(outPath, hTable);
                LOG.log(Level.INFO, "Stage #{0} of {1} completed..", new Object[]{stage, stages});
            } else {
                return -1;
            }
        }
    }
    LOG.info("Bulk Update Completed..");
    return 0;
}
 

@Override
public int run(String[] args) throws Exception {
  try {
    Configuration conf = HBaseConfiguration.create();
    Connection connection = ConnectionFactory.createConnection(conf);

    String inputPath = args[0];
    String outputPath = args[1];
    final TableName tableName = TableName.valueOf(args[2]);

    // tag::SETUP[]
    Table table = connection.getTable(tableName);

    Job job = Job.getInstance(conf, "ConvertToHFiles: Convert CSV to HFiles");

    HFileOutputFormat2.configureIncrementalLoad(job, table,
                                      connection.getRegionLocator(tableName)); // <1>
    job.setInputFormatClass(TextInputFormat.class); // <2>

    job.setJarByClass(ConvertToHFiles.class); // <3>
    job.setJar("/home/cloudera/ahae/target/ahae.jar"); // <3>

    job.setMapperClass(ConvertToHFilesMapper.class); // <4>
    job.setMapOutputKeyClass(ImmutableBytesWritable.class); // <5>
    job.setMapOutputValueClass(KeyValue.class); // <6>

    FileInputFormat.setInputPaths(job, inputPath);
    HFileOutputFormat2.setOutputPath(job, new Path(outputPath));
    // end::SETUP[]

    if (!job.waitForCompletion(true)) {
      LOG.error("Failure");
    } else {
      LOG.info("Success");
      return 0;
    }
  } catch (Exception e) {
    e.printStackTrace();
  }
  return 1;
}