示例#1
0
  public static void main(String[] args) throws Exception {
    String inputDirectory = "/home/cs246/Desktop/HW2/input";
    String outputDirectory = "/home/cs246/Desktop/HW2/output";
    String centroidDirectory = "/home/cs246/Desktop/HW2/config";

    int iterations = 20;

    for (int i = 1; i <= iterations; i++) {
      Configuration conf = new Configuration();

      String cFile = centroidDirectory + "/c" + i + ".txt";
      String nextCFile = centroidDirectory + "/c" + (i + 1) + ".txt";
      conf.set("CFILE", cFile);
      conf.set("NEXTCFILE", nextCFile);

      String cFile = centroidDirectory + "/c" + i + ".txt";
      String nextCFile = centroidDirectory + "/c" + (i + 1) + ".txt";
      conf.set("CFILE", cFile);
      conf.set("NEXTCFILE", nextCFile);

      Job job = new Job(conf, "HW2_Q4." + i);
      job.setJarByClass(HW2_Q4.class);
      job.setOutputKeyClass(IntWritable.class);
      job.setOutputValueClass(Text.class);
      job.setMapperClass(Map1.class);
      job.setReducerClass(Reduce1.class);
      job.setInputFormatClass(TextInputFormat.class);
      job.setOutputFormatClass(TextOutputFormat.class);

      FileInputFormat.addInputPath(job, new Path(inputDirectory));
      FileOutputFormat.setOutputPath(job, new Path(outputDirectory + "/output" + i));

      job.waitForCompletion(true);
    }
  }
  public int run(String[] args) throws Exception {
    Configuration argConf = getConf();

    // JobConf conf = new JobConf(diffdb.class);
    Configuration config = HBaseConfiguration.create();
    HBaseAdmin hbAdmin = new HBaseAdmin(config);
    dbutil db_util = new dbutil(config);

    HTable runTable = new HTable(config, "gestore_runs");
    Get runGet = new Get(argConf.get("id").getBytes());
    Result pipeline = runTable.get(runGet);

    NavigableMap<byte[], byte[]> pipeMap = pipeline.getFamilyMap("d".getBytes());

    Map.Entry<byte[], byte[]> results = pipeMap.pollFirstEntry();

    HashMap<String, HashMap<String, String>> resultMap =
        new HashMap<String, HashMap<String, String>>();

    while (results != null) {
      String resultKey = new String(results.getKey());
      String resultValue = new String(results.getValue());
      String field = "type";
      HashMap<String, String> tempMap = new HashMap<String, String>();
      String entry = resultKey;

      if (resultKey.endsWith("_db_timestamp")) {
        field = "db_timestamp";
        entry = resultKey.substring(0, resultKey.lastIndexOf("_db_timestamp"));
      } else if (resultKey.endsWith("_filename")) {
        field = "filename";
        entry = resultKey.substring(0, resultKey.lastIndexOf("_filename"));
      } else if (resultKey.endsWith("_regex")) {
        field = "regex";
        entry = resultKey.substring(0, resultKey.lastIndexOf("_regex"));
      }

      if (resultMap.containsKey(entry)) {
        tempMap = resultMap.get(entry);
      }

      tempMap.put(field, resultValue);
      resultMap.put(entry, tempMap);

      // System.out.println("Key: " + resultKey + " Value: " + resultValue);
      results = pipeMap.pollFirstEntry();
    }

    for (String key : resultMap.keySet()) {
      System.out.println("File ID: " + key);
      for (String subKey : resultMap.get(key).keySet()) {
        // System.out.println("\t " + subKey + "\t\t" + resultMap.get(key).get(subKey));
        System.out.format("  %1$-20s  %2$s\n", subKey, resultMap.get(key).get(subKey));
      }
    }

    return 0;
  }
示例#3
0
 private static String getConfigString(Configuration config) {
   String output = "";
   Iterator<Map.Entry<String, String>> iterConfig = config.iterator();
   while (iterConfig.hasNext()) {
     Map.Entry<String, String> curEntry = iterConfig.next();
     output = output + "Key: \t" + curEntry.getKey() + "\nValue: \t" + curEntry.getValue() + "\n";
   }
   return output;
 }
  /**
   * @param args
   * @throws IOException
   * @throws ClassNotFoundException
   * @throws InterruptedException
   */
  public static void main(String[] args)
      throws IOException, InterruptedException, ClassNotFoundException {
    // TODO Auto-generated method stub

    Configuration conf = new Configuration();
    conf.set("inParameter", toString(args));
    Job job = new Job(conf, "MovieGenreIdentifier");

    job.setOutputKeyClass(Text.class);
    job.setOutputValueClass(IntWritable.class);
    job.setJarByClass(MovieGenreIdentifier.class);
    job.setMapperClass(Map.class);
    job.setCombinerClass(Reduce.class);
    job.setReducerClass(Reduce.class);

    job.setInputFormatClass(TextInputFormat.class);
    job.setOutputFormatClass(TextOutputFormat.class);

    FileInputFormat.addInputPath(job, new Path(args[0]));
    FileOutputFormat.setOutputPath(job, new Path(args[1]));

    job.waitForCompletion(true);
  }
  @Override
  public int run(String[] args) throws Exception {

    String locatorHost = args[0];
    int locatorPort = Integer.parseInt(args[1]);
    String hdfsHomeDir = args[2];

    System.out.println(
        "KnownKeysMRv2 invoked with args (locatorHost = "
            + locatorHost
            + " locatorPort = "
            + locatorPort
            + " hdfsHomeDir = "
            + hdfsHomeDir);

    Configuration conf = getConf();
    conf.set(GFInputFormat.INPUT_REGION, "partitionedRegion");
    conf.set(GFInputFormat.HOME_DIR, hdfsHomeDir);
    conf.setBoolean(GFInputFormat.CHECKPOINT, false);
    conf.set(GFOutputFormat.REGION, "validationRegion");
    conf.set(GFOutputFormat.LOCATOR_HOST, locatorHost);
    conf.setInt(GFOutputFormat.LOCATOR_PORT, locatorPort);

    Job job = Job.getInstance(conf, "knownKeysMRv2");
    job.setInputFormatClass(GFInputFormat.class);
    job.setOutputFormatClass(GFOutputFormat.class);

    job.setMapperClass(KnownKeysMRv2Mapper.class);
    job.setMapOutputKeyClass(GFKey.class);
    job.setMapOutputValueClass(PEIWritable.class);

    job.setReducerClass(KnownKeysMRv2Reducer.class);
    // job.setOutputKeyClass(String.class);
    // job.setOutputValueClass(ValueHolder.class);

    return job.waitForCompletion(false) ? 0 : 1;
  }
 protected void setup(Context context) {
   Configuration config = context.getConfiguration();
   inMovies = config.get("inParameter");
 }
示例#7
0
  public int run(String[] args) throws Exception {
    // printUsage();
    /*
     * SETUP
     */
    Configuration argConf = getConf();
    Hashtable<String, String> confArg = new Hashtable<String, String>();
    setup(confArg, argConf);
    Date currentTime = new Date();
    Date endDate = new Date(new Long(confArg.get("timestamp_stop")));
    Boolean full_run = confArg.get("intermediate").matches("(?i).*true.*");
    Boolean quick_add = confArg.get("quick_add").matches("(?i).*true.*");
    logger.info("Running GeStore");

    // ZooKeeper setup
    Configuration config = HBaseConfiguration.create();
    zkWatcher = new ZooKeeperWatcher(config, "Testing", new HBaseAdmin(config));
    zkInstance =
        new ZooKeeper(
            ZKConfig.getZKQuorumServersString(config),
            config.getInt("zookeeper.session.timeout", -1),
            zkWatcher);

    if (!confArg.get("task_id").isEmpty()) {
      confArg.put("temp_path", confArg.get("temp_path") + confArg.get("task_id"));
    }

    String lockRequest = confArg.get("file_id");
    if (!confArg.get("run_id").isEmpty())
      lockRequest = lockRequest + "_" + confArg.get("run_id") + "_";
    if (!confArg.get("task_id").isEmpty())
      lockRequest = lockRequest + "_" + confArg.get("task_id") + "_";

    // Get type of movement
    toFrom type_move = checkArgs(confArg);
    if (type_move == toFrom.LOCAL2REMOTE && !confArg.get("format").equals("unknown")) {
      List<String> arguments = new ArrayList<String>();
      arguments.add("-Dinput=" + confArg.get("local_path"));
      arguments.add("-Dtable=" + confArg.get("file_id"));
      arguments.add("-Dtimestamp=" + confArg.get("timestamp_stop"));
      arguments.add("-Dtype=" + confArg.get("format"));
      arguments.add("-Dtarget_dir=" + confArg.get("base_path") + "_" + confArg.get("file_id"));
      arguments.add("-Dtemp_hdfs_path=" + confArg.get("temp_path"));
      arguments.add("-Drun_id=" + confArg.get("run_id"));
      if (!confArg.get("run_id").isEmpty()) arguments.add("-Drun_id=" + confArg.get("run_id"));
      if (!confArg.get("task_id").isEmpty()) arguments.add("-Dtask_id=" + confArg.get("task_id"));
      if (quick_add) arguments.add("-Dquick_add=" + confArg.get("quick_add"));
      String lockName = lock(lockRequest);
      String[] argumentString = arguments.toArray(new String[arguments.size()]);
      adddb.main(argumentString);
      unlock(lockName);
      System.exit(0);
    }

    // Database registration

    dbutil db_util = new dbutil(config);
    db_util.register_database(confArg.get("db_name_files"), true);
    db_util.register_database(confArg.get("db_name_runs"), true);
    db_util.register_database(confArg.get("db_name_updates"), true);
    FileSystem hdfs = FileSystem.get(config);
    FileSystem localFS = FileSystem.getLocal(config);

    // Get source type
    confArg.put("source", getSource(db_util, confArg.get("db_name_files"), confArg.get("file_id")));
    confArg.put(
        "database", isDatabase(db_util, confArg.get("db_name_files"), confArg.get("file_id")));
    if (!confArg.get("source").equals("local")
        && type_move == toFrom.REMOTE2LOCAL
        && !confArg.get("timestamp_stop").equals(Integer.toString(Integer.MAX_VALUE))) {
      confArg.put("timestamp_stop", Long.toString(latestVersion(confArg, db_util)));
    }

    /*
     * Get previous timestamp
     */
    Get run_id_get = new Get(confArg.get("run_id").getBytes());
    Result run_get = db_util.doGet(confArg.get("db_name_runs"), run_id_get);
    KeyValue run_file_prev =
        run_get.getColumnLatest(
            "d".getBytes(), (confArg.get("file_id") + "_db_timestamp").getBytes());
    String last_timestamp = new String("0");
    if (null != run_file_prev && !confArg.get("source").equals("local")) {
      long last_timestamp_real = run_file_prev.getTimestamp();
      Long current_timestamp = new Long(confArg.get("timestamp_real"));
      if ((current_timestamp - last_timestamp_real) > 36000) {
        last_timestamp = new String(run_file_prev.getValue());
        Integer lastTimestamp = new Integer(last_timestamp);
        lastTimestamp += 1;
        last_timestamp = lastTimestamp.toString();
        logger.info("Last timestamp: " + last_timestamp + " End data: " + endDate);
        Date last_run = new Date(run_file_prev.getTimestamp());
        if (last_run.before(endDate) && !full_run) {
          confArg.put("timestamp_start", last_timestamp);
        }
      }
    }

    Integer tse = new Integer(confArg.get("timestamp_stop"));
    Integer tss = new Integer(confArg.get("timestamp_start"));
    if (tss > tse) {
      logger.info("No new version of requested file.");
      return 0;
    }

    /*
     * Generate file
     */

    String lockName = lock(lockRequest);

    Get file_id_get = new Get(confArg.get("file_id").getBytes());
    Result file_get = db_util.doGet(confArg.get("db_name_files"), file_id_get);
    if (!file_get.isEmpty()) {
      boolean found =
          hasFile(
              db_util,
              hdfs,
              confArg.get("db_name_files"),
              confArg.get("file_id"),
              getFullPath(confArg));
      if (confArg.get("source").equals("fullfile")) {
        found = false;
      }
      String filenames_put =
          getFileNames(
              db_util, confArg.get("db_name_files"), confArg.get("file_id"), getFullPath(confArg));
      // Filename not found in file database
      if (!found && type_move == toFrom.REMOTE2LOCAL) {
        if (!confArg.get("source").equals("local")) {
          // Generate intermediate file
          if (getFile(hdfs, confArg, db_util) == null) {
            unlock(lockName);
            return 1;
          }
          // Put generated file into file database
          if (!confArg.get("format").equals("fullfile")) {
            putFileEntry(
                db_util,
                hdfs,
                confArg.get("db_name_files"),
                confArg.get("file_id"),
                confArg.get("full_file_name"),
                confArg.get("source"));
          }
        } else {
          logger.warn("Remote file not found, and cannot be generated! File: " + confArg);
          unlock(lockName);
          return 1;
        }
      }
    } else {
      if (type_move == toFrom.REMOTE2LOCAL) {
        logger.warn("Remote file not found, and cannot be generated.");
        unlock(lockName);
        return 1;
      }
    }

    /*
     * Copy file
     * Update tables
     */

    if (type_move == toFrom.LOCAL2REMOTE) {
      if (!confArg.get("format").equals("fullfile")) {
        putFileEntry(
            db_util,
            hdfs,
            confArg.get("db_name_files"),
            confArg.get("file_id"),
            getFullPath(confArg),
            confArg.get("source"));
      }
      putRunEntry(
          db_util,
          confArg.get("db_name_runs"),
          confArg.get("run_id"),
          confArg.get("file_id"),
          confArg.get("type"),
          confArg.get("timestamp_real"),
          confArg.get("timestamp_stop"),
          getFullPath(confArg),
          confArg.get("delimiter"));
      hdfs.copyFromLocalFile(new Path(confArg.get("local_path")), new Path(getFullPath(confArg)));
    } else if (type_move == toFrom.REMOTE2LOCAL) {
      FileStatus[] files = hdfs.globStatus(new Path(getFullPath(confArg) + "*"));
      putRunEntry(
          db_util,
          confArg.get("db_name_runs"),
          confArg.get("run_id"),
          confArg.get("file_id"),
          confArg.get("type"),
          confArg.get("timestamp_real"),
          confArg.get("timestamp_stop"),
          getFullPath(confArg),
          confArg.get("delimiter"));
      unlock(lockName);
      for (FileStatus file : files) {
        Path cur_file = file.getPath();
        Path cur_local_path =
            new Path(new String(confArg.get("local_path") + confArg.get("file_id")));
        String suffix = getSuffix(getFileName(confArg), cur_file.getName());
        if (suffix.length() > 0) {
          cur_local_path = cur_local_path.suffix(new String("." + suffix));
        }
        if (confArg.get("copy").equals("true")) {
          String crc = hdfs.getFileChecksum(cur_file).toString();
          if (checksumLocalTest(cur_local_path, crc)) {
            continue;
          } else {
            hdfs.copyToLocalFile(cur_file, cur_local_path);
            writeChecksum(cur_local_path, crc);
          }
        } else {
          System.out.println(cur_local_path + "\t" + cur_file);
        }
      }
    }
    unlock(lockName);
    return 0;
  }
示例#8
0
  /** Sets up configuration based on params */
  private static boolean setup(Hashtable<String, String> curConf, Configuration argConf) {

    if (argConf.get("file") == null) {
      logger.fatal("Missing file parameter");
      System.exit(1);
    }

    if (argConf.get("hdfs_base_path") == null) {
      logger.fatal("Missing HDFS base path, check gestore-conf.xml");
      System.exit(1);
    }

    if (argConf.get("hdfs_temp_path") == null) {
      logger.fatal("Missing HDFS temp path, check gestore-conf.xml");
      System.exit(1);
    }

    if (argConf.get("local_temp_path") == null) {
      logger.fatal("Missing local temp path, check gestore-conf.xml");
      System.exit(1);
    }

    // Input paramaters
    curConf.put("run_id", argConf.get("run", ""));
    curConf.put("task_id", argConf.get("task", ""));
    curConf.put("file_id", argConf.get("file"));
    curConf.put("local_path", argConf.get("path", ""));
    curConf.put("type", argConf.get("type", "l2r"));
    curConf.put("timestamp_start", argConf.get("timestamp_start", "1"));
    curConf.put(
        "timestamp_stop", argConf.get("timestamp_stop", Integer.toString(Integer.MAX_VALUE)));
    curConf.put("delimiter", argConf.get("regex", "ID=.*"));
    curConf.put("taxon", argConf.get("taxon", "all"));
    curConf.put("intermediate", argConf.get("full_run", "false"));
    curConf.put("quick_add", argConf.get("quick_add", "false"));
    Boolean full_run = curConf.get("intermediate").matches("(?i).*true.*");
    curConf.put("format", argConf.get("format", "unknown"));
    curConf.put("split", argConf.get("split", "1"));
    curConf.put("copy", argConf.get("copy", "true"));

    // Constants
    curConf.put("base_path", argConf.get("hdfs_base_path"));
    curConf.put("temp_path", argConf.get("hdfs_temp_path"));
    curConf.put("local_temp_path", argConf.get("local_temp_path"));
    curConf.put("db_name_files", argConf.get("hbase_file_table"));
    curConf.put("db_name_runs", argConf.get("hbase_run_table"));
    curConf.put("db_name_updates", argConf.get("hbase_db_update_table"));

    // Timestamps
    Date currentTime = new Date();
    Date endDate = new Date(new Long(curConf.get("timestamp_stop")));
    curConf.put("timestamp_real", Long.toString(currentTime.getTime()));

    return true;
  }