Esempio n. 1
0
  // the method which actually copies the caches locally and unjars/unzips them
  // and does chmod for the files
  private static Path localizeCache(
      Configuration conf, URI cache, long confFileStamp, CacheStatus cacheStatus, boolean isArchive)
      throws IOException {
    FileSystem fs = getFileSystem(cache, conf);
    FileSystem localFs = FileSystem.getLocal(conf);
    Path parchive = null;

    if (isArchive) {
      parchive =
          new Path(
              cacheStatus.localizedLoadPath, new Path(cacheStatus.localizedLoadPath.getName()));
    } else {
      parchive = cacheStatus.localizedLoadPath;
    }
    if (!localFs.mkdirs(parchive.getParent())) {
      throw new IOException(
          "Mkdirs failed to create directory " + cacheStatus.localizedLoadPath.toString());
    }
    String cacheId = cache.getPath();

    fs.copyToLocalFile(new Path(cacheId), parchive);
    if (isArchive) {
      String tmpArchive = parchive.toString().toLowerCase();
      File srcFile = new File(parchive.toString());
      File destDir = new File(parchive.getParent().toString());
      if (tmpArchive.endsWith(".jar")) {
        RunJar.unJar(srcFile, destDir);
      } else if (tmpArchive.endsWith(".zip")) {
        FileUtil.unZip(srcFile, destDir);
      } else if (isTarFile(tmpArchive)) {
        FileUtil.unTar(srcFile, destDir);
      }
      // else will not do anyhting
      // and copy the file into the dir as it is
    }
    long cacheSize = FileUtil.getDU(new File(parchive.getParent().toString()));
    cacheStatus.size = cacheSize;
    addCacheInfoUpdate(cacheStatus);

    // do chmod here
    try {
      // Setting recursive permission to grant everyone read and execute
      Path localDir = new Path(cacheStatus.localizedBaseDir, cacheStatus.uniqueParentDir);
      LOG.info("Doing chmod on localdir :" + localDir);
      FileUtil.chmod(localDir.toString(), "ugo+rx", true);
    } catch (InterruptedException e) {
      LOG.warn("Exception in chmod" + e.toString());
    }

    // update cacheStatus to reflect the newly cached file
    cacheStatus.mtime = getTimestamp(conf, cache);
    return cacheStatus.localizedLoadPath;
  }
  /**
   * Method to move files from HDFS to local filesystem
   *
   * <p>localPath: Path on the machines filesystem fs:FileSystem object from HDFS pathList:List of
   * paths for files that might need to be backed up size:max size in bytes to be backed up
   *
   * <p>ReturnsDate of the last files backed up if reached size limit, else, zero
   */
  public long backupFiles(
      String localPath, String preservePath, FileSystem fs, ArrayList<Path> pathList, long size) {
    Path fsPath;
    long tmpSize = 0;
    long tmpDate = 0;

    // Start iterating over all paths
    for (Path hdfsPath : pathList) {
      try {
        long nFileSize = fs.getContentSummary(hdfsPath).getLength();
        tmpSize = tmpSize + nFileSize;

        if ((tmpSize <= size) || (size == 0)) {
          FileStatus stat = fs.getFileStatus(hdfsPath);

          System.err.println(
              "File "
                  + hdfsPath.toUri().getPath()
                  + " "
                  + nFileSize
                  + " bytes, "
                  + "perms: "
                  + stat.getOwner()
                  + "/"
                  + stat.getGroup()
                  + ", "
                  + stat.getPermission().toString());

          tmpDate = stat.getModificationTime() / 1000;

          String sFsPath = localPath + hdfsPath.toUri().getPath();
          fsPath = new Path(sFsPath);

          File f = new File(sFsPath);

          // COMMENTED OUT: until a few backup cycles run
          // and the mtime gets in fact set on all copied
          // files.
          //
          // ignore it if the file exists and has the same mtime
          // if (f.exists() && f.isFile() && f.lastModified() == stat.getModificationTime())
          // {
          // System.out.println("no need to backup " + f.toString() + ", mtime matches hdfs");
          // continue;
          // }

          if (false == m_bDryRun) {
            // check if we need to back up the local file
            // (not directory), if it already exists.
            if (f.exists() && f.isFile()) {
              // ignore files with substrings in the
              // no-preserve file
              if (true == doPreserveFile(sFsPath)) {
                // move it to the backup path
                String sNewPath = preservePath + hdfsPath.toUri().getPath();
                File newFile = new File(sNewPath);

                // create directory structure for new file?
                if (false == newFile.getParentFile().exists()) {
                  if (false == newFile.getParentFile().mkdirs()) {
                    System.err.println("Failed to mkdirs " + newFile.getParentFile().toString());
                    System.exit(1);
                  }
                }

                // rename existing file to new location
                if (false == f.renameTo(newFile)) {
                  System.err.println(
                      "Failed to renameTo " + f.toString() + " to " + newFile.toString());
                  System.exit(1);
                }

                System.out.println("preserved " + f.toString() + " into " + newFile.toString());
              } else {
                System.out.println("skipped preservation of " + f.toString());
              }
            }

            // copy from hdfs to local filesystem
            fs.copyToLocalFile(hdfsPath, fsPath);

            // set the mtime to match hdfs file
            f.setLastModified(stat.getModificationTime());

            // compare checksums on both files
            compareChecksums(fs, hdfsPath, sFsPath);
          }

          // don't print the progress after every file -- go
          // by at least 1% increments
          long nPercentDone = (long) (100 * tmpSize / m_nTotalBytes);
          if (nPercentDone > m_nLastPercentBytesDone) {
            System.out.println(
                "progress: copied "
                    + prettyPrintBytes(tmpSize)
                    + ", "
                    + nPercentDone
                    + "% done"
                    + ", tstamp="
                    + tmpDate);

            m_nLastPercentBytesDone = nPercentDone;
          }

          if (m_nSleepSeconds > 0) {
            try {
              Thread.sleep(1000 * m_nSleepSeconds);
            } catch (Exception e2) {
              // ignore
            }
          }
        } else {
          return tmpDate;
        }
      } catch (IOException e) {
        System.err.println("FATAL ERROR: Something wrong with the file");
        System.err.println(e);
        System.out.println(tmpDate);
        System.exit(1);

        return 0;
      }
    }

    return 0;
  }
Esempio n. 3
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;
  }