Java DistCpOptions.setSyncFolder方法代码示例

import org.apache.hadoop.tools.DistCpOptions; //导入方法依赖的package包/类
public DistCpOptions getDistCpOptions(String[] args) throws ParseException {
    Options options = new Options();
    Option opt;
    opt = new Option("maxMaps", true,
            "max number of maps to use for this copy");
    opt.setRequired(true);
    options.addOption(opt);

    opt = new Option("sourcePaths", true,
            "comma separtated list of source paths to be copied");
    opt.setRequired(true);
    options.addOption(opt);

    opt = new Option("targetPath", true, "target path");
    opt.setRequired(true);
    options.addOption(opt);

    CommandLine cmd = new GnuParser().parse(options, args);
    String[] paths = cmd.getOptionValue("sourcePaths").trim().split(",");
    List<Path> srcPaths = getPaths(paths);
    String trgPath = cmd.getOptionValue("targetPath").trim();

    DistCpOptions distcpOptions = new DistCpOptions(srcPaths, new Path(
            trgPath));
    distcpOptions.setSyncFolder(true);
    distcpOptions.setBlocking(true);
    distcpOptions
            .setMaxMaps(Integer.valueOf(cmd.getOptionValue("maxMaps")));

    return distcpOptions;
}
 

import org.apache.hadoop.tools.DistCpOptions; //导入方法依赖的package包/类
@Test
public void testDeleteMissing() {
  TaskAttemptContext taskAttemptContext = getTaskAttemptContext(config);
  JobContext jobContext = new JobContextImpl(taskAttemptContext.getConfiguration(),
      taskAttemptContext.getTaskAttemptID().getJobID());
  Configuration conf = jobContext.getConfiguration();

  String sourceBase;
  String targetBase;
  FileSystem fs = null;
  try {
    OutputCommitter committer = new CopyCommitter(null, taskAttemptContext);
    fs = FileSystem.get(conf);
    sourceBase = TestDistCpUtils.createTestSetup(fs, FsPermission.getDefault());
    targetBase = TestDistCpUtils.createTestSetup(fs, FsPermission.getDefault());
    String targetBaseAdd = TestDistCpUtils.createTestSetup(fs, FsPermission.getDefault());
    fs.rename(new Path(targetBaseAdd), new Path(targetBase));

    DistCpOptions options = new DistCpOptions(Arrays.asList(new Path(sourceBase)),
        new Path("/out"));
    options.setSyncFolder(true);
    options.setDeleteMissing(true);
    options.appendToConf(conf);

    CopyListing listing = new GlobbedCopyListing(conf, CREDENTIALS);
    Path listingFile = new Path("/tmp1/" + String.valueOf(rand.nextLong()));
    listing.buildListing(listingFile, options);

    conf.set(DistCpConstants.CONF_LABEL_TARGET_WORK_PATH, targetBase);
    conf.set(DistCpConstants.CONF_LABEL_TARGET_FINAL_PATH, targetBase);

    committer.commitJob(jobContext);
    if (!TestDistCpUtils.checkIfFoldersAreInSync(fs, targetBase, sourceBase)) {
      Assert.fail("Source and target folders are not in sync");
    }
    if (!TestDistCpUtils.checkIfFoldersAreInSync(fs, sourceBase, targetBase)) {
      Assert.fail("Source and target folders are not in sync");
    }

    //Test for idempotent commit
    committer.commitJob(jobContext);
    if (!TestDistCpUtils.checkIfFoldersAreInSync(fs, targetBase, sourceBase)) {
      Assert.fail("Source and target folders are not in sync");
    }
    if (!TestDistCpUtils.checkIfFoldersAreInSync(fs, sourceBase, targetBase)) {
      Assert.fail("Source and target folders are not in sync");
    }
  } catch (Throwable e) {
    LOG.error("Exception encountered while testing for delete missing", e);
    Assert.fail("Delete missing failure");
  } finally {
    TestDistCpUtils.delete(fs, "/tmp1");
    conf.set(DistCpConstants.CONF_LABEL_DELETE_MISSING, "false");
  }
}
 

import org.apache.hadoop.tools.DistCpOptions; //导入方法依赖的package包/类
@Test
public void testDeleteMissingFlatInterleavedFiles() {
  TaskAttemptContext taskAttemptContext = getTaskAttemptContext(config);
  JobContext jobContext = new JobContextImpl(taskAttemptContext.getConfiguration(),
      taskAttemptContext.getTaskAttemptID().getJobID());
  Configuration conf = jobContext.getConfiguration();


  String sourceBase;
  String targetBase;
  FileSystem fs = null;
  try {
    OutputCommitter committer = new CopyCommitter(null, taskAttemptContext);
    fs = FileSystem.get(conf);
    sourceBase = "/tmp1/" + String.valueOf(rand.nextLong());
    targetBase = "/tmp1/" + String.valueOf(rand.nextLong());
    TestDistCpUtils.createFile(fs, sourceBase + "/1");
    TestDistCpUtils.createFile(fs, sourceBase + "/3");
    TestDistCpUtils.createFile(fs, sourceBase + "/4");
    TestDistCpUtils.createFile(fs, sourceBase + "/5");
    TestDistCpUtils.createFile(fs, sourceBase + "/7");
    TestDistCpUtils.createFile(fs, sourceBase + "/8");
    TestDistCpUtils.createFile(fs, sourceBase + "/9");

    TestDistCpUtils.createFile(fs, targetBase + "/2");
    TestDistCpUtils.createFile(fs, targetBase + "/4");
    TestDistCpUtils.createFile(fs, targetBase + "/5");
    TestDistCpUtils.createFile(fs, targetBase + "/7");
    TestDistCpUtils.createFile(fs, targetBase + "/9");
    TestDistCpUtils.createFile(fs, targetBase + "/A");

    DistCpOptions options = new DistCpOptions(Arrays.asList(new Path(sourceBase)), 
        new Path("/out"));
    options.setSyncFolder(true);
    options.setDeleteMissing(true);
    options.appendToConf(conf);

    CopyListing listing = new GlobbedCopyListing(conf, CREDENTIALS);
    Path listingFile = new Path("/tmp1/" + String.valueOf(rand.nextLong()));
    listing.buildListing(listingFile, options);

    conf.set(DistCpConstants.CONF_LABEL_TARGET_WORK_PATH, targetBase);
    conf.set(DistCpConstants.CONF_LABEL_TARGET_FINAL_PATH, targetBase);

    committer.commitJob(jobContext);
    if (!TestDistCpUtils.checkIfFoldersAreInSync(fs, targetBase, sourceBase)) {
      Assert.fail("Source and target folders are not in sync");
    }
    Assert.assertEquals(fs.listStatus(new Path(targetBase)).length, 4);

    //Test for idempotent commit
    committer.commitJob(jobContext);
    if (!TestDistCpUtils.checkIfFoldersAreInSync(fs, targetBase, sourceBase)) {
      Assert.fail("Source and target folders are not in sync");
    }
    Assert.assertEquals(fs.listStatus(new Path(targetBase)).length, 4);
  } catch (IOException e) {
    LOG.error("Exception encountered while testing for delete missing", e);
    Assert.fail("Delete missing failure");
  } finally {
    TestDistCpUtils.delete(fs, "/tmp1");
    conf.set(DistCpConstants.CONF_LABEL_DELETE_MISSING, "false");
  }

}
 

import org.apache.hadoop.tools.DistCpOptions; //导入方法依赖的package包/类
private void deleteMissing(Configuration conf) throws IOException {
  LOG.info("-delete option is enabled. About to remove entries from " +
      "target that are missing in source");

  // Sort the source-file listing alphabetically.
  Path sourceListing = new Path(conf.get(DistCpConstants.CONF_LABEL_LISTING_FILE_PATH));
  FileSystem clusterFS = sourceListing.getFileSystem(conf);
  Path sortedSourceListing = DistCpUtils.sortListing(clusterFS, conf, sourceListing);

  // Similarly, create the listing of target-files. Sort alphabetically.
  Path targetListing = new Path(sourceListing.getParent(), "targetListing.seq");
  CopyListing target = new GlobbedCopyListing(new Configuration(conf), null);

  List<Path> targets = new ArrayList<Path>(1);
  Path targetFinalPath = new Path(conf.get(DistCpConstants.CONF_LABEL_TARGET_FINAL_PATH));
  targets.add(targetFinalPath);
  Path resultNonePath = Path.getPathWithoutSchemeAndAuthority(targetFinalPath)
      .toString().startsWith(DistCpConstants.HDFS_RESERVED_RAW_DIRECTORY_NAME)
      ? DistCpConstants.RAW_NONE_PATH : DistCpConstants.NONE_PATH;
  DistCpOptions options = new DistCpOptions(targets, resultNonePath);
  //
  // Set up options to be the same from the CopyListing.buildListing's perspective,
  // so to collect similar listings as when doing the copy
  //
  options.setOverwrite(overwrite);
  options.setSyncFolder(syncFolder);
  options.setTargetPathExists(targetPathExists);
  
  target.buildListing(targetListing, options);
  Path sortedTargetListing = DistCpUtils.sortListing(clusterFS, conf, targetListing);
  long totalLen = clusterFS.getFileStatus(sortedTargetListing).getLen();

  SequenceFile.Reader sourceReader = new SequenceFile.Reader(conf,
                               SequenceFile.Reader.file(sortedSourceListing));
  SequenceFile.Reader targetReader = new SequenceFile.Reader(conf,
                               SequenceFile.Reader.file(sortedTargetListing));

  // Walk both source and target file listings.
  // Delete all from target that doesn't also exist on source.
  long deletedEntries = 0;
  try {
    CopyListingFileStatus srcFileStatus = new CopyListingFileStatus();
    Text srcRelPath = new Text();
    CopyListingFileStatus trgtFileStatus = new CopyListingFileStatus();
    Text trgtRelPath = new Text();

    FileSystem targetFS = targetFinalPath.getFileSystem(conf);
    boolean srcAvailable = sourceReader.next(srcRelPath, srcFileStatus);
    while (targetReader.next(trgtRelPath, trgtFileStatus)) {
      // Skip sources that don't exist on target.
      while (srcAvailable && trgtRelPath.compareTo(srcRelPath) > 0) {
        srcAvailable = sourceReader.next(srcRelPath, srcFileStatus);
      }

      if (srcAvailable && trgtRelPath.equals(srcRelPath)) continue;

      // Target doesn't exist at source. Delete.
      boolean result = (!targetFS.exists(trgtFileStatus.getPath()) ||
          targetFS.delete(trgtFileStatus.getPath(), true));
      if (result) {
        LOG.info("Deleted " + trgtFileStatus.getPath() + " - Missing at source");
        deletedEntries++;
      } else {
        throw new IOException("Unable to delete " + trgtFileStatus.getPath());
      }
      taskAttemptContext.progress();
      taskAttemptContext.setStatus("Deleting missing files from target. [" +
          targetReader.getPosition() * 100 / totalLen + "%]");
    }
  } finally {
    IOUtils.closeStream(sourceReader);
    IOUtils.closeStream(targetReader);
  }
  LOG.info("Deleted " + deletedEntries + " from target: " + targets.get(0));
}
 

import org.apache.hadoop.tools.DistCpOptions; //导入方法依赖的package包/类
public DistCpOptions getDistCpOptions(String[] args) throws ParseException {
	Options options = new Options();
	Option opt;
	opt = new Option("update", false,
			"specify update for synching folders");
	opt.setRequired(true);
	options.addOption(opt);

	opt = new Option("blocking", true,
			"should DistCp be running in blocking mode");
	opt.setRequired(true);
	options.addOption(opt);

	opt = new Option("maxMaps", true,
			"max number of maps to use for this copy");
	opt.setRequired(true);
	options.addOption(opt);

       opt = new Option("sourcePaths", true,
			"comma separtated list of source paths to be copied");
	opt.setRequired(true);
	options.addOption(opt);

       opt = new Option("targetPath", true, "target path");
	opt.setRequired(true);
	options.addOption(opt);

	CommandLine cmd = new GnuParser().parse(options, args);
	String[] paths = cmd.getOptionValue("sourcePaths").trim().split(",");
	List<Path> srcPaths = getPaths(paths);
	String trgPath = cmd.getOptionValue("targetPath").trim();

	DistCpOptions distcpOptions = new DistCpOptions(srcPaths, new Path(
			trgPath));
       distcpOptions.setSyncFolder(true);
	distcpOptions.setBlocking(Boolean.valueOf(cmd
			.getOptionValue("blocking")));
	distcpOptions
			.setMaxMaps(Integer.valueOf(cmd.getOptionValue("maxMaps")));

	return distcpOptions;
}