Examples of DistCp

org.apache.hadoop.tools.DistCp
DistCp is the main driver-class for DistCpV2. For command-line use, DistCp::main() orchestrates the parsing of command-line parameters and the launch of the DistCp job. For programmatic use, a DistCp object can be constructed by specifying options (in a DistCpOptions object), and DistCp::execute() may be used to launch the copy-job. DistCp may alternatively be sub-classed to fine-tune behaviour.
org.springframework.data.hadoop.fs.DistCp
p.apache.org/common/docs/current/distcp.html">distcp as an embeddable API. Due to the number of options available in DistCp, one can either specify them in a command-like style (through one or multiple {@link String}s) through {@link #copy(String)} or specify individual arguments through the rest of the methods. @author Costin Leau @author Thomas Risberg

Examples of org.apache.hadoop.tools.DistCp

      cluster = new MiniDFSCluster(conf, 1, true, null);
      final FileSystem hdfs = cluster.getFileSystem();
      final String namenode = hdfs.getUri().toString();
      if (namenode.startsWith("hdfs://")) {
        MyFile[] files = createFiles(LOCAL_FS, TEST_ROOT_DIR+"/srcdat");
        ToolRunner.run(new DistCp(conf), new String[] {
                                         "-log",
                                         namenode+"/logs",
                                         "file:///"+TEST_ROOT_DIR+"/srcdat",
                                         namenode+"/destdat"});
        assertTrue("Source and destination directories do not match.",

View Full Code Here

Examples of org.apache.hadoop.tools.DistCp

      cluster = new MiniDFSCluster(conf, 1, true, null);
      final FileSystem hdfs = cluster.getFileSystem();
      final String namenode = FileSystem.getDefaultUri(conf).toString();
      if (namenode.startsWith("hdfs://")) {
        MyFile[] files = createFiles(URI.create(namenode), "/srcdat");
        ToolRunner.run(new DistCp(conf), new String[] {
                                         "-log",
                                         "/logs",
                                         namenode+"/srcdat",
                                         "file:///"+TEST_ROOT_DIR+"/destdat"});
        assertTrue("Source and destination directories do not match.",

View Full Code Here

Examples of org.apache.hadoop.tools.DistCp

      cluster = new MiniDFSCluster(conf, 2, true, null);
      final FileSystem hdfs = cluster.getFileSystem();
      final String namenode = hdfs.getUri().toString();
      if (namenode.startsWith("hdfs://")) {
        MyFile[] files = createFiles(URI.create(namenode), "/srcdat");
        ToolRunner.run(new DistCp(conf), new String[] {
                                         "-p",
                                         "-log",
                                         namenode+"/logs",
                                         namenode+"/srcdat",
                                         namenode+"/destdat"});
        assertTrue("Source and destination directories do not match.",
                   checkFiles(hdfs, "/destdat", files));
        FileSystem fs = FileSystem.get(URI.create(namenode+"/logs"), conf);
        assertTrue("Log directory does not exist.",
                    fs.exists(new Path(namenode+"/logs")));


        FileStatus[] dchkpoint = getFileStatus(hdfs, "/destdat", files);
        final int nupdate = NFILES>>2;
        updateFiles(cluster.getFileSystem(), "/srcdat", files, nupdate);
        deldir(hdfs, "/logs");


        ToolRunner.run(new DistCp(conf), new String[] {
                                         "-p",
                                         "-update",
                                         "-log",
                                         namenode+"/logs",
                                         namenode+"/srcdat",
                                         namenode+"/destdat"});
        assertTrue("Source and destination directories do not match.",
                   checkFiles(hdfs, "/destdat", files));
        assertTrue("Update failed to replicate all changes in src",
                 checkUpdate(hdfs, dchkpoint, "/destdat", files, nupdate));


        deldir(hdfs, "/logs");
        ToolRunner.run(new DistCp(conf), new String[] {
                                         "-p",
                                         "-overwrite",
                                         "-log",
                                         namenode+"/logs",
                                         namenode+"/srcdat",

View Full Code Here

Examples of org.apache.hadoop.tools.DistCp


  public void testCopyDuplication() throws Exception {
    final FileSystem localfs = FileSystem.get(LOCAL_FS, new Configuration());
    try {    
      MyFile[] files = createFiles(localfs, TEST_ROOT_DIR+"/srcdat");
      ToolRunner.run(new DistCp(new Configuration()),
          new String[] {"file:///"+TEST_ROOT_DIR+"/srcdat",
                        "file:///"+TEST_ROOT_DIR+"/src2/srcdat"});
      assertTrue("Source and destination directories do not match.",
                 checkFiles(localfs, TEST_ROOT_DIR+"/src2/srcdat", files));
  
      assertEquals(DistCp.DuplicationException.ERROR_CODE,
          ToolRunner.run(new DistCp(new Configuration()),
          new String[] {"file:///"+TEST_ROOT_DIR+"/srcdat",
                        "file:///"+TEST_ROOT_DIR+"/src2/srcdat",
                        "file:///"+TEST_ROOT_DIR+"/destdat",}));
    }
    finally {

View Full Code Here

Examples of org.apache.hadoop.tools.DistCp

    FileSystem fs = FileSystem.get(LOCAL_FS, new Configuration());
    Path root = new Path(TEST_ROOT_DIR+"/srcdat");
    try {    
      MyFile[] files = {createFile(root, fs)};
      //copy a dir with a single file
      ToolRunner.run(new DistCp(new Configuration()),
          new String[] {"file:///"+TEST_ROOT_DIR+"/srcdat",
                        "file:///"+TEST_ROOT_DIR+"/destdat"});
      assertTrue("Source and destination directories do not match.",
                 checkFiles(fs, TEST_ROOT_DIR+"/destdat", files));
      
      //copy a single file
      String fname = files[0].getName();
      Path p = new Path(root, fname);
      FileSystem.LOG.info("fname=" + fname + ", exists? " + fs.exists(p));
      ToolRunner.run(new DistCp(new Configuration()),
          new String[] {"file:///"+TEST_ROOT_DIR+"/srcdat/"+fname,
                        "file:///"+TEST_ROOT_DIR+"/dest2/"+fname});
      assertTrue("Source and destination directories do not match.",
          checkFiles(fs, TEST_ROOT_DIR+"/dest2", files));     
      //copy single file to existing dir
      deldir(fs, TEST_ROOT_DIR+"/dest2");
      fs.mkdirs(new Path(TEST_ROOT_DIR+"/dest2"));
      MyFile[] files2 = {createFile(root, fs, 0)};
      String sname = files2[0].getName();
      ToolRunner.run(new DistCp(new Configuration()),
          new String[] {"-update",
                        "file:///"+TEST_ROOT_DIR+"/srcdat/"+sname,
                        "file:///"+TEST_ROOT_DIR+"/dest2/"});
      assertTrue("Source and destination directories do not match.",
          checkFiles(fs, TEST_ROOT_DIR+"/dest2", files2));     
      updateFiles(fs, TEST_ROOT_DIR+"/srcdat", files2, 1);
      //copy single file to existing dir w/ dst name conflict
      ToolRunner.run(new DistCp(new Configuration()),
          new String[] {"-update",
                        "file:///"+TEST_ROOT_DIR+"/srcdat/"+sname,
                        "file:///"+TEST_ROOT_DIR+"/dest2/"});
      assertTrue("Source and destination directories do not match.",
          checkFiles(fs, TEST_ROOT_DIR+"/dest2", files2));

View Full Code Here

Examples of org.apache.hadoop.tools.DistCp

        MyFile[] files = createFiles(URI.create(nnUri), "/srcdat");
        FileStatus[] srcstat = getFileStatus(fs, "/srcdat", files);
        for(int i = 0; i < srcstat.length; i++) {
          fs.setOwner(srcstat[i].getPath(), "u" + i, null);
        }
        ToolRunner.run(new DistCp(conf),
            new String[]{"-pu", nnUri+"/srcdat", nnUri+"/destdat"});
        assertTrue("Source and destination directories do not match.",
                   checkFiles(fs, "/destdat", files));
        
        FileStatus[] dststat = getFileStatus(fs, "/destdat", files);
        for(int i = 0; i < dststat.length; i++) {
          assertEquals("i=" + i, "u" + i, dststat[i].getOwner());
        }
        deldir(fs, "/destdat");
        deldir(fs, "/srcdat");
      }


      {//test preserving group
        MyFile[] files = createFiles(URI.create(nnUri), "/srcdat");
        FileStatus[] srcstat = getFileStatus(fs, "/srcdat", files);
        for(int i = 0; i < srcstat.length; i++) {
          fs.setOwner(srcstat[i].getPath(), null, "g" + i);
        }
        ToolRunner.run(new DistCp(conf),
            new String[]{"-pg", nnUri+"/srcdat", nnUri+"/destdat"});
        assertTrue("Source and destination directories do not match.",
                   checkFiles(fs, "/destdat", files));
        
        FileStatus[] dststat = getFileStatus(fs, "/destdat", files);
        for(int i = 0; i < dststat.length; i++) {
          assertEquals("i=" + i, "g" + i, dststat[i].getGroup());
        }
        deldir(fs, "/destdat");
        deldir(fs, "/srcdat");
      }


      {//test preserving mode
        MyFile[] files = createFiles(URI.create(nnUri), "/srcdat");
        FileStatus[] srcstat = getFileStatus(fs, "/srcdat", files);
        FsPermission[] permissions = new FsPermission[srcstat.length];
        for(int i = 0; i < srcstat.length; i++) {
          permissions[i] = new FsPermission((short)(i & 0666));
          fs.setPermission(srcstat[i].getPath(), permissions[i]);
        }


        ToolRunner.run(new DistCp(conf),
            new String[]{"-pp", nnUri+"/srcdat", nnUri+"/destdat"});
        assertTrue("Source and destination directories do not match.",
                   checkFiles(fs, "/destdat", files));
  
        FileStatus[] dststat = getFileStatus(fs, "/destdat", files);

View Full Code Here

Examples of org.apache.hadoop.tools.DistCp

      for (MyFile f : files) {
        totsize += f.getSize();
      }
      Configuration job = mr.createJobConf();
      job.setLong("distcp.bytes.per.map", totsize / 3);
      ToolRunner.run(new DistCp(job),
          new String[] {"-m", "100",
                        "-log",
                        namenode+"/logs",
                        namenode+"/srcdat",
                        namenode+"/destdat"});
      assertTrue("Source and destination directories do not match.",
                 checkFiles(fs, "/destdat", files));


      String logdir = namenode + "/logs";
      System.out.println(execCmd(shell, "-lsr", logdir));
      FileStatus[] logs = fs.listStatus(new Path(logdir));
      // rare case where splits are exact, logs.length can be 4
      assertTrue("Unexpected map count, logs.length=" + logs.length,
          logs.length == 5 || logs.length == 4);


      deldir(fs, "/destdat");
      deldir(fs, "/logs");
      ToolRunner.run(new DistCp(job),
          new String[] {"-m", "1",
                        "-log",
                        namenode+"/logs",
                        namenode+"/srcdat",
                        namenode+"/destdat"});

View Full Code Here

Examples of org.apache.hadoop.tools.DistCp

    MiniDFSCluster cluster = null;
    try {
      cluster = new MiniDFSCluster(conf, 2, true, null);
      final String nnUri = FileSystem.getDefaultUri(conf).toString();
      final FileSystem fs = FileSystem.get(URI.create(nnUri), conf);
      final DistCp distcp = new DistCp(conf);
      final FsShell shell = new FsShell(conf);  


      final String srcrootdir =  "/src_root";
      final Path srcrootpath = new Path(srcrootdir); 
      final String dstrootdir =  "/dst_root";

View Full Code Here

Examples of org.apache.hadoop.tools.DistCp


      final Path srcrootpath = new Path(home, "src_root"); 
      final String srcrootdir =  srcrootpath.toString();
      final Path dstrootpath = new Path(home, "dst_root"); 
      final String dstrootdir =  dstrootpath.toString();
      final DistCp distcp = new DistCp(userConf);


      FileSystem.mkdirs(fs, srcrootpath, new FsPermission((short)0700));
      final String[] args = {"hftp://"+httpAdd+srcrootdir, nnUri+dstrootdir};


      { //copy with permission 000, should fail

View Full Code Here

Examples of org.apache.hadoop.tools.DistCp

      cluster = new MiniDFSCluster(conf, 2, true, null);
      final URI nnURI = FileSystem.getDefaultUri(conf);
      final String nnUri = nnURI.toString();
      final FileSystem fs = FileSystem.get(URI.create(nnUri), conf);


      final DistCp distcp = new DistCp(conf);
      final FsShell shell = new FsShell(conf);  


      final String srcrootdir = "/src_root";
      final String dstrootdir = "/dst_root";

View Full Code Here

0 1 2 3 4 5

TOP

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.