apache/hadoop · error · IOException

Mkdirs failed to create " + untarDir

Error message

Mkdirs failed to create " + untarDir

What it means

unTar(InputStream inputStream, File untarDir, boolean gzipped) starts by calling untarDir.mkdirs(); if that returns false and untarDir.isDirectory() is still false it throws IOException("Mkdirs failed to create <untarDir>"). The re-check tolerates a directory that already exists or was concurrently created, so the throw means the extraction target could not be created at all. This fails before any archive bytes are read.

Solutions

1. Check for and remove any regular file occupying untarDir, or pick a fresh unique directory
2. Ensure the parent of untarDir is writable by the process user (mkdir -p by hand to see the OS error)
3. Pre-create the directory yourself with Files.createDirectories() and pass it in — unTar tolerates existing dirs
4. On denials, fix mount options or policy (rw, exec, SELinux context)

Example fix

// before
FileUtil.unTar(in, new File("/tmp/work"), false);
// Mkdirs failed to create /tmp/work

// after
File untarDir = new File("/tmp/work");
if (untarDir.exists() && !untarDir.isDirectory()) {
  Files.delete(untarDir.toPath());
}
Files.createDirectories(untarDir.toPath());
FileUtil.unTar(in, untarDir, false);

File untarDir = new File("/tmp/work");
if (untarDir.exists() && !untarDir.isDirectory()) {
  throw new IOException("Path occupied by a file: " + untarDir);
}
Files.createDirectories(untarDir.toPath()); // pre-create; unTar tolerates existing dirs

When it happens

Trigger: untarDir's parent is not writable; a regular FILE exists at the untarDir path; read-only mount or SELinux denial; the path component is a symlink to a nonexistent location.

Common situations: Extracting to /tmp/${user}-dirs after /tmp was locked down; target path pre-existing as a file from a prior run; containers with read-only rootfs; NFS permission mismatches.

Related errors

• Mkdirs failed to create tar internal dir " + outputDir
• Mkdirs failed to create " + file.getParentFile().toString()
• Mkdirs failed to create " + parent.getAbsolutePath()
• Cannot create directory
• Error executing command.

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/56199c0b8c0c1282. Report an issue: GitHub.

Appendix: source

Thrown at hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/FileUtil.java:990

   * Given a Tar File as input it will untar the file in a the untar directory
   * passed as the second parameter
   *
   * This utility will untar ".tar" files and ".tar.gz","tgz" files.
   *
   * @param inputStream The tar file as input.
   * @param untarDir The untar directory where to untar the tar file.
   * @param gzipped The input stream is gzipped
   *                TODO Use magic number and PusbackInputStream to identify
   * @throws IOException an exception occurred
   * @throws InterruptedException command interrupted
   * @throws ExecutionException task submit failed
   */
  public static void unTar(InputStream inputStream, File untarDir,
                           boolean gzipped)
      throws IOException, InterruptedException, ExecutionException {
    if (!untarDir.mkdirs()) {
      if (!untarDir.isDirectory()) {
        throw new IOException("Mkdirs failed to create " + untarDir);
      }
    }

    if(Shell.WINDOWS) {
      // Tar is not native to Windows. Use simple Java based implementation for
      // tests and simple tar archives
      unTarUsingJava(inputStream, untarDir, gzipped);
    } else {
      // spawn tar utility to untar archive for full fledged unix behavior such
      // as resolving symlinks in tar archives
      unTarUsingTar(inputStream, untarDir, gzipped);
    }
  }

  /**
   * Given a Tar File as input it will untar the file in a the untar directory
   * passed as the second parameter
   *

View on GitHub (pinned to 2add963021)