apache/hadoop · error · FileNotFoundException

Hadoop bin directory is not a directory.: {}

Error message

Hadoop bin directory is not a directory.: {}

What it means

Shell.getQualifiedBinInner() found an entry named bin under the validated Hadoop home, but it is not a directory (java.io.File.isDirectory() is false), so executables cannot live inside it and FileNotFoundException is thrown. Typically a regular file, a broken symlink, or a mount oddity occupies the name 'bin'.

Source

Thrown at hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/util/Shell.java:665

  /**
   * Inner logic of {@link #getQualifiedBin(String)}, accessible
   * for tests.
   * @param hadoopHomeDir home directory (assumed to be valid)
   * @param executable executable
   * @return path to the binary
   * @throws FileNotFoundException if the executable was not found/valid
   */
  static File getQualifiedBinInner(File hadoopHomeDir, String executable)
      throws FileNotFoundException {
    String binDirText = "Hadoop bin directory ";
    File bin = new File(hadoopHomeDir, "bin");
    if (!bin.exists()) {
      throw new FileNotFoundException(addOsText(binDirText + E_DOES_NOT_EXIST
          + ": " + bin));
    }
    if (!bin.isDirectory()) {
      throw new FileNotFoundException(addOsText(binDirText + E_NOT_DIRECTORY
          + ": " + bin));
    }

    File exeFile = new File(bin, executable);
    if (!exeFile.exists()) {
      throw new FileNotFoundException(
          addOsText(E_NO_EXECUTABLE + ": " + exeFile));
    }
    if (!exeFile.isFile()) {
      throw new FileNotFoundException(
          addOsText(E_NOT_EXECUTABLE_FILE + ": " + exeFile));
    }
    try {
      return exeFile.getCanonicalFile();
    } catch (IOException e) {
      // this isn't going to happen, because of all the upfront checks.
      // so if it does, it gets converted to a FNFE and rethrown
      throw fileNotFoundException(e.toString(), e);

View on GitHub (pinned to 2add963021)

Solutions

1. Inspect it: ls -la "$HADOOP_HOME/bin" to see whether it is a file, symlink, or directory
2. If a file, remove/rename it and restore the real bin/ directory from the distribution
3. Repoint broken symlinks at the actual bin directory of the extracted distribution
4. Re-extract the Hadoop (or winutils) archive to rebuild a correct directory tree

Example fix

# before
$ ls -la /opt/hadoop/bin
-rw-r--r-- 1 user user 4096 bin   # a file, not a directory

# after
mv /opt/hadoop/bin /opt/hadoop/bin.broken
tar -xzf hadoop-3.3.6.tar.gz -C /opt/hadoop --strip-components=1 hadoop-3.3.6/bin
ls -lad /opt/hadoop/bin   # drwxr-xr-x ... bin

File bin = new File(new File(System.getenv("HADOOP_HOME")), "bin");
if (bin.exists() && !bin.isDirectory()) {
  throw new IllegalStateException(bin + " exists but is not a directory — reinstall Hadoop bin/");
}

try { Shell.getQualifiedBin(exe); } catch (FileNotFoundException e) { /* inspect bin entry type, repair, retry */ }

When it happens

Trigger: A file named bin (e.g. a script or archive accidentally named 'bin') exists directly under HADOOP_HOME when Shell.getQualifiedBin() resolves an executable; likewise a dangling symlink at ${HADOOP_HOME}/bin.

Common situations: Someone created a helper script literally called bin; a truncated copy left a file mid-transfer; symlinks broken by moving the install; Windows junctions pointing at deleted targets.

Related errors

• Hadoop bin directory does not exist: {}
• %s does not exist or is not file.
• Dest filesystem '${fs.getUri().getScheme()}' doesn't support
• Hadoop home directory {} does not exist
• Hadoop home directory {} is not a directory.

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/086265d40889b0eb. Report an issue: GitHub.