apache/hadoop · error · IllegalArgumentException

Path cannot be empty.

Error message

Path ${files} cannot be empty.

What it means

After every entry of a -files/-libjars/-archives value is resolved and qualified (and wildcards expanded), validateFiles requires at least one final path; an empty result throws IllegalArgumentException('Path <files> cannot be empty.'). The realistic route is wildcard expansion: expandWildcard only logs '<dir> does not have jars in it. It will be ignored.' for a jar-less directory and adds nothing, so `-libjars 'dir/*'` with zero .jar files in dir ends with an empty list and this error.

Solutions

1. Ensure the wildcard directory actually contains .jar files
2. List the jar files explicitly instead of using '/*' when the directory is not a jar directory
3. Drop the option entirely when nothing needs to be shipped

Example fix

# before: lib/ holds .class files, no jars
hadoop jar job.jar -libjars 'build/lib/*' Driver
# after: point at a real jar directory or name the jar
hadoop jar job.jar -libjars 'build/jars/*' Driver

import java.nio.file.*;
boolean hasJars(String wildcardDir) {
  try (DirectoryStream<Path> s = Files.newDirectoryStream(Paths.get(wildcardDir), "*.jar")) {
    return s.iterator().hasNext();
  } catch (IOException e) {
    return false;
  }
}
// only pass -libjars 'dir/*' when hasJars("dir") is true

When it happens

Trigger: A -libjars/-archives wildcard directory containing no .jar files (only classes, configs, or nothing); every entry filtered out during expansion so finalPaths stays empty.

Common situations: Pointing -libjars at a directory of class files or resources instead of jars; empty lib directories on freshly built projects; CI jobs referencing a lib dir before artifacts are copied in.

Related errors

• File name can't be empty string
• is not a directory.
• Input quota value should be a positive number.
• [ ] = [ ] exceeds max len [ ]
• [ ] = [ ] must be

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/2d1a9f9c6c2b4c1a. Report an issue: GitHub.

Appendix: source

Thrown at hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/util/GenericOptionsParser.java:485

        }
      } else {
        // check if the file exists in this file system
        // we need to recreate this filesystem object to copy
        // these files to the file system ResourceManager is running
        // on.
        FileSystem fs = path.getFileSystem(conf);
        // existence check
        fs.getFileStatus(path);
        if (isWildcard) {
          expandWildcard(finalPaths, path, fs);
        } else {
          finalPaths.add(path.makeQualified(fs.getUri(),
              fs.getWorkingDirectory()).toString());
        }
      }
    }
    if (finalPaths.isEmpty()) {
      throw new IllegalArgumentException("Path " + files + " cannot be empty.");
    }
    return StringUtils.join(",", finalPaths);
  }

  private boolean matchesCurrentDirectory(String path) {
    return path.isEmpty() || path.equals(Path.CUR_DIR) ||
        path.equals(Path.CUR_DIR + File.separator);
  }

  private void expandWildcard(List<String> finalPaths, Path path, FileSystem fs)
      throws IOException {
    FileStatus status = fs.getFileStatus(path);
    if (!status.isDirectory()) {
      throw new FileNotFoundException(path + " is not a directory.");
    }
    // get all the jars in the directory
    List<Path> jars = FileUtil.getJarsInDirectory(path.toString(),
        fs.equals(FileSystem.getLocal(conf)));

View on GitHub (pinned to 2add963021)