apache/hadoop · error · IllegalArgumentException

${pmemDir} is not a directory

Error message

${pmemDir} is not a directory

What it means

Error "${pmemDir} is not a directory" thrown in apache/hadoop.

Source

Thrown at hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/fsdataset/impl/PmemVolumeManager.java:327

      usedBytesCounts.get(volumeIndex).reserve(usedBytes);
    }
    return keyToMappableBlock;
  }

  public void recoverBlockKeyToVolume(ExtendedBlockId key, byte volumeIndex) {
    blockKeyToVolume.put(key, volumeIndex);
  }

  @VisibleForTesting
  static File verifyIfValidPmemVolume(File pmemDir)
      throws IOException {
    if (!pmemDir.exists()) {
      final String message = pmemDir + " does not exist";
      throw new IOException(message);
    }
    if (!pmemDir.isDirectory()) {
      final String message = pmemDir + " is not a directory";
      throw new IllegalArgumentException(message);
    }

    File realPmemDir = new File(getRealPmemDir(pmemDir.getPath()));
    if (!realPmemDir.exists() && !realPmemDir.mkdir()) {
      throw new IOException("Failed to create " + realPmemDir.getPath());
    }

    String uuidStr = UUID.randomUUID().toString();
    String testFilePath = realPmemDir.getPath() + "/.verify.pmem." + uuidStr;
    byte[] contents = uuidStr.getBytes(StandardCharsets.UTF_8);
    RandomAccessFile testFile = null;
    MappedByteBuffer out = null;
    try {
      testFile = new RandomAccessFile(testFilePath, "rw");
      out = testFile.getChannel().map(FileChannel.MapMode.READ_WRITE, 0,
          contents.length);
      if (out == null) {
        throw new IOException(

View on GitHub (pinned to 2add963021)

Solutions

1. Point dfs.datanode.pmem.cache.dirs at a directory, not a regular file; fix or replace the offending path.
2. Recreate the path as a directory on the pmem mount.

When it happens

Trigger: Pmem cache initialization when a configured dfs.datanode.pmem.cache.dirs entry points to a file rather than a directory.

Common situations: See trigger scenarios.

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/817c6b418bfbf911. Report an issue: GitHub.