apache/hadoop · error · IllegalArgumentException

Unknown OpenFileType: {}

Error message

Unknown OpenFileType: {}

What it means

FSNamesystem.listOpenFiles dispatches on the EnumSet of OpenFilesType supplied by the client. The deployed enum has only ALL_OPEN_FILES and BLOCKING_DECOMMISSION; if the set contains neither (empty set or a value unknown to this NameNode), control falls into the final else branch and throws IllegalArgumentException('Unknown OpenFileType: ...').

Source

Thrown at hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/FSNamesystem.java:2091

      EnumSet<OpenFilesType> openFilesTypes, String path) throws IOException {
    INode.checkAbsolutePath(path);
    final String operationName = "listOpenFiles";
    checkSuperuserPrivilege(operationName, path);
    checkOperation(OperationCategory.READ);
    BatchedListEntries<OpenFileEntry> batchedListEntries;
    String normalizedPath = new Path(path).toString(); // normalize path.
    try {
      readLock(RwLockMode.FS);
      try {
        checkOperation(OperationCategory.READ);
        if (openFilesTypes.contains(OpenFilesType.ALL_OPEN_FILES)) {
          batchedListEntries = leaseManager.getUnderConstructionFiles(prevId,
              normalizedPath);
        } else {
          if (openFilesTypes.contains(OpenFilesType.BLOCKING_DECOMMISSION)) {
            batchedListEntries = getFilesBlockingDecom(prevId, normalizedPath);
          } else {
            throw new IllegalArgumentException("Unknown OpenFileType: "
                + openFilesTypes);
          }
        }
      } finally {
        readUnlock(RwLockMode.FS, operationName, getLockReportInfoSupplier(null));
      }
    } catch (AccessControlException e) {
      logAuditEvent(false, operationName, null);
      throw e;
    }
    logAuditEvent(true, operationName, null);
    return batchedListEntries;
  }

  public BatchedListEntries<OpenFileEntry> getFilesBlockingDecom(long prevId,
      String path) {
    assert hasReadLock(RwLockMode.FS);
    final List<OpenFileEntry> openFileEntries = Lists.newArrayList();

View on GitHub (pinned to 2add963021)

Solutions

1. Pass a valid non-empty set: EnumSet.of(OpenFilesType.ALL_OPEN_FILES) or EnumSet.of(OpenFilesType.BLOCKING_DECOMMISSION)
2. Rebuild the client against the same Hadoop version as the NameNode to remove enum skew
3. Use `hdfs dfsadmin -listOpenFiles` (optionally -blockingDecommission), which always sends a valid set

Example fix

// before
OpenFilesType[] types = typesFromConfig; // may be empty -> IllegalArgumentException
RemoteIterator<OpenFileEntry> it = dfs.listOpenFiles(types);
// after
EnumSet<OpenFilesType> set = (types == null || types.length == 0)
    ? EnumSet.of(OpenFilesType.ALL_OPEN_FILES)
    : EnumSet.noneOf(OpenFilesType.class);
if (types != null) Collections.addAll(set, types);
RemoteIterator<OpenFileEntry> it = dfs.listOpenFiles(set.toArray(new OpenFilesType[0]));

Set<OpenFilesType> t = (types == null || types.isEmpty())
    ? EnumSet.of(OpenFilesType.ALL_OPEN_FILES) : EnumSet.copyOf(types);
boolean valid = t.stream().allMatch(v ->
    v == OpenFilesType.ALL_OPEN_FILES || v == OpenFilesType.BLOCKING_DECOMMISSION);
if (!valid) t = EnumSet.of(OpenFilesType.ALL_OPEN_FILES);

static boolean isOpenFilesQueryValid(Collection<OpenFilesType> c) {
  if (c == null || c.isEmpty()) return false;
  for (OpenFilesType v : c) {
    if (v != OpenFilesType.ALL_OPEN_FILES && v != OpenFilesType.BLOCKING_DECOMMISSION) return false;
  }
  return true;
}

catch (IllegalArgumentException e) { if (e.getMessage() != null && e.getMessage().contains("Unknown OpenFileType")) { query = EnumSet.of(OpenFilesType.ALL_OPEN_FILES); continue; } throw e; }

When it happens

Trigger: Calling ClientProtocol.listOpenFiles / DistributedFileSystem.listOpenFiles with an empty OpenFilesType array (e.g. EnumSet.noneOf(OpenFilesType.class)), or a client compiled against a Hadoop version whose OpenFilesType values differ from the NameNode's so the deserialized set matches no branch.

Common situations: Custom monitoring tooling that builds the filter set from config and ends up empty; client/server version skew after a rolling upgrade; code that treats a null filter as 'none' instead of defaulting to 'all'.

Related errors

• Unexpected HAServiceStateProto:
• Illagal value: {NAME} = {value} < MIN = {MIN}
• Unknown nameservice: {}
• Configuration has multiple addresses that match local node's
• Cannot delete/rename non-empty protected directory {}

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/10665d1bafa38f31. Report an issue: GitHub.