apache/hadoop · error · FileConflictException

File conflicts during rename, {}

Error message

File conflicts during rename, {}

What it means

An ObsException with response code CONFLICT from the POSIX RenameRequest becomes FileConflictException('File conflicts during rename, <responseStatus>'). POSIX-protocol OBS buckets perform a server-side atomic rename that never overwrites: if the destination key already exists, the service rejects the rename.

Source

Thrown at hadoop-cloud-storage-project/hadoop-huaweicloud/src/main/java/org/apache/hadoop/fs/obs/OBSPosixBucketUtils.java:282

  static void innerFsRenameFile(final OBSFileSystem owner,
      final String srcKey,
      final String dstKey) throws IOException {
    LOG.debug("RenameFile path {} to {}", srcKey, dstKey);

    try {
      final RenameRequest renameObjectRequest = new RenameRequest();
      renameObjectRequest.setBucketName(owner.getBucket());
      renameObjectRequest.setObjectKey(srcKey);
      renameObjectRequest.setNewObjectKey(dstKey);
      owner.getObsClient().renameFile(renameObjectRequest);
      owner.getSchemeStatistics().incrementWriteOps(1);
    } catch (ObsException e) {
      if (e.getResponseCode() == OBSCommonUtils.NOT_FOUND_CODE) {
        throw new FileNotFoundException(
            "No such file or directory: " + srcKey);
      }
      if (e.getResponseCode() == OBSCommonUtils.CONFLICT_CODE) {
        throw new FileConflictException(
            "File conflicts during rename, " + e.getResponseStatus());
      }
      throw OBSCommonUtils.translateException(
          "renameFile(" + srcKey + ", " + dstKey + ")", srcKey, e);
    }
  }

  /**
   * Used to rename a source object to a destination object which is not existed
   * before rename.
   *
   * @param owner  OBS File System instance
   * @param srcKey source object key
   * @param dstKey destination object key
   * @throws IOException io exception
   */
  static void fsRenameToNewObject(final OBSFileSystem owner,
      final String srcKey,

View on GitHub (pinned to 2add963021)

Solutions

1. Check fs.exists(dst) first and delete the destination when overwrite is intended
2. Make commits idempotent: missing source plus existing destination means the commit already happened
3. Use unique per-attempt destination names and finalize with a rename of a not-yet-existing key

Example fix

// before
fs.rename(src, dst);

// after
if (fs.exists(dst)) {
  fs.delete(dst, false);
}
fs.rename(src, dst);

if (fs.exists(dst)) {
  fs.delete(dst, false); // deliberate overwrite of an existing destination
}

try {
  fs.rename(src, dst);
} catch (FileConflictException e) {
  if (fs.exists(dst)) {
    // destination taken: delete it or pick a new name, then retry once
  } else {
    throw e;
  }
}

When it happens

Trigger: rename(src, dst) where dst already exists; two clients renaming different sources onto the same destination concurrently; a retried commit whose destination was already written by the first attempt.

Common situations: Commit-by-rename algorithms ported from filesystems with overwrite semantics; concurrent writers to one output path; exactly-once sink implementations that assume rename is idempotent.

Related errors

• {} is not a directory
• No such file or directory: {}
• {} is root directory
• new destination is an existed directory
• new destination is an existed file

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/33da429fc20913cf. Report an issue: GitHub.