apache/hadoop · error · ExitUtil.ExitException

46

46

Error message

The -zone option is only supported for Amazon S3 Express One Zone Storage

What it means

Fail-fast check in the bucket tool before instantiating the filesystem: if the bucket name carries the S3 Express One Zone suffix (directory buckets are named like 'data--use1-az4--x-s3') but the effective endpoint is not an AWS endpoint (NetworkBinding.isAwsEndpoint), the tool exits 46 (EXIT_NOT_ACCEPTABLE) with 'The -zone option is only supported for Amazon S3 Express One Zone Storage'. Directory buckets only exist in real AWS S3, so combining an express-style name with a third-party S3-compatible endpoint is rejected up front.

Source

Thrown at hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/tools/BucketTool.java:221

    // may not exist
    String bucketPrefix = "fs.s3a.bucket." + bucket + '.';
    conf.setInt(bucketPrefix + S3A_BUCKET_PROBE, 0);
    conf.setBoolean(bucketPrefix + REJECT_OUT_OF_SPAN_OPERATIONS, false);
    // propagate an option
    BiFunction<String, Optional<String>, Boolean> propagate = (key, opt) ->
        opt.map(v -> {
          conf.set(key, v);
          LOG.info("{} = {}", key, v);
          return true;
        }).orElse(false);


    propagate.apply(AWS_REGION, region);
    propagate.apply(ENDPOINT, endpoint);

    // fail fast on third party store
    if (hasS3ExpressSuffix(bucket) && !isAwsEndpoint(endpoint.orElse(""))) {
      throw new ExitUtil.ExitException(EXIT_NOT_ACCEPTABLE, UNSUPPORTED_ZONE_ARG);
    }
    S3AFileSystem fs;
    try {
      fs = (S3AFileSystem) FileSystem.newInstance(fsURI, conf);
    } catch (FileNotFoundException e) {
      // this happens if somehow the probe wasn't disabled.
      errorln(PROBE_FAILURE);
      throw new ExitUtil.ExitException(EXIT_BAD_CONFIGURATION, PROBE_FAILURE);
    }

    try {

      // now build the configuration
      final CreateBucketConfiguration.Builder builder = CreateBucketConfiguration.builder();

      if (fs.hasPathCapability(new Path("/"), STORE_CAPABILITY_S3_EXPRESS_STORAGE)) {
        //  S3 Express store requires a zone and some other other settings
        final String az = zone.orElseThrow(() ->

View on GitHub (pinned to 2add963021)

Solutions

1. For third-party S3-compatible stores, use a normal bucket name without the '--x-s3' availability-zone suffix
2. For a real S3 Express One Zone bucket, drop the custom -endpoint or set it to an AWS endpoint (s3.amazonaws.com or a regional s3.<region>.amazonaws.com)
3. Check for leftover fs.s3a.endpoint or fs.s3a.bucket.<bucket>.endpoint overrides steering the client at a non-AWS URL

Example fix

# before (express-style name against MinIO)
hadoop aws s3guard bucket -create -endpoint http://minio.example:9000 s3a://data--use1-az4--x-s3
# after (plain bucket name for the third-party store)
hadoop aws s3guard bucket -create -endpoint http://minio.example:9000 s3a://data

String bucket = new Path(bucketArg).toUri().getHost();
String endpoint = effectiveEndpoint(conf, bucket); // -endpoint or fs.s3a.bucket.<b>.endpoint
if (hasS3ExpressSuffix(bucket) && !isAwsEndpoint(endpoint)) {
  failFast("express-suffixed bucket names require an AWS endpoint");
}

try {
  int rc = new BucketTool(conf).exec(args);
} catch (ExitUtil.ExitException e) {
  if (e.getExitCode() == 46) {
    // EXIT_NOT_ACCEPTABLE: fix bucket name or endpoint; retrying unchanged cannot help
  }
}

When it happens

Trigger: Bucket argument ending in '--x-s3' together with '-endpoint http://minio.example:9000' or a config override (fs.s3a.endpoint / fs.s3a.bucket.<bucket>.endpoint) pointing at MinIO, Ceph or another S3-compatible store; also an endpoint string malformed enough that isAwsEndpoint() does not recognize it as AWS.

Common situations: Testing S3 Express-style bucket naming against on-prem stores; endpoint typos or missing scheme/host in the endpoint URL; configs copied from third-party store documentation applied to express-named buckets.

Related errors

• Bucket does not exist. Accessing with fs.s3a.endpoint set t
• Bucket does not exist
• 42
• Wrong bucket: %s, in path: %s, expected bucket: %s
• GCS bucket name cannot be empty.

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/2a74a8ce322251f2. Report an issue: GitHub.