apache/hadoop · error · IllegalArgumentException

output name cannot be 'part'

Error message

output name cannot be 'part'

What it means

Thrown as IllegalArgumentException from MultipleOutputs.checkBaseOutputPath (MultipleOutputs.java:250) when the string equals FileOutputFormat.PART ('part'). 'part' is the reserved prefix of the default output files (part-r-00000 etc.), so named outputs and base output paths may not claim it verbatim; otherwise output would collide with the framework's default channel.

Source

Thrown at hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/lib/output/MultipleOutputs.java:250

      }
      if ((ch >= '0') && (ch <= '9')) {
        continue;
      }
      throw new IllegalArgumentException(
        "Name cannot be have a '" + ch + "' char");
    }
  }

  /**
   * Checks if output name is valid.
   *
   * name cannot be the name used for the default output
   * @param outputPath base output Name
   * @throws IllegalArgumentException if the output name is not valid.
   */
  private static void checkBaseOutputPath(String outputPath) {
    if (outputPath.equals(FileOutputFormat.PART)) {
      throw new IllegalArgumentException("output name cannot be 'part'");
    }
  }
  
  /**
   * Checks if a named output name is valid.
   *
   * @param namedOutput named output Name
   * @throws IllegalArgumentException if the output name is not valid.
   */
  private static void checkNamedOutputName(JobContext job,
      String namedOutput, boolean alreadyDefined) {
    checkTokenName(namedOutput);
    checkBaseOutputPath(namedOutput);
    List<String> definedChannels = getNamedOutputsList(job);
    if (alreadyDefined && definedChannels.contains(namedOutput)) {
      throw new IllegalArgumentException("Named output '" + namedOutput +
        "' already alreadyDefined");
    } else if (!alreadyDefined && !definedChannels.contains(namedOutput)) {

View on GitHub (pinned to 2add963021)

Solutions

1. Rename the base path to anything except exactly 'part': 'mypart', 'part-agg', 'data'
2. If you wanted the default channel, use context.write(key, value) instead of MultipleOutputs

Example fix

// before
mos.write("summary", key, value, "part");

// after
mos.write("summary", key, value, "summary-part");

// guard the reserved default-output name
static String requireBasePath(String p) {
  if (FileOutputFormat.PART.equals(p)) throw new IllegalArgumentException("baseOutputPath 'part' is reserved");
  return p;
}

When it happens

Trigger: mos.write("channel", key, value, "part") with baseOutputPath exactly 'part'; or addNamedOutput(job, "part", ...) — checkNamedOutputName applies checkBaseOutputPath to the channel name too. Substrings like 'partA' or 'part-r-5' are fine.

Common situations: Generic write helpers that pass a fixed base name 'part' assuming it works like the default output; migrating code from direct context.write() where 'part' is the implicit name.

Related errors

• Name cannot be NULL or emtpy
• Name cannot be have a '{ch}' char
• Name cannot be NULL or emtpy
• Name cannot be have a '{}' char
• Named output name cannot be 'part'

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/825a4fc9c99a36a7. Report an issue: GitHub.