apache/hadoop · error · IllegalArgumentException

Cannot serialize field {} into JSON

Error message

Cannot serialize field {} into JSON

What it means

RBFMetrics.getJson() serializes a BaseRecord (e.g. RouterState, MountTable, MembershipState) into a JSONObject for JMX output by reflectively reading every declared field. If reading any field via getField(record, fieldName) throws (inaccessible field, incompatible type in the record, or a nested value that fails conversion), it rethrows IllegalArgumentException('Cannot serialize field <name> into JSON'). The original exception is deliberately swallowed, so the field name in the message is your only clue.

Source

Thrown at hadoop-hdfs-project/hadoop-hdfs-rbf/src/main/java/org/apache/hadoop/hdfs/server/federation/metrics/RBFMetrics.java:1004

   *
   * @return Map representing the data for the JSON representation.
   */
  private static Map<String, Object> getJson(BaseRecord record) {
    Map<String, Object> json = new HashMap<>();
    Map<String, Class<?>> fields = getFields(record);

    for (String fieldName : fields.keySet()) {
      if (!fieldName.equalsIgnoreCase("proto")) {
        try {
          Object value = getField(record, fieldName);
          if (value instanceof BaseRecord) {
            BaseRecord recordField = (BaseRecord) value;
            json.putAll(getJson(recordField));
          } else {
            json.put(fieldName, value == null ? JSONObject.NULL : value);
          }
        } catch (Exception e) {
          throw new IllegalArgumentException(
              "Cannot serialize field " + fieldName + " into JSON");
        }
      }
    }
    return json;
  }

  /**
   * Returns all serializable fields in the object.
   *
   * @return Map with the fields.
   */
  private static Map<String, Class<?>> getFields(BaseRecord record) {
    Map<String, Class<?>> getters = new HashMap<>();
    for (Method m : record.getClass().getDeclaredMethods()) {
      if (m.getName().startsWith("get")) {
        try {
          Class<?> type = m.getReturnType();

View on GitHub (pinned to 2add963021)

Solutions

1. Identify the failing field from the message and inspect that field's type in the corresponding BaseRecord class
2. If the field is a custom type, implement/verify its serialization support (make it a BaseRecord, String-primitive, or add a getter the serializer uses)
3. Check for state-store schema/version drift after upgrade: refresh records (e.g. re-register Router/NN heartbeats) so new-format records replace old ones
4. As a maintainer, chain the cause: throw new IllegalArgumentException(msg, e) to expose the root reason

Example fix

// before
} catch (Exception e) {
  throw new IllegalArgumentException(
      "Cannot serialize field " + fieldName + " into JSON");
}

// after (preserves root cause)
} catch (Exception e) {
  throw new IllegalArgumentException(
      "Cannot serialize field " + fieldName + " into JSON", e);
}

try {
  String json = routerMetrics.getNamenodeRegistrations(); // any record-dumping JMX attr
} catch (IllegalArgumentException e) {
  if (e.getMessage() != null && e.getMessage().startsWith("Cannot serialize field")) {
    // extract field name, inspect that BaseRecord field's type; likely schema drift or custom type
    LOG.warn("JMX record serialization failed: {}", e.getMessage());
  } else { throw e; }
}

When it happens

Trigger: JMX queries on RBFMetrics that dump state-store records, e.g. getRouters(), getMountTable(), getNameservices(), getNamenodeRegistrations() returning records whose fields include enums/Date/nested BaseRecords the reflective reader cannot handle; a custom BaseRecord subclass registered in the state store whose field types break getField(); accessing a field on a partially-updated record.

Common situations: Scraping RBF JMX beans with monitoring after adding a custom record type or after a Hadoop version change altered record schemas; records loaded from an older state-store driver (serialization version mismatch); reflective access blocked under stricter JRE module rules.

Related errors

• Federated metrics is not initialized
• Cannot find namenode resolver.
• Cannot find subcluster resolver
• Namenode metrics is not initialized
• Mount table state store is not available.

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/6d31c0af01e411bb. Report an issue: GitHub.