apache/druid · error · IllegalStateException

Unsupported nested array type

Error message

Unsupported nested array type: [%s]

What it means

FieldTypeInfo.add maps a nested value type to its array-type bitmask; only STRING/LONG/DOUBLE (and their ARRAY forms) are supported. Encountering an array of an unsupported type (e.g. complex/array-of-arrays beyond supported set) in the default branch throws ISE 'Unsupported nested array type'.

Solutions

1. Flatten or remove nested-of-nested arrays in the input data before ingestion
2. Cast array elements to supported scalar types (string/long/double) during ingestion
3. Upgrade to a Druid version supporting the desired nested array type
4. Guard the ingestion spec with transforms/filtered aggregations that skip unsupported array types

Example fix

// before
// input: {"tags": [["a","b"], ["c"]]} // array of arrays
// after
// transform to flat array
"transforms": [{"type": "expression", "name": "tags", "expression": "array_concat(tag0, tag1)"}]

for (Object v : list) { ValueType t = ValueType.fromRawType(v); if (!(t == ValueType.STRING || t == ValueType.LONG || t == ValueType.DOUBLE)) throw new IllegalArgumentException("Unsupported array element type " + t); }

boolean supportedArrayType(Object v) { return v instanceof List && !((List<?>) v).isEmpty() && (v.toString().startsWith("[") ) && !firstElementIsComplex((List<?>) v); }

try { types = fieldTypeInfo.add(type, types); } catch (ISE e) { log.warn("Skipping unsupported nested array type: %s", e.getMessage()); return nullTypeMask(); }

When it happens

Trigger: Collecting field type info while ingesting nested data when an array element's ValueType falls through the switch's default case — e.g. nested arrays, arrays of complex objects, or future/unknown types.

Common situations: Ingesting JSON with arrays containing objects/arrays into a nested column on a Druid version without support for that array type; type inference producing unexpected ValueType from user data.

Related errors

• Cannot make array element selector for path
• Multi-value row not supported
• Aggregator[ ] cannot vectorize
• AppenderatorsManager methods should only called by services…
• ApproximateHistogramAggregator does not support getFloat()

AI-assisted analysis of apache/druid@9b90983fd2 (2026-09-07). Data as JSON: /api/errors/dc99e610df9145d7. Report an issue: GitHub.

Appendix: source

Thrown at processing/src/main/java/org/apache/druid/segment/nested/FieldTypeInfo.java:316

        types |= LONG_MASK;
        break;
      case DOUBLE:
        types |= DOUBLE_MASK;
        break;
      case ARRAY:
        Preconditions.checkNotNull(type.getElementType(), "ElementType must not be null");
        switch (type.getElementType().getType()) {
          case STRING:
            types |= STRING_ARRAY_MASK;
            break;
          case LONG:
            types |= LONG_ARRAY_MASK;
            break;
          case DOUBLE:
            types |= DOUBLE_ARRAY_MASK;
            break;
          default:
            throw new ISE("Unsupported nested array type: [%s]", type.asTypeString());
        }
        break;
      default:
        throw new ISE("Unsupported nested type: [%s]", type.asTypeString());
    }
    return types;
  }

  public static Set<ColumnType> convertToSet(byte types)
  {
    final Set<ColumnType> theTypes = Sets.newHashSetWithExpectedSize(4);
    if ((types & STRING_MASK) > 0) {
      theTypes.add(ColumnType.STRING);
    }
    if ((types & LONG_MASK) > 0) {
      theTypes.add(ColumnType.LONG);
    }
    if ((types & DOUBLE_MASK) > 0) {

View on GitHub (pinned to 9b90983fd2)