apache/druid · error · UnsupportedColumnTypeException

Cannot handle column [%s] with type [%s]

Error message

Cannot handle column [%s] with type [%s]

What it means

FieldWriters.create dispatches on the column type to build writers; supported scalar and scalar-array types return writers, but some branches (e.g. COMPLEX inside the array switch, and the outer default) throw UnsupportedColumnTypeException ('Cannot handle column [%s] with type [%s]'). Error [989] is the outer default: a top-level type the frame writer layer cannot write.

Source

Thrown at processing/src/main/java/org/apache/druid/frame/field/FieldWriters.java:98

      case STRING:
        return makeStringWriter(columnSelectorFactory, columnName, removeNullBytes);

      case COMPLEX:
        return makeComplexWriter(columnSelectorFactory, columnName, columnType.getComplexTypeName());

      case ARRAY:
        switch (columnType.getElementType().getType()) {
          case STRING:
            return makeStringArrayWriter(columnSelectorFactory, columnName, removeNullBytes);
          case LONG:
            return makeLongArrayWriter(columnSelectorFactory, columnName);
          case FLOAT:
            return makeFloatArrayWriter(columnSelectorFactory, columnName, frameType);
          case DOUBLE:
            return makeDoubleArrayWriter(columnSelectorFactory, columnName, frameType);
          case ARRAY:
          case COMPLEX:
            throw new UnsupportedColumnTypeException(columnName, columnType);
        }
      default:
        throw new UnsupportedColumnTypeException(columnName, columnType);
    }
  }

  private static FieldWriter makeLongWriter(
      final ColumnSelectorFactory selectorFactory,
      final String columnName
  )
  {
    final ColumnValueSelector<?> selector =
        TypeCastSelectors.makeColumnValueSelector(selectorFactory, columnName, ColumnType.LONG);
    return LongFieldWriter.forPrimitive(selector);
  }

  private static FieldWriter makeFloatWriter(
      final ColumnSelectorFactory selectorFactory,

View on GitHub (pinned to 9b90983fd2)

Solutions

  1. Read the reported type; if it is COMPLEX, avoid routing complex-metric columns through frame writing or convert them to a serializable representation.
  2. Cast/derive supported columns (LONG/FLOAT/DOUBLE/STRING/arrays of those) before the frame-writing stage.
  3. Align Druid versions so any newly supported types are available on all nodes.
  4. Extend FieldWriters/FieldReaders together if adding support for a new type.

Example fix

// before
SELECT sketch_col FROM ... -- complex column hits frame writer
// after
SELECT APPROX_QUANTILE_DS(sketch_col, 0.5) AS p50 FROM ...  -- aggregate to a supported scalar before shuffle
Defensive patterns

Strategy: validation

Validate before calling

if (columnType != null && columnType.getType() == ValueType.COMPLEX) { throw new IllegalStateException("complex column " + columnName + " not frame-writable; aggregate or convert first"); }

Try / catch

try { FieldWriters.create(factory, name, t, false); } catch (UnsupportedColumnTypeException e) { /* rewrite query to convert the column to a supported type */ }

Prevention

When it happens

Trigger: Calling FieldWriters.create with a ColumnType whose top-level type is not handled by the switch — for instance a COMPLEX column (hit in the array-switch COMPLEX arm) or any unknown/newer top-level type — when writing frames.

Common situations: MSQ work involving complex-metric columns that cannot be shuffled as frames; new column types introduced by upgrades but not yet supported by the frame writer; extension-provided types routed into generic frame writing.

Understand the failure class

Background: UnsupportedOperationException and "is not supported" errors: when a library deliberately refuses a call — this error's family across 30 libraries.

Related errors

AI-assisted analysis of apache/druid@9b90983fd2 (2026-09-07). Data as JSON: /api/errors/ed88489b51cc5e95. Report an issue: GitHub.