apache/druid · warning · UnsupportedOperationException

Vectorized groupBys on ARRAY columns are not yet implemented

Error message

Vectorized groupBys on ARRAY columns are not yet implemented

What it means

GroupByVectorColumnProcessorFactory.makeArrayProcessor unconditionally throws UnsupportedOperationException: grouping on ARRAY-typed columns is not supported by the vectorized group-by engine. Any vectorized query grouping by an ARRAY column reaches this factory method and fails.

Source

Thrown at processing/src/main/java/org/apache/druid/query/groupby/epinephelinae/vector/GroupByVectorColumnProcessorFactory.java:112

    return new NullableDoubleGroupByVectorColumnSelector(selector);
  }

  @Override
  public GroupByVectorColumnSelector makeLongProcessor(
      final ColumnCapabilities capabilities,
      final VectorValueSelector selector
  )
  {
    if (capabilities.hasNulls().isFalse()) {
      return new LongGroupByVectorColumnSelector(selector);
    }
    return new NullableLongGroupByVectorColumnSelector(selector);
  }

  @Override
  public GroupByVectorColumnSelector makeArrayProcessor(ColumnCapabilities capabilities, VectorObjectSelector selector)
  {
    throw new UnsupportedOperationException(
        "Vectorized groupBys on ARRAY columns are not yet implemented"
    );
  }

  @Override
  public GroupByVectorColumnSelector makeObjectProcessor(
      final ColumnCapabilities capabilities,
      final VectorObjectSelector selector
  )
  {
    if (capabilities.is(ValueType.STRING)) {
      if (capabilities.hasMultipleValues().isTrue()) {
        throw new UnsupportedOperationException(
            "Vectorized groupBys on multi-value dictionary-encoded dimensions are not yet implemented"
        );
      }
      return new DictionaryBuildingSingleValueStringGroupByVectorColumnSelector(selector);
    } else if (capabilities.is(ValueType.COMPLEX)) {

View on GitHub (pinned to 9b90983fd2)

Solutions

1. Disable vectorization for the query (context {"vectorize":"false"}).
2. Convert the array to a scalar before grouping, e.g. via array_to_string or element access in an expression virtual column.
3. Normalize arrays into separate rows/columns at ingestion if grouping per element is the goal.

Example fix

// before: GROUP BY arrayCol (vectorized)
// after: query context
{"vectorize":"false"}
// or group by expression: array_to_string(arrayCol, ',')

ColumnCapabilities caps = pool.getColumnCapabilities(column);
if (caps != null && caps.is(ValueType.ARRAY)) { /* route to non-vectorized engine */ }

boolean isArrayColumn(ColumnCapabilities caps) {
  return caps != null && caps.is(ValueType.ARRAY);
}

try {
  runVectorized(query);
} catch (UnsupportedOperationException e) {
  return runWithVectorizeDisabled(query);
}

When it happens

Trigger: A vectorized groupBy query with a grouping key whose column capabilities report ValueType.ARRAY, routed to the vector column processor factory's makeArrayProcessor.

Common situations: Druid 28+ ARRAY types used in groupBy dimensions; users experimenting with new ARRAY columns in aggregations; auto-vectorization picking up an array-typed virtual column.

Related errors

• Cardinality aggregator does not support[%s] inputs
• Vectorized matcher cannot make string matcher for ARRAY type
• Vectorized matcher cannot make object matcher for ARRAY type
• Vectorized groupBys on multi-value dictionary-encoded dimens
• cannot vectorize fixed bucket histogram aggregation for type

AI-assisted analysis of apache/druid@9b90983fd2 (2026-09-07). Data as JSON: /api/errors/c01574974d5a4454. Report an issue: GitHub.