apache/druid · error · UnsupportedOperationException

HistogramBufferAggregator does not support getLong()

Error message

HistogramBufferAggregator does not support getLong()

What it means

HistogramBufferAggregator.getLong() throws UnsupportedOperationException by design: the aggregator's result is a Histogram object stored in an off-heap buffer, and it cannot be converted to a single long. Druid's BufferAggregator interface offers getLong(), but the histogram implementation rejects it. Seeing this means a query path requested LONG access on a histogram metric.

Source

Thrown at processing/src/main/java/org/apache/druid/query/aggregation/HistogramBufferAggregator.java:99

    ByteBuffer mutationBuffer = buf.duplicate();
    mutationBuffer.position(position);
    mutationBuffer.asLongBuffer().get(bins);

    float min = mutationBuffer.getFloat(position + minOffset);
    float max = mutationBuffer.getFloat(position + maxOffset);
    return new Histogram(breaks, bins, min, max);
  }

  @Override
  public float getFloat(ByteBuffer buf, int position)
  {
    throw new UnsupportedOperationException("HistogramBufferAggregator does not support getFloat()");
  }

  @Override
  public long getLong(ByteBuffer buf, int position)
  {
    throw new UnsupportedOperationException("HistogramBufferAggregator does not support getLong()");
  }

  @Override
  public double getDouble(ByteBuffer buf, int position)
  {
    throw new UnsupportedOperationException("HistogramBufferAggregator does not support getDouble");
  }

  @Override
  public void close()
  {
    // no resources to cleanup
  }

  @Override
  public void inspectRuntimeShape(RuntimeShapeInspector inspector)
  {
    inspector.visit("selector", selector);

View on GitHub (pinned to 9b90983fd2)

Solutions

Use get(ByteBuffer, int) to obtain the Histogram and process it as an object result.
Choose a long-producing aggregator (longSum, longMax) if a BIGINT result is required.
Remove LONG-typed post-aggregators or casts from the histogram metric in the query spec.
Adjust custom code to branch on aggregator type before calling getLong().

Example fix

// before
long l = bufferAgg.getLong(buf, position); // throws
// after
Object val = bufferAgg.get(buf, position);
if (val instanceof Histogram) {
  int binCount = ((Histogram) val).bins.length;
}

Defensive patterns

Strategy: type-guard

Validate before calling

if (bufferAggregator instanceof HistogramBufferAggregator) {
  // use get(buf, position) (Histogram), not getLong()
}

Type guard

boolean supportsLong(BufferAggregator agg) {
  return !(agg instanceof HistogramBufferAggregator);
}

Try / catch

try {
  value = agg.getLong(buf, position);
} catch (UnsupportedOperationException e) {
  Histogram h = (Histogram) agg.get(buf, position);
}

Prevention

Avoid BIGINT coercion of histogram columns in SQL
Match the aggregator type to the expected output type in ingestion specs
Handle Histogram results as objects in custom extraction code
Review post-aggregator chains for long-typed ops over histogram metrics

When it happens

Trigger: A groupBy/timeseries query with a 'histogram' aggregator whose metric is read via getLong(ByteBuffer, int); LONG coercion of the histogram column in SQL; long-typed post-aggregators (arithmetic with long ops, longGreatest) applied to the histogram metric.

Common situations: Migrating queries from longSum to histogram without adjusting output handling; SQL planning that assigns BIGINT to the aggregation expression; custom aggregation pipelines that call getLong() generically on all buffer aggregators.

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/80b792e9f685331a. Report an issue: GitHub.