apache/iceberg · error · IllegalArgumentException

Unexpected object type for TIME logical type. Received: " +

Error message

Unexpected object type for TIME logical type. Received: " + object

What it means

convertToTime converts the Avro-decoded object for a TIME logical type into millis-of-day int. It supports java.time.LocalTime and Joda objects via JodaConverter; otherwise it throws IllegalArgumentException("Unexpected object type for TIME logical type. Received: " + object). It indicates the runtime decoded object doesn't match the TIME logical type.

Source

Thrown at flink/v2.2/flink/src/main/java/org/apache/iceberg/flink/formats/avro/AvroToRowDataConverters.java:284

      } else {
        throw new IllegalArgumentException(
            "Unexpected object type for DATE logical type. Received: " + object);
      }
    }
  }

  private static int convertToTime(Object object) {
    final int millis;
    if (object instanceof Integer) {
      millis = (Integer) object;
    } else if (object instanceof LocalTime) {
      millis = ((LocalTime) object).get(ChronoField.MILLI_OF_DAY);
    } else {
      JodaConverter jodaConverter = JodaConverter.getConverter();
      if (jodaConverter != null) {
        millis = jodaConverter.convertTime(object);
      } else {
        throw new IllegalArgumentException(
            "Unexpected object type for TIME logical type. Received: " + object);
      }
    }
    return millis;
  }

  private static byte[] convertToBytes(Object object) {
    if (object instanceof GenericFixed) {
      return ((GenericFixed) object).bytes();
    } else if (object instanceof ByteBuffer) {
      ByteBuffer byteBuffer = (ByteBuffer) object;
      byte[] bytes = new byte[byteBuffer.remaining()];
      byteBuffer.get(bytes);
      return bytes;
    } else {
      return (byte[]) object;
    }
  }

View on GitHub (pinned to 86d9c8fc54)

Solutions

Ensure the Avro schema declares the time logicalType so decoding yields LocalTime.
Add joda-time to the runtime classpath for legacy Joda-encoded times.
Align the Flink column type (e.g. INT/TIME) with what the data actually contains.

Example fix

// before
{ "name": "t", "type": "int" } // decoded as Integer
// after
{ "name": "t", "type": { "type": "int", "logicalType": "time-millis" } }

Defensive patterns

Strategy: type-guard

Validate before calling

if (!(avroObject instanceof LocalTime) && JodaConverter.getConverter() == null) { throw new IllegalStateException("TIME field decodes to " + avroObject.getClass() + " and Joda converter unavailable"); }

Type guard

boolean isSupportedTimeObject(Object o) { return o instanceof LocalTime || (isJodaTime(o) && JodaConverter.getConverter() != null); }

Try / catch

try { int millis = convertToTime(avroObject); } catch (IllegalArgumentException e) { /* inspect received type in message */ }

Prevention

Ensure Avro schemas declare time-millis/time-micros logicalType for time fields.
Include joda-time when consuming legacy-encoded times.
Round-trip test Avro serialize/deserialize for time columns.

When it happens

Trigger: An Avro 'time-millis'/'time-micros' field decodes to an unexpected object (e.g. plain int because the schema omitted the logical type), or Joda data with joda-time missing from the runtime.

Common situations: Schema evolution where the producer dropped the time logicalType; missing joda-time dependency; hand-built Avro GenericRecord values bypassing logical type conversion.

Related errors

AI-assisted analysis of apache/iceberg@86d9c8fc54 (2026-09-12). Data as JSON: /api/errors/bfb5fa90aa40b184. Report an issue: GitHub.