apache/iceberg · error · java.lang.UnsupportedOperationException

Unsupported timestamp type:

Error message

Unsupported timestamp type: 

What it means

LogicalTypeWriterBuilder.visit(TimestampLogicalTypeAnnotation) supports writing timestamps as NANOS or MICROS units. Any other TimestampUnit (e.g. MILLIS, or future units) falls to the default branch and throws UnsupportedOperationException naming the timestamp annotation.

Solutions

1. Change the column annotation to TIMESTAMP(MICROS) or TIMESTAMP(NANOS) in the write schema.
2. Convert the data at the source (e.g. scale millis to micros) and re-write with a supported unit.
3. Add a MILLIS case (writing via the existing micros/nanos writer after conversion) in visit(...) if support is needed.

Example fix

// before
Types.TimestampType.withZone() -> Parquet TIMESTAMP_MILLIS
// after
use TIMESTAMP_MICROS annotation (TimestampLogicalTypeAnnotation(true, MICROS))

// java
if (annotation instanceof TimestampLogicalTypeAnnotation) {
  TimestampLogicalTypeAnnotation ts = (TimestampLogicalTypeAnnotation) annotation;
  if (ts.getUnit() != TimeUnit.MICROS && ts.getUnit() != TimeUnit.NANOS) {
    throw new IllegalArgumentException("Timestamp unit must be MICROS or NANOS, got " + ts.getUnit());
  }
}

// java
try {
  writer = FlinkParquetWriters.createWriter(...);
} catch (UnsupportedOperationException e) {
  // convert column annotation to MICROS/NANOS and retry
}

When it happens

Trigger: Creating a Flink Parquet writer for a column annotated with TimestampLogicalTypeAnnotation using a unit other than MICROS or NANOS — most commonly isAdjustedToUTC/NANOS variants aside, a MILLIS-annotated timestamp column.

Common situations: Schemas copied from engines that emit MILLIS timestamp annotations (e.g. some Spark/Hive writers); hand-built Parquet schemas with TIMESTAMP_MILLIS; format conversion that preserved the source unit.

Related errors

• Unsupported base type for decimal:
• Unsupported logical type:
• Unsupported logical type
• Unsupported timestamp type:
• Unsupported timestamp type

AI-assisted analysis of apache/iceberg@86d9c8fc54 (2026-09-12). Data as JSON: /api/errors/ee2389c613950af7. Report an issue: GitHub.

Appendix: source

Thrown at flink/v2.3/flink/src/main/java/org/apache/iceberg/flink/data/FlinkParquetWriters.java:285

      Preconditions.checkArgument(
          LogicalTypeAnnotation.TimeUnit.MICROS.equals(times.getUnit()),
          "Cannot write time in %s, only MICROS is supported",
          times.getUnit());
      return Optional.of(timeMicros(desc));
    }

    @Override
    public Optional<ParquetValueWriter<?>> visit(TimestampLogicalTypeAnnotation timestamps) {
      ParquetValueWriter<TimestampData> writer;
      switch (timestamps.getUnit()) {
        case NANOS:
          writer = timestampNanos(desc);
          break;
        case MICROS:
          writer = timestamps(desc);
          break;
        default:
          throw new UnsupportedOperationException("Unsupported timestamp type: " + timestamps);
      }

      return Optional.of(writer);
    }

    @Override
    public Optional<ParquetValueWriter<?>> visit(IntLogicalTypeAnnotation type) {
      Preconditions.checkArgument(type.isSigned(), "Cannot write unsigned integer type: %s", type);
      ParquetValueWriter<?> writer;
      if (type.getBitWidth() < 64) {
        writer = ints(flinkType, desc);
      } else {
        writer = ParquetValueWriters.longs(desc);
      }

      return Optional.of(writer);
    }

View on GitHub (pinned to 86d9c8fc54)