apache/iceberg · error · UnsupportedOperationException

Unsupported type - map

Error message

Unsupported type - map

What it means

IcebergArrowColumnVector supports primitive and array (list) accessors but not map accessors, so getMap() unconditionally throws UnsupportedOperationException. Map-typed columns cannot be read through this Arrow-backed vector.

Source

Thrown at spark/v4.0/spark/src/main/java/org/apache/iceberg/spark/data/vectorized/IcebergArrowColumnVector.java:128

    return accessor.getFloat(rowId);
  }

  @Override
  public double getDouble(int rowId) {
    return accessor.getDouble(rowId);
  }

  @Override
  public ColumnarArray getArray(int rowId) {
    if (isNullAt(rowId)) {
      return null;
    }
    return accessor.getArray(rowId);
  }

  @Override
  public ColumnarMap getMap(int rowId) {
    throw new UnsupportedOperationException("Unsupported type - map");
  }

  @Override
  public Decimal getDecimal(int rowId, int precision, int scale) {
    if (isNullAt(rowId)) {
      return null;
    }
    return accessor.getDecimal(rowId, precision, scale);
  }

  @Override
  public UTF8String getUTF8String(int rowId) {
    if (isNullAt(rowId)) {
      return null;
    }
    return accessor.getUTF8String(rowId);
  }

View on GitHub (pinned to 86d9c8fc54)

Solutions

  1. Exclude the map column from the vectorized read (select only needed non-map columns)
  2. Disable vectorization: set table property read.split.vectorization.enabled=false for the read
  3. Use a non-vectorized/row-based reader (e.g. file-format-level parquet reader path)
  4. Upgrade Iceberg — map support in vectorized reads is extended in newer versions

Example fix

// before
spark.read.format("iceberg").load("t").select("m") // map column + vectorized
// after
spark.conf.set("read.split.vectorization.enabled", "false")
spark.read.format("iceberg").load("t").select("m")
Defensive patterns

Strategy: validation

Validate before calling

boolean hasMap = table.schema().columns().stream().anyMatch(c -> c.type().typeId() == Types.MapType.class.cast(c.type()).typeId()); if (hasMap) { spark.conf.set("read.split.vectorization.enabled", "false"); }

Try / catch

try { map = vector.getMap(rowId); } catch (UnsupportedOperationException e) { // retry read without vectorization }
spark.conf.set("read.split.vectorization.enabled", "false"); reRunQuery();

Prevention

When it happens

Trigger: A vectorized batch scan selects a map-typed column, and Spark's columnar reader calls getMap(rowId) on the IcebergArrowColumnVector wrapping the map column.

Common situations: Querying tables containing Iceberg map columns with vectorized reads enabled (read.split.vectorization.enabled=true or Spark vectorized reader defaults for the format).

Related errors

AI-assisted analysis of apache/iceberg@86d9c8fc54 (2026-09-12). Data as JSON: /api/errors/a050afa617cd49ac. Report an issue: GitHub.