position-delete file

Error message

position-delete file {} has unsupported format {:?}; expected Puffin or Parquet

What it means

When reading Iceberg position-delete files to build delete vectors, only Puffin (deletion vector) and Parquet formats are supported. If the delete file's `DataFileFormat` is anything else (e.g. Avat/Orc-style formats), the reader bails because it cannot interpret the file's delete records.

Solutions

Rewrite the source table's delete files to Puffin/Parquet (e.g. run a compaction/rewrite with a Spark/Flink job) so RisingWave can read them.
Re-create or rewrite the table avoiding non-Puffin/Parquet position deletes (use copy-on-write or DVs).
Check which engine/format produced the delete files and whether upgrading RisingWave adds support for that format.
File/verify support for the specific `DataFileFormat` you hit, since the match already covers Puffin and Parquet.

Example fix

// before: engine wrote ORC position-deletes, unreadable here
// after: compact/rewrite deletes to Puffin or Parquet, e.g. Spark:
//   CALL catalog.system.rewrite_deletes(table => 'db.tbl', options => map('format','parquet'))

Defensive patterns

Strategy: validation

Validate before calling

fn check_delete_file_format(fmt: &str) -> Result<(), String> {
    match fmt {
        "puffin" | "parquet" => Ok(()),
        other => Err(format!("unsupported position-delete format {other}; rewrite deletes as puffin/parquet")),
    }
}

Prevention

Inspect table metadata for delete file formats before pointing a RisingWave iceberg sink at an external table.
Configure upstream writers to use Puffin DVs or Parquet position deletes.
Run periodic compaction on the source table to rewrite legacy delete formats.

When it happens

Trigger: Encountering a position-delete entry in the Iceberg table metadata whose format is neither `puffin` nor `parquet` — typically from tables written by another engine with different delete-file formats — while the sink collects input DVs or flushes.

Common situations: Consuming an Iceberg table produced by third-party writers (Spark/Flink with ORC deletes); tables where a writer stored position deletes in an unsupported format; stale/mismatched metadata after a writer upgrade.

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 risingwavelabs/risingwave@6469eb736d (2026-09-11). Data as JSON: /api/errors/604e117e688995c0. Report an issue: GitHub.

Appendix: source

Thrown at src/connector/src/sink/iceberg/position_delete.rs:142

            })?;
            delete_vector.insert(pos as u64);
        }
    }

    Ok(delete_vector)
}

/// Reads the deleted positions of a single position-delete `DataFile` regardless of on-disk format,
pub async fn read_position_deletes_from_file(
    file_io: &FileIO,
    delete_file: &DataFile,
) -> Result<DeleteVector> {
    match delete_file.file_format() {
        DataFileFormat::Puffin => read_dv_positions_from_data_file(file_io, delete_file).await,
        DataFileFormat::Parquet => {
            read_parquet_position_deletes_from_file(file_io, delete_file).await
        }
        other => bail!(
            "position-delete file {} has unsupported format {:?}; expected Puffin or Parquet",
            delete_file.file_path(),
            other
        ),
    }
}

/// Writes `delete_vector` as a single Puffin deletion-vector blob referencing `data_file_path`,
/// and returns its [`DataFile`] metadata (content `PositionDeletes`, format `Puffin`) with
/// `referenced_data_file` set.
pub async fn write_dv_puffin_file(
    table: &Table,
    location_generator: &DefaultLocationGenerator,
    file_name_generator: &DefaultFileNameGenerator,
    data_file_path: String,
    delete_vector: &DeleteVector,
    partition_key: Option<&PartitionKey>,
) -> Result<DataFile> {

View on GitHub (pinned to 6469eb736d)