Functions and Settings Reference

Search Shortcut cmd + k | ctrl + k

Documentation / Core Extensions / Iceberg

This page is a reference for all functions and settings provided by the iceberg extension. For task-oriented documentation, see the Overview (reading), Writing to Iceberg, and Iceberg REST Catalogs pages.

Functions that take a table accept either a path to a table's metadata (e.g., 'data/iceberg/lineitem_iceberg') or, when a catalog is attached, a fully qualified table name (e.g., my_catalog.default.my_table).

Common Parameters

The read and metadata table functions below accept the following named parameters:

Parameter	Type	Default	Description
`allow_moved_paths`	`BOOLEAN`	`false`	Allows scanning Iceberg tables that are moved
`metadata_compression_codec`	`VARCHAR`	`''`	Set to `'gzip'` to read gzip-compressed metadata files
`snapshot_from_id`	`UBIGINT`	`NULL`	Access the snapshot with a specific `id`
`snapshot_from_timestamp`	`TIMESTAMP`	`NULL`	Access the snapshot as of a specific `timestamp`
`version`	`VARCHAR`	`'?'`	Explicit version string, hint file, or `'?'` for guessing
`version_name_format`	`VARCHAR`	`'v%s%s.metadata.json,%s%s.metadata.json'`	Controls how versions are converted to metadata file names

See Selecting Metadata Versions for details.

Read and Metadata Functions

Function	Description
`iceberg_scan(table, ⟨options⟩)`	Reads the data of an Iceberg table. Returns the table's columns.
`iceberg_metadata(table, ⟨options⟩)`	Returns one row per manifest entry. Columns: `manifest_path`, `manifest_sequence_number`, `manifest_content`, `status`, `content`, `file_path`, `file_format`, `record_count`.
`iceberg_snapshots(table, ⟨options⟩)`	Returns one row per snapshot. Columns: `sequence_number`, `snapshot_id`, `timestamp_ms`, `manifest_list`.
`iceberg_column_stats(table, ⟨options⟩)`	Returns per-data-file, per-column statistics. Columns include `file_path`, `column_name`, `column_type`, `lower_bound`, `upper_bound`, `value_count`, `null_value_count`, `nan_value_count`.
`iceberg_partition_stats(table, ⟨options⟩)`	Returns per-partition-field statistics. Columns include `partition_field_name`, `partition_source_columns`, `partition_field_transform`, `lower_bound`, `upper_bound`.
`iceberg_load_table_response(table)`	Advanced, REST-catalog only. Returns the raw catalog `LoadTable` response: `metadata_location`, `metadata` (`VARIANT`), `config` (`MAP`), `storage_credentials`, `request_url`.

SELECT * FROM iceberg_snapshots('data/iceberg/lineitem_iceberg');
SELECT * FROM iceberg_column_stats('my_catalog.default.events');

Table and Schema Property Functions

These functions read and modify Iceberg table properties and Iceberg schema (namespace) properties. They require an attached catalog. See Writing to Iceberg.

Function	Description
`iceberg_table_properties(table)`	Returns all properties of the table.
`set_iceberg_table_properties(table, properties)`	Sets properties on the table. `properties` is a `MAP(VARCHAR, VARCHAR)`.
`remove_iceberg_table_properties(table, property_list)`	Removes the listed properties (`VARCHAR[]`) from the table.
`iceberg_schema_properties(schema)`	Returns all properties of the schema (namespace).
`set_iceberg_schema_properties(schema, properties)`	Sets properties on the schema.
`remove_iceberg_schema_properties(schema, property_list)`	Removes the listed properties from the schema.

DuckLake Interoperability

Function	Description
`iceberg_to_ducklake(iceberg_catalog, ducklake_catalog, skip_tables := [...])`	Performs a metadata-only copy of an attached Iceberg catalog into a DuckLake catalog. The optional `skip_tables` parameter (`VARCHAR[]`) excludes tables.

See Interoperability with DuckLake.

Partition Transform Functions

These scalar functions implement the Iceberg partition transforms. They are most often used in a PARTITIONED BY clause (see Partitioning), but can also be called directly.

Function	Description
`iceberg_bucket(num_buckets, value)`	Returns the Iceberg bucket partition value (an `INTEGER`) for `value`. Supported `value` types: `INTEGER`, `BIGINT`, `DECIMAL`, `DATE`, `TIME`, `TIMESTAMP`, `TIMESTAMP WITH TIME ZONE`, `TIMESTAMP_NS`, `VARCHAR`, `BLOB`, `UUID`.
`iceberg_truncate(width, value)`	Returns the Iceberg truncate partition value for `value`, with the same type as `value`. Supported types: `INTEGER`, `BIGINT`, `DECIMAL`, `VARCHAR`, `BLOB`.

SELECT iceberg_bucket(16, 'duckdb');
SELECT iceberg_truncate(10, 1234);

Settings

Setting	Type	Default	Description
`unsafe_enable_version_guessing`	`BOOLEAN`	`false`	Enable globbing the filesystem (if possible) to find the latest metadata version. This may read an uncommitted version, so it is disabled by default.
`iceberg_use_metadata_log`	`BOOLEAN`	`true`	Use a table's optional `metadata-log` to preserve atomicity guarantees, at the cost of an additional metadata `GET` in rare cases.
`ignore_target_file_size_for_partitioned_tables`	`BOOLEAN`	`false`	Ignore the unsupported `write.target-file-size-bytes` table property on partitioned tables instead of raising an error.
`ignore_row_group_size_for_partitioned_tables`	`BOOLEAN`	`false`	Ignore the unsupported `write.parquet.row-group-size-bytes` table property on partitioned tables instead of raising an error.
`iceberg_via_aws_sdk_for_catalog_interactions`	`BOOLEAN`	`false`	Use the legacy AWS SDK code path to interact with AWS-based catalogs instead of DuckDB's HTTP client.

About this page

DuckDB Home Code of Conduct Trademark Use Blog