ingestr v1.1.13 - Iceberg Storage And Catalog Fixes

ingestr v1.1.13 was published on July 30, 2026. The main operator change is a tighter Iceberg storage and catalog path, including GCS and local storage selection, scheme aware bucket handling, and automatic S3 compatibility mode.

The full release notes and downloads are on the GitHub release page.

The release accepts storage=gcs when the Iceberg destination uses the bucket shorthand. It also adds explicit GCS and local storage handling with scheme aware bucket parsing. Those changes matter when a pipeline configuration supplies a compact bucket name in one environment and a scheme qualified location in another.

S3 compatible Iceberg deployments get a related adjustment. Compatibility mode is now enabled automatically where the connector identifies that requirement. This removes one source of configuration mismatch between standard S3 and an S3 compatible catalog deployment.

SQL catalog failures should also be easier to diagnose. The release notes call out a clearer SQL catalog error and accompanying documentation. They do not describe a new migration step or a breaking configuration change, so the useful difference here is earlier and more specific failure reporting rather than a new catalog model.

Finally, the project pins iceberg-go to a revision containing storage and catalog fixes. That pin makes this release dependent on a specific upstream state. Operators should treat the behavior as part of v1.1.13, not assume that an arbitrary iceberg-go update has the same fixes.

The highest risk fix outside Iceberg is the alignment of arrow-adbc with arrow-go 18.7. The stated reason is DuckDB read corruption. The notes do not identify affected types, query shapes, or data volumes, so there is no basis for narrowing the exposure further. Any pipeline that reads DuckDB through this path deserves an output comparison after the upgrade.

The release also pins arrow-adbc to the merged SQL driver fix. Both Arrow changes are dependency corrections rather than new user facing features. That distinction matters for rollout: connector syntax may remain unchanged while the bytes returned by a previously faulty read can change.

gRPC moves to v1.82.1 to clear GO-2026-6061. The release notes provide no severity or exploit conditions, so this should be recorded as a dependency remediation without adding assumptions about exposure. It is still a concrete reason to avoid carrying the older dependency in a long lived ingestion image.

This is a normal release, not a prerelease. Start validation with Iceberg jobs that use GCS bucket shorthand, local storage, SQL catalogs, or S3 compatible object storage. Confirm catalog creation and lookup, then compare object locations with the previous version. For DuckDB sources, compare row counts and checksums against a known result rather than relying only on successful process exit.

Keep the current pipeline configuration for the first pass. Change the ingestr version, capture connector logs, and compare emitted Iceberg paths before adjusting catalog options. This separates dependency effects from configuration edits during rollback analysis.

The narrow scope is useful. v1.1.13 concentrates on Iceberg address and catalog handling, Arrow reader correctness, and dependency maintenance. There is no announced CLI migration or schema migration. Existing production checks should therefore focus on destination resolution and data correctness, not on broad command changes.