ingestr v1.0.51 - Oracle Destination And Vitess CDC

bruin-data/ingestr v1.0.51 was published on June 29, 2026, and is not marked as a prerelease. The main operational change is Oracle destination support, with Vitess CDC via VStream added beside new Square and Recurly sources for teams moving application and database data into analytical stores.

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

The release notes lead with Oracle destination support. That matters because ingestr is usually evaluated by the shape of its supported source and destination graph, not by a single connector in isolation. Adding Oracle on the destination side gives teams another target for jobs that already pass through ingestr.

The notes do not describe Oracle specific limits, data type mappings, batch size behavior, or transaction semantics. Treat this as a connector addition that needs a small validation run before it enters scheduled production work. A useful first pass is a narrow sync that covers nullable columns, numeric precision, timestamp fields, and a table with enough rows to expose write path behavior.

There is no breaking change called out for the destination layer. That is useful, but it should not replace local testing. Destination connectors are where quiet schema and type assumptions usually show up.

Vitess CDC via VStream is the other database side change in v1.0.51. The changelog also says the project documented Vitess and PlanetScale CDC through VStream, which is the right place to look before wiring this into a real replication path.

The fixes around the feature are more interesting than the headline. ingestr now matches keyspace qualified table names, streams the copy phase from explicit shards, and binds vtcombo gRPC in tests. Those are operator visible details. They touch table selection, shard handling, and test coverage around the local Vitess path.

For platform teams, the practical question is how this behaves during initial copy and then CDC follow through. The release notes do not claim a new migration contract, so the sane rollout is one keyspace, one known table set, and a comparison pass against the source before widening scope.

v1.0.51 adds Square as a source and also includes a Recurly source entry. The Square work has follow up fixes in the same release, which makes it worth reading as more than a connector checkbox.

Square catalog tombstones are now always surfaced. That is important for pipelines that need delete or retire semantics to remain visible downstream. The release also changes cash_drawers to use a composite primary key and materializes location_id on cash_drawer_shifts rows. Those fields are not cosmetic. They affect merge keys, downstream joins, and how analysts separate activity across locations.

The Recurly note is sparse. There is no stream list or schema detail in the release notes, so teams should inspect the connector behavior directly before assuming coverage for every billing object they care about.

The security related maintenance is small but worth noting. go-chi/chi/v5 moves to v5.3.0 for IP spoofing advisories, and the license audit lock is refreshed for both the Vitess dependency tree and the go-chi/chi/v5 update.

There is also a license audit usability change. The audit failure message now includes remediation guidance, and the docs add an AGENTS.md license audit note. That is internal plumbing, but it helps in CI when a dependency update fails on policy instead of tests.

This release is not just new connector surface. It also tightens some of the operational feedback loops around dependencies, advisories, and audit failures.