Decaf Blog Data pipelines, open-source tooling, reverse engineering, and systems notes

ingestr v1.0.51 - Vitess CDC and Oracle Destination

3 min read

ingestr v1.0.51 was published on June 29, 2026, with Vitess CDC via VStream as the main pipeline change. The release also adds Oracle destination support, Square and Recurly source work, and a dependency fix for go-chi/chi/v5. It is not marked as a prerelease.

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

Vitess CDC through VStream

The most important data movement item in this release is CDC support for Vitess via VStream. The notes also list Vitess and PlanetScale CDC documentation, plus fixes for keyspace qualified table names and copy phase streaming from explicit shards.

That combination matters for real Vitess pipelines. Table identity and shard selection are not cosmetic details in a backfill plus change stream workflow. If either is wrong, a pipeline can copy the wrong slice, miss rows, or fail once names are qualified by keyspace.

The test work also matters. The changelog says vtcombo gRPC is bound in tests. That reads like internal plumbing, but it is the kind of plumbing that keeps CDC behavior testable instead of only exercised against a live service. For teams using Vitess as an operational database layer, this is the part of v1.0.51 to inspect first on the GitHub release page.

New Oracle, Square, and Recurly coverage

The destination side gets Oracle support. For ELT operators, that widens the set of places where ingestr can land extracted data without a separate loading step. The release notes do not include detailed Oracle type mapping notes, so the practical read is destination availability, not a promise about every edge case in schema conversion.

The source side adds Square and Recurly work. Square gets more detail in the changelog than Recurly. The notes call out a new Square source, catalog tombstone visibility, a composite cash_drawers primary key, and location_id materialization on cash_drawer_shifts rows.

Those Square fixes are not just connector polish. Tombstones are how downstream systems can see objects that have been removed or retired. Primary key shape decides whether a load can merge cleanly. A materialized location_id on cash drawer shift rows gives consumers a direct join key for location scoped reporting instead of making them recover that context later.

Recurly is listed as a source in this release. The notes do not describe table coverage or incremental behavior, so it should be treated as connector availability until the project publishes more schema detail.

Security, tests, and audit feedback

ingestr also bumps go-chi/chi/v5 to v5.3.0 for IP spoofing advisories. For platform operators, IP address handling can feed audit logs, rate limits, allow lists, and proxy trust decisions. A router dependency fix is therefore relevant even when the headline of the release is data movement.

There is also license audit maintenance. The changelog mentions refreshed audit locks for the Vitess dependency tree and the Chi dependency tree, plus remediation guidance in the license audit failure message. This is not a user visible pipeline feature, but it changes the failure mode for builds and policy checks. A better audit message can reduce the time between a blocked build and a clear dependency action.

Some of the remaining entries are internal refactor or maintenance work. That is fine. In this release, the user facing shape is still clear: Vitess CDC, Oracle as a destination, new source coverage, and a small security pass around HTTP request handling.

Where to get it