New: Cookbooks and AI ExplanationsStep-by-Step recipes to solve problems connected to Roadmaps and Cheat Sheets. Need more details? Use AI buttons for structured and simple explanations with concrete examples throughout the whole platform.Take a look
A developer on another team adds one column to a table, and three of your downstream jobs go down within the hour.
What you'll have at the end
A registry subject locked to backward-compatible mode, tested against both a safe column addition and a genuinely breaking one, with the connector recovered and every consumer confirmed caught up afterward.
You need
A live CDC pipeline already reading a source table through a connector, publishing schema-carrying change events to a topic backed by a schema registry, with more than one downstream job consuming it.
Not covered
Coordinating a rename or a drop that must genuinely reach every consumer's own code: that needs a planned migration window on its own, and this recipe's guardrail is built to block that kind of change here until such planning happens.
Leans on
Catch a replication slot before it fills your Postgres disk
when the failed task in this recipe stays down long enough that the source's own replication backlog risks filling disk before anyone restarts it.
Every consumer breaks without a Kafka Schema Registry check
for the same guardrail on a topic your own application writes to directly, where the schema comes from a producer's code instead of a connector reacting to someone else's migration.
Checked 26 Aug 2026
Part of the Change Data Capture (CDC) cookbook