Is it possible to change the destination schema in Airbyte to a different schema without performing a full refresh? Context: We currently have incremental syncs configured, and we’d like to move the destination tables to another schema (within the same database) without triggering a full reload of all data. Is there any supported way to update the destination schema .

Based on the available knowledge sources, Airbyte does not natively support changing the destination schema (i.e., moving tables to a different schema within the same database) for an existing connection without performing a full refresh. The standard approach to schema changes in Airbyte typically involves either a full refresh or manual intervention. However, there is an advanced upgrade path described for upgrading to Destinations V2 that uses a "dual-writing" method. This process involves: 1. Creating a new destination pointing to the same database but with a different schema. 2. Creating a new connection using the same source and the new destination. 3. For incremental syncs, manually copying the raw data already replicated to the new schema (using SQL in your data warehouse). 4. Copying the connection state from the old connection to the new one, so historical data is not replicated again. 5. Enabling the new connection, which will continue incremental syncs without reloading all historical data. This method allows you to effectively "move" your destination tables to a new schema without a full refresh, but it requires manual steps outside of the Airbyte UI, particularly for copying existing data and state management. There is no direct, in-place schema change for the destination in Airbyte's UI or connection settings that avoids a full refresh or manual intervention for incremental syncs. For more details, see the official documentation on [Upgrading Connections One by One with Dual-Writing](https://docs.airbyte.com/release_notes/upgrading_to_destinations_v2#upgrading-connections-one-by-one-with-dual-writing). If you are not upgrading to Destinations V2, or if you are looking for a simpler, UI-based solution, the knowledge sources do not indicate that such a feature currently exists.

Depending on the destination, you should be able to just copy the table manually, then update the reference on the connection—but it depends on the destination. For example, in the modern BigQuery destination, there are no more raw tables. In older versions, you'd need to move/rename the tables. With some connectors and auto-naming and such, you might need to update the database row for the connection in OSS (obviously that isn't an option for cloud). I would use a small test connection to confirm, and always back up your tables before doing things like this just to be safe.

I just tried for Snowflake; some tables worked, and others didn't. Super weird :s

out of curiosity, were the tables that didn't work incremental and the ones that did set to full refresh or something like that?

No, most of those tables are incremental append + dedup.

interesting. IIRC, Snowflake hasn't yet been updated to use the newer Direct-Load paradigm like BigQuery, so you'd need to move the internal tables as well

Yes I'm using 4.0>