@kapa.ai What is the exact SQL sort condition that the current version of Airbyte uses to determine transaction order when delivering Postgres CDC records to Snowflake?

The knowledge sources do not contain enough information to answer this query with confidence. Specifically, none of the provided documents describe the exact SQL sort/ordering condition used when processing Postgres CDC records destined for Snowflake. What the knowledge sources do show is the sort condition used in Airbyte's normalization layer (SCD Type 2 model) for ordering records generally (not Snowflake-CDC-specific):

ORDER BY
    <cursor_field> IS NULL ASC,
    <cursor_field> DESC,
    _airbyte_emitted_at DESC

This is used in the

lag()

and

row_number()

window functions to determine

_airbyte_end_at

and

_airbyte_active_row

in the SCD table. [BigQuery SQL optimization; incremental sync tutorial] For CDC sources, Airbyte adds metadata columns like

_ab_cdc_updated_at

and

_ab_cdc_deleted_at

, and these are factored into the normalization SQL via

cdc_updated_at_order

and

cdc_active_row

variables in the SCD model. [CDC docs] However, the exact SQL expression for those CDC-specific ordering variables is not included in the provided knowledge sources. For the precise, up-to-date SQL used for Postgres CDC → Snowflake ordering, I'd recommend checking the Airbyte source code directly, particularly the normalization templates in stream_processor.py and the Snowflake destination connector. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.