@kapa.ai I am using convex as a source and bigquery as a destination. The setup works fine and it outputs to bigquery in 4 different tables I selected. However, each table outputted has a

_table

property that includes other tables besides the table I selected. The schema is correct, so for rows matching the wrong _table almost all properties are null. What is going on?

The knowledge sources don’t explicitly mention a

_table

field for the Convex source, so I can’t give a definitive, documented explanation of that specific behavior. I can only rely on what is documented and what can be inferred. From the Convex source docs: - Each Convex table is synced as a separate Airbyte stream. [[Convex source](https://docs.airbyte.com/integrations/sources/convex)] - Every record includes: - Your client-defined fields - System fields:

_id

,

_creationTime

,

_ts

,

_deleted

[[Convex source](https://docs.airbyte.com/integrations/sources/convex)] There is no mention of a

_table

field being added by the Convex source itself. [[Convex source](https://docs.airbyte.com/integrations/sources/convex)] On the destination side (BigQuery with Destinations V2), Airbyte adds its own

_airbyte_*

metadata columns and may change how JSON / metadata is stored, but again there is no documented

_table

column there either. [[Destinations V2](https://docs.airbyte.com/release_notes/upgrading_to_destinations_v2)] Given that: -

_table

is not documented as a Convex system field. -

_table

is not documented as a standard Airbyte metadata column for Destinations V2. - You see rows in each BigQuery table where

_table

refers to other tables and all the other fields are null. The most likely explanation (but not confirmed by docs) is: - Some upstream process (either in your Convex data model, or in a custom transformation / prior pipeline) is including a

_table

field in the source data itself, and that field is being carried through by Airbyte into all streams/tables. - When that

_table

value doesn’t match the current stream’s logical table, the other columns don’t have values, so they appear as null in BigQuery. However, because the docs do not describe

_table

at all, I cannot confirm whether this is: - An intentional behavior of the Convex connector, - A bug in the Convex source, - Or something specific to your data model / transformations. To move forward, based on the available documentation: 1. Verify in Convex directly Check the raw documents in the Convex dashboard “Data” view for each table and see whether a

_table

field exists there. [[Convex source](https://docs.airbyte.com/integrations/sources/convex)] 2. Check if

_table

is a user-defined field If you find

_table

in your Convex documents, it’s likely a client-defined field that Airbyte is just syncing as-is. 3. **If

_table

is not in Convex** Then this is behavior not covered by the docs; you should open an issue or support ticket with logs and example rows, as it may be a connector bug or an undocumented feature. The Convex connector docs explicitly invite you to reach out to Convex support for connector-related performance/behavior issues. [[Convex source](https://docs.airbyte.com/integrations/sources/convex)] Because the knowledge sources don’t document

_table

at all, I can’t give a more precise or guaranteed explanation than this.