Open acsbendi opened 2 months ago
Our organization is experiencing this issue and we're not sure how to work around it. We've tried turning on full replica identity for the source table with the generated column.
It's unfortunate because the only reason we're using a generated column is to get around another airbyte postgres issue where it can't sync interval columns to Snowflake destinations.
@tanderson-hp Our workaround is to define a view on top of the synced table, and in this view, fill the NULL
column values using the same logic as in the source table.
Thanks @acsbendi, so do you have a separate connection/stream to handle the view sync? My understanding is CDC replication can only be based off tables.
The view is created in the destination (Snowflake in our case) and it reads the latest data from the synced table so you don't need Airbyte for that.
Connector Name
destination-postgres
Connector Version
3.6.18
What step the error happened?
None
Relevant information
Steps for reproduction:
Incremental | Append + Deduped
orIncremental | Append + Deduped
), run initial sync ->day_inserted
has correct valueinsert into main.generated_column_test(id, timestamp_inserted) values (default, default);
)day_inserted
hasnull
value, incorrectlyThis may be related to #38470.
Relevant log output
Contribute