Is the normalisation routine performed on the entire "raw data" table when a new sync happens? I am using my own dbt transformations with MongoDB as a source and I'm wondering what happens if I remove all the tables but the the "raw data" while changing schema directly in dbt. Will final table be fully replicated from the raw data table on a subsequent sync?
Hey Eugene yes when the normalisation runs again it gets replicated.
Can I ask you where's the current cursor value is stored in case of incremental sync?
@Eugene Krall you can check the following table in our Airbyte DB:
At which point this cursor refreshes? I have a situation when i had a couple of failed syncs but they failed at the normalization stage, so the raw data was already there, after a couple attempts I noticed that there were duplicate records in my raw data. I suspect that cause may be the cursor that updates only after the entire sync was successful.
It depends of the source connector you're using, the checkpointing can be done in different manners. You can read more about this here. Feel free to share which connector you're using and I'll have a look to its code.