we were wondering about the full refresh logic in the model and have a question about it. Why are only the manifest tables protected ? If you are doing a full refresh and the data in the page view and session models are lost, then the data in the manifest tables are not much use and you would want to reprocess anyway ? Is there anything against protecting your own tables with the macro ?
Hey @Stella_Oppermann, the logic for this was introduced a little before my time but I can give an idea about it. One reason for this protection is that the manifest table is used by all (derived) tables, which means even if you did a dbt run with a full refresh but selected to just your page views table for a full refresh, it would refresh the whole manifest table which isn’t great!
You’re right if you just do a full-package full-refresh that there’s not a load of point. In general we combine it with a select flag and setting the models_to_remove variable to refresh a specific derived table from the start date (you can see an example here)
You’re welcome to use the macro to protect your own tables against refresh, just be aware that it only works in your prod target, you can full-refresh in your dev target and everything will be refreshed including the manifest tables. It won’t be applied to the config of the package derived tables so you’ll need to add that yourself as well.