Hi everyone, I want to integrate a dbt project dbt in Dagster and I am trying everything to be able to restart the materialization of DAGs "from the point of failure". Anyone did something similar? I can't find a way to make it work. When I import my assets via manifest or via dbt project, dagster restarts everytime the whole DAG, even if only the last node failed. I tried to load each model as a single asset and it works, but for every run dagster repeats the command "dbt ls" for the whole project and the run time spikes massively.

hi @Lorenzo -- one option that might work for you would be to load everything with a single call to

load_assets_from_dbt_(manifest or project)

, then execute all of your assets. When one fails, instead of trying to restart the job from the point of failure on the job run page, you should be able to go back to the asset graph, where there will be a "Materialize stale and missing" button on the top left corner (you may need to hit the dropdown menu to access that). This will execute all of the assets whose upstream assets have been updated more recently than them, which in the case of a failed dbt job will just be the failed models and their downstream models.

Thank you so much @owen, it works! dagster evolution And it's also pretty straightforeward... I honestly don't know why it did not work previously. I tried so many combinations of import and/or job creations and I could not make it work 🥲