Small question (or 2) about prefect 2.0 Trying to build a CICD solution with prefect 2.0. There are a few things I’m still struggling with; • What is the advantage of storing my flow on another source than Git? Using GIT gives me a single source of truth inline with the standards in my organization. • Where does my flow store its cache while running? I want to be able to restart a task when my flow crashed, while running on K8s. Am I correct when assuming that I need to configure External Storage with put-directory and write-path for this? • The MINIO example on the website is trying to connect to an AWS endpoint, probably due to s3fs. Can I just follow the example on https://docs.prefect.io/concepts/blocks/, create my own PyMinio based block and use that? Many thanks in advance

#1 Git is way more than just storage, typically engineers use it for version control and collaboration and then CI/CD deploys the application upon merge to main -- a common scenario in Prefect is that your flow along with your custom modules gets deployed to S3 and your dependencies are built into Docker image and pushed to a registry; or maybe you are leveraging both as part of flow code stored in the Docker image as shown here - we do have plans to support Git as storage but if you want to read more on that check this Discourse page #2 no caching, Prefect always pulls flow code from storage at runtime due to the hybrid model, only container image can be cached using image pull policies

Hi Anna, thnxs for the quick reply #1 Fair point. I will have to look into this how this works in best practice with Prefect. Will look at the link you send me. #2 So if I understand correctly, when a task fails, the flow should run as a whole again. There is no option to start te flow from the task that failed when running in a Docker / k8s environment.

correct, unless you leverage automatic retries which are configurable both on a task and a flow level in Prefect 2