Best practice for deployments in S3

Archive
1

I’m evaluating Prefect hosted in k8s using an agent also in k8s. I’m using S3 for storage. Right now if I want two deployments without their files conflicting with each other in my S3 bucket, it looks like I need to create two separate Blocks each one pointing to a different s3 path in the same S3 bucket, and then build each deployment with --storage-block pointing to their own block.
Is this the preferred way? I wasn’t sure if there was a way to just make each deployment create its own directory within the same block which would avoid having to create a block for each one.

Also, a more basic question: is the above best practice for a situation where we might have users creating, running and removing deployments on an ad-hoc basis?

2

We’ve just merged a PR that may help you here - it allows to add a prefix to the S3 paths:

e.g. --storage-block s3/dev/subdirectory/path

3

Thanks, this looks like just what I need.
For my image, will the following pick it up, or do I need to reference a developer image until its released?
prefecthq/prefect:2-latest

4

yup, that’s how you would need to do it

you can build the image from main branch using prefect dev build-image and use prefecthq/prefect:dev-python3.9