1. If I create a deployment, it seems to refer to the local path of the flow (docs indicate the default is to upload the flow so it’s portable) 2. If I try to update the deployment by running

prefect deployment create my_deployment_spec.py

I get

prefect.exceptions.ObjectAlreadyExists

3. If I try to configure the deployment spec in yaml, on create I get

AttributeError: 'set' object has no attribute 'items'

4. There doesn’t seem to be any way to delete deployments to try again and updating is erroring as above… so I’m just sort of stuck with them.

Sorry this is a little rough around the edges still, there are significant changes being made to the “storage” of flows for deployments and it’s introduced some friction into the UX.

Only other big outstanding question I had was around configuring storage. It seems like storage is presented in the API as ‘blocks’ that have some special inner params. Is that right? If so is there an example of them anywhere?

If I can figure that part out then I think I can contribute a Kubernetes operator that will do all the prod-level deployment automation

Storage is a real tricky bit because the CLI for it is interactive, so can’t even fall back to that without it being really ropey

I’m trying to think of a good way for you to workaround it in the meantime…

You don’t technically have to interact with blocks at all to use a deployment right now, it’s just the interface we’re exposing

I can’t see any place in the client, APIs etc to create that storage config in a straightforward way and couldn’t figure out the actual spec used for one when they’re added to the server

Never mind! Just found the spec by looking at the DB table

block_spec

We’re finalizing the design changes right now so things will probably be quite different in the coming weeks

This has been super helpful, thank you. I think I might be able to work around everything now. I’ll see what I can put together and hopefully share with everyone.