Hi all, Quick question about flow.from_source(). Can I use this from within the same repo? As in, can I do something like this from within the same repo defined by the GitRepo url? @flow async def my_flow(name: str = "world"): print(f"Hello {name}!") if name == "__main__": DEFAULT_GH_SECRET_BLOCK_KEY = "github-secret" access_token = Secret.load(DEFAULT_GH_SECRET_BLOCK_KEY).value flow.from_source( source=GitRepository( url="https://github.com/company/my-repo.git", branch="main", credentials={"access_token": access_token}, ), entrypoint="/builtin_worklflows/paige-test.py:my_flow", ).deploy( name="my-deployment", work_pool_name="my-k8s-job", job_variables=dict(pull_policy="Never"), schedule=(CronSchedule(cron="0 0 * * *", timezone="America/Chicago")), tags=["test1","test2"], paramaters=dict(name='Paige') )

hi @Paige Fuller! is there a reason that you wouldn't want to do something like this? you don't need

from_source

if you already have the flow object handy

I guess this goes into my lack of understanding of how prefect deployments work! We currently copy all of our files into GCS bucket and then deploy from there, but we are trying to move away from there. If we do this, we werent sure where prefect would get the code from when the deployment is actually run? We are using the open source and not the cloud offering

this will be helpful I think

I mean, we have all of the code living in GitHub so ideally we just pull from there! We just seem to be struggling to make it happen haha

nice! i like keeping my code in github so i dont have to rebuild images every time my code changes best approach is always different for folks, but here's what I'd say (given what I know now from chatting): • to minimize a lot of changes for you all (since you were already using

from_source

) - I would stick with

from_source

and just point it at github like your original example, this will: ◦ allow you to write some GHA like you did above (just set

PREFECT_API_URL

there and you should be good!) ◦

from_source

will make it so that the eventual

Deployment

you register with the server has the information about where the flow code lives when the worker picks up a flow run from that deployment if you're trying this

@flow
async def my_flow(name: str = "world"):
    print(f"Hello {name}!")

if __name__ == "__main__":
    flow.from_source(
      ...
    ).deploy(...)

like above and its not working for you, what's the error look like that you're getting?

ok fab, will give it a try! I havent had it approved to merge into dev yet until I can convince people its a good idea, so this chat is good ammo haha! Pointint to the github repo from source works fine when its at the same repo in which its running? Just double confirming!

Pointing to the github repo from source works fine when its at the same repo in which its running?

yes it does! because all you're doing at the moment you call

.deploy()

is telling your prefect server that "hey, this Deployment I'm giving you lives at • source=

<https://github.com/company/my-repo.git>

• entrpoint=

path/to/my/flow/relative/to/repo/root.py:my_flow_function

so when its time to run, that's where you should go

pull

it from!" (behind the scenes, we translate that

from_source

bit into a

pull

step for your worker that instructs it on how to retrieve your flow code at runtime)

ahh thanks! Will change that - thank you!! Appreciate you always responding to my chaotic questions 🙂

@Nate actually going to ask you one more question while I have you trapped in this thread!! If we want to deploy a flow as a k8s job, if we deploy to a kubernetes agent pool, will it automatically run as a k8s job? We had some before we defined in deployment.yamls that ran as a process