How do I create and run flows on a schedule in Prefect Orion

Question

How do I create and run flows on a schedule in Prefect Orion Let s say I have this ```schedule = IntervalSchedule interval=datetime timedelta seconds=10 deployment spec = DeploymentSpec name= hola flow=greetings flow schedule=schedule ``` Do I now use OrionClient and some methods to deploy We currently use Prefect core in a pod and simply do flow run with a schedule attached and I m looking for a working example to do the equivalent in Orion even if I gather behind the scenes it will do something else via a server and DB Basically I m looking for this <https orion docs prefect io concepts deployments running deployments with the api> slightly smiling face

Michael Adkins · Answer

Hey < Christoph Deil> the easiest way to deploy from a specification like that is from the CLI `prefect deployment create file py` Then you d start the server `prefect orion start` We actually don t have a great way to support the `flow run ` on a schedule pattern yet we ll have to think about the best way to address that use case

Michael Adkins · Answer

I d recommend using cron to call your flow script every 10 seconds instead

Christoph Deil · Answer

This seems to almost work and do the equivalent ``` schedule = IntervalSchedule interval=datetime timedelta seconds=10 deployment spec = DeploymentSpec name= hola flow=greetings flow schedule=schedule client = OrionClient deployment id = create deployment from spec spec=deployment spec client=client subprocess run prefect orion start shell=True ``` But I see some errors ```22 59 32 526 | Exception in callback SubprocessStreamProtocol pipe data received 1 b EXCEPTION 8 XCEPTION 81 n handle lt Handle SubprocessStreamProtocol pipe data received 1 b EXCEPTION 8 XCEPTION 81 n gt Traceback most recent call last File Users cdeil opt anaconda3 envs prefect orion lib python3 9 asyncio events py line 80 in run self context run self callback self args File Users cdeil opt anaconda3 envs prefect orion lib python3 9 asyncio subprocess py line 73 in pipe data received reader feed data data File Users cdeil opt anaconda3 envs prefect orion lib python3 9 asyncio streams py line 472 in feed data assert not self eof feed data after feed eof AssertionError feed data after feed eof 22 59 32 535 | Flow run 3f327c9d dbd5 43af a61f 1c6e6df30c86 exited with exception BrokenWorkerProcess ``` So I guess trying to make it work from one Python process isn t a good idea it s what we currently have a click CLI which does some argument stuff and then creates a flow and schedule and runs it based on user arguments to the CLI

Michael Adkins · Answer

Yeah I wouldn t recommend trying to do it that way

Michael Adkins · Answer

For your ~use case~ example ```while True greetings flow time sleep 10 ``` should be sufficient

Michael Adkins · Answer

You can handle an arbitrary schedule with something like but the `Schedule get dates` interface is only intended to be consumed by our scheduler ```import sys from datetime import timedelta import anyio import pendulum from prefect import flow from prefect orion schemas schedules import IntervalSchedule flow def my flow print f Hello Running with sys executable schedule = IntervalSchedule interval=timedelta seconds=10 async def main while True now = pendulum now utc next date = await schedule get dates start=now n=1 0 wait time = now diff next date total seconds print f Waiting wait time seconds until next date await anyio sleep wait time my flow anyio run main ```

Christoph Deil · Answer

< Michael Adkins> thank you We ll try it out and discuss in the team today which style we prefer either sleep in function or only deploy from the script but then run `prefect orion start` separately as main blocking command in the pod

Michael Adkins · Answer

The sleep solution is better if you want to pull all of the run data into a single Orion instance in the future you can just set a `PREFECT ORION HOST` env var and they ll report to one server

Michael Adkins · Answer

In the second you ll be running multiple servers and if you mix their data they ll start running flows that belong in other pods