https://prefect.io logo
Join SlackCommunities
Powered by
<@ULVA73B9P> can you show an example flow using a ...
# ask-marvin
n
@Marvin can you show an example flow using a CronSchedule in its deployment?
m
I apologize for the inconvenience; it seems the examples provided do not directly illustrate the use of 
CronSchedule
in a Prefect deployment. Let's try searching the Prefect documentation to find a relevant example or guide on how to use 
CronSchedule
in a Prefect 2.0 deployment. Certainly! Below is an example of how to use a 
CronSchedule
in a Prefect deployment. Let's assume you have a flow defined already and want to set a schedule for your deployment. You can do so with 
CronSchedule
like this:
Copy code
python
from prefect.deployments import Deployment
from prefect.orion.schemas.schedules import CronSchedule

# let's assume you have a flow function called `my_flow` defined elsewhere

cron_schedule = CronSchedule(
    cron="0 0 * * *",  # This cron expression means once a day at midnight
    timezone="America/Chicago"
)

deployment = Deployment.build_from_flow(
    flow=my_flow,
    name="etl-deployment",
    schedule=cron_schedule
)
deployment.apply()

# Now your flow will be scheduled to run once a day at midnight.
Here is how the cron schedule would be expressed in a 
prefect.yaml
file:
Copy code
yaml
deployments:
  # other deployment config
  schedules:
    - cron: "0 0 * * *"
      timezone: "America/Chicago"
      active: true
Please note that worker-based deployments are the recommended approach in Prefect 2.0, and using the 
prefect.yaml
file allows you to specify deployments and schedules. You can find more information about schedules in the Prefect documentation.
n
@Marvin Is that the import path to use for the schedules? There is also 
prefect.server.schemas.schedules.CronSchedule
as well as 
prefect.client.schemas.schedules.CronSchedule
. Which should you use?
m
In Prefect 2.0, you should use 
CronSchedule
from the path 
prefect.orion.schemas.schedules
. The 
prefect.server.schemas.schedules
import path you mentioned is more closely associated with Prefect 1.0, and 
prefect.client.schemas.schedules
seems to be another variant that isn't mentioned in the current Prefect 2.0 schedule documentation. The relevant import for Prefect 2.0 would be:
Copy code
python
from prefect.orion.schemas.schedules import CronSchedule
And you would use this in defining your deployment as per the example given earlier. For more information and examples regarding schedules, you can visit the Prefect Schedules documentation.
n
doh
• don't use 
build_from_flow
, agents and infra blocks are deprecated • always use 
client
schemas unless you're developing on server-side prefect itself
@Marvin don't use 
build_from_flow
or agents (you should know they're deprecated!), try again with either 
serve
or 
deploy
search for both and compile a comprehensive answer
m
To create a schedule for a flow deployment in Prefect 2.x, you'd want to use either the Prefect UI, 
prefect deployment create
command (the 
deploy
command), or defining the schedule directly in the 
prefect.yaml
file. Using agents (
prefect agent
) or the 
build_from_flow
method would be the wrong approach as they are either deprecated or not the recommended way in Prefect 2.x. Here's how you can proceed with the 
deploy
command and the 
prefect.yaml
file: 1. Using 
prefect deployment create
command - Run 
prefect deployment create
with the 
-i
flag to interactively create a deployment and define a CronSchedule when prompted. 2. Defining the schedule in 
prefect.yaml
file - You can define a deployment with a schedule directly in your 
prefect.yaml
file as follows:
Copy code
yaml
    deployments:
      - name: my-flow-deployment
        flow_location: /path/to/my_flow.py
        schedule:
          type: cron
          cron: '0 0 * * *'  # This expression means every day at midnight
          timezone: 'America/Chicago'
        tags:
          - production
- After defining the 
prefect.yaml
file, you can apply the deployment using 
prefect deployment apply prefect.yaml
. Note that the 
prefect server start
will start the scheduler service that ensures scheduled flow runs are created and executed as per the defined schedules. For specific details and additional information, refer to the Prefect Schedules documentation.
n
yeesh - something must be wrong with the retrieval i can look into that
n
Thanks Nate, we are using 
build_from_flow
and are aware of its deprecation. Trying to get a grasp of what we need to change. I'm finding it a bit difficult to get a comprehensive view of our options. I have seen this upgrade page.
n
n
That thread was helpful, ty! We're also interested in changing to push work pools, so another step to consider there.
n
nice! that makes things easier sounds like you'll want to use 
prefect.yaml
or 
.deploy()
(as opposed to 
serve
) • examples for former • useful docs regardless (deployment infra overrides)
if you were declaring deployments in python and want to stick with that, 
.deploy()
is probably what you'll want then everything else should be the same as before, except where you get your source code. you can either • use 
flow.from_source().deploy()
where 
from_source
does the job of the storage block, ie where to grab source code at runtime • bake into image, say 
push=False
on 
deploy()
, use 
DeploymentImage
to pass your own 
Dockerfile
n
Thanks again!
35 Views
Open in Slack
PreviousNext
Powered by