Ryan Peden
03/17/2023, 12:02 AMprefect.runtime
namespace:
from prefect.runtime import flow_run
from prefect import flow
from prefect_dask.task_runners import DaskTaskRunner
@flow(task_runner=DaskTaskRunner(client_kwargs = {"name": flow_run.id}))
def my_flow():
...
This will create a Dask client whose name mirrors the flow run ID. Similarly, you can use prefect.runtime
to access parameters that were passed to this deployment run via prefect.runtime.deployment.parameters
. Note that all of these attributes will be empty if they are not available.
See #8790 for details.
A few other key enhancements and fixes:
Enhancements
⢠Add deployment ID support to run_deployment
:identification_card: ā #7958
⢠Disable Postgres JIT for performance improvements š ā #8804
Fixes
⢠Fix blocking file read in async method Deployment.load_from_yaml
š ļø ā #8798
⢠Allow tasks and flows to make redundant transitions such as RUNNING
-> RUNNING
šāāļø šāāļø ā #8802
Documentation
⢠Update workspace roles table to emphasize differences between roles ā #8787
⢠Add Webhook block docs :hook: ā #8773
⢠Update info on Ray's support for hardware and software :ray:ā #8811
Helm chart
⢠Helm charts are now automatically published on each Prefect release :kubernetes: ā #8776
Contributors
⢠@devanshdoshi9
See the release notes for a full list of changes!