You can set it with the

--concurrency-limit-collision-strategy

flag on

prefect deploy

. Example (adding CANCEL_NEW to your command):

uv run prefect deploy ./workflows/example_workflow.py:example_flow \
  --name example-deployment \
  --pool local-pool \
  --cron "* * * * *" \
  --concurrency-limit 1 \
  --concurrency-limit-collision-strategy CANCEL_NEW

Notes: - Valid strategies include at least

CANCEL_NEW

(do not queue; cancel new runs when the limit is reached). If you omit this flag, Prefect will use its default behavior (typically queueing). - You can also set this in

prefect.yaml

if you prefer config over CLI:

deployments:
  - name: example-deployment
    entrypoint: ./workflows/example_workflow.py:example_flow
    work_pool: local-pool
    schedules:
      - cron: "* * * * *"
    concurrency_limit:
      limit: 1
      collision_strategy: CANCEL_NEW

References: - Global concurrency limits (context on strategies) - I verified the

prefect deploy

interface exposes

concurrency_limit

and

concurrency_limit_collision_strategy

options. If you want, I can paste the full

--help

output.