That usually means the run was created, but nothing is picking it up to start execution. In Prefect 3.x,

Late

means the scheduled start time passed and the run never moved to `Pending`/`Running`. Most common causes: - no worker is running - worker is polling the wrong work pool - the work pool is paused - concurrency limits are blocking pickup - the deployment is assigned to a different pool/queue than the worker First things to check 1. Is a worker actually running? - In the UI, go to Work Pools and check whether your pool has an online worker. 2. Is the deployment using the same work pool your worker is polling? - If your worker is started against pool

A

, but the deployment sends runs to pool

B

, runs will sit in

Scheduled

and then become

Late

. 3. Is the work pool paused? - Run:

bash
prefect work-pool ls
prefect work-pool inspect <your-pool-name>

4. Are concurrency limits full? - Check whether the pool or queue has a concurrency limit that’s preventing new runs from starting. 5. Preview what the pool thinks is scheduled

bash
prefect work-pool preview <your-pool-name> --hours 24

What “Late” means technically - Prefect marks a run

Late

when its scheduled time has passed and it has not been claimed quickly enough by a worker. - So this is usually a worker / work pool / infrastructure pickup problem, not a flow code problem. Good next debugging info to gather - how you are starting the worker - the deployment’s work pool name - whether this is flow runs or task runs - worker logs around the time a run becomes late If you want, send me: - your

prefect worker start ...

command - the output of

prefect work-pool inspect <pool>

- and a screenshot or copy of one affected run’s details and I can help pinpoint the exact mismatch. Helpful docs: - Work pools and workers - Prefect states

I can’t file it directly from here unless we’re turning this thread into a GitHub discussion, and I wasn’t able to confirm whether one already exists because GitHub search is failing right now. That said, this sounds like a good product suggestion: If a deployment is triggered and there are no online workers for its work pool, show a clear UI warning instead of just letting runs drift into

Late

. You’ve got two good self-serve options: 1. Open a GitHub discussion / feature request in the Prefect repo: GitHub discussions 2. Contact Prefect directly with product feedback: prefect.io/contact If helpful, you can paste this:

text
Feature request: surface a UI warning when a deployment is run but no workers are online for its assigned work pool.

Problem:
Today, a deployment can be triggered successfully, but if no workers are running for that work pool, the run sits in Scheduled, then transitions to Late, which can be confusing.

Suggested behavior:
When a deployment is triggered, Prefect should detect whether the assigned work pool has any online workers capable of picking up the run. If not, show a prominent UI warning on the deployment/run page.

Why this helps:
This would make a common misconfiguration much easier to diagnose and reduce time spent debugging Scheduled/Late runs.

If you want, I can also help you tighten that wording for either a GitHub discussion or an internal feature request form.

fixed!