What you’re seeing is expected worker behavior in Prefect 3.x, but probably not the root cause you think it is. Key point: If you start a worker without

--work-queue

, the worker will poll all queues in the work pool — not just

default

. So this:

bash
prefect worker start --pool my-pool

means:

text
poll every queue in my-pool

not:

text
poll only the default queue

I verified the CLI help for

prefect worker start

, and

--work-queue/-q

is optional specifically to restrict which queues the worker pulls from. Most likely, your deployments are being assigned to the

default

queue because

work_queue_name

was not set when deploying. Common cause If your deployment config looks like this:

yaml
deployments:
  - name: my-deployment
    work_pool:
      name: my-pool

then Prefect will use the pool’s default queue unless you explicitly set:

yaml
deployments:
  - name: my-deployment
    work_pool:
      name: my-pool
      work_queue_name: my-other-queue

Or in Python:

python
my_flow.deploy(
    name="my-deployment",
    work_pool_name="my-pool",
    work_queue_name="my-other-queue",
)

How to confirm Check your deployment in the UI or inspect it via CLI and see which queue it is assigned to. If it says

default

, then the worker is just correctly picking up runs from that queue. If you want queue-specific workers Start workers like this:

bash
prefect worker start --pool my-pool --work-queue queue-a

or multiple queues:

bash
prefect worker start --pool my-pool --work-queue queue-a --work-queue queue-b

Summary - No

--work-queue

on the worker => worker polls all queues in the pool - No

work_queue_name

on the deployment => deployment usually goes to the pool’s

default

queue - So flows running on

default

usually means the deployment is targeting

default

Relevant docs: - Worker CLI docs - prefect.yaml deployment config If you want, paste your

prefect.yaml

or deployment code and I can point out exactly where the queue is being set to

default

.

@Marvin - during worker starts, worker is polling only default queue and other queues are showing as Not ready, even though the worker is online. worker queue doesn't have any actively polling workers ready to execute work. This is message i see on worker queue info button

Thanks — that extra detail changes the diagnosis. If the UI says for non-default queues:

text
This work queue doesn't have any actively polling workers ready to execute work

that means those queues are not being polled, even if the worker itself is online. Important distinction: - Worker online = worker heartbeat is being received - Queue ready = that specific queue is being polled for scheduled flow runs Those are not the same thing in Prefect 3.x. What should happen normally If you start a worker with no queue restriction:

bash
prefect worker start --pool my-pool

it should poll all queues in that pool, and those queues should become ready. If only

default

becomes ready and the others stay

Not ready

, then usually one of these is true: 1. The worker is actually restricted to the default queue For example, it may have been started with:

bash
prefect worker start --pool my-pool --work-queue default

or equivalent env/config values. 2. The worker is connected to a different work pool than the queues you’re checking A worker only polls queues in its own pool. 3. The worker has not yet successfully polled those queues Heartbeat alone does not mark queues ready. What to check Check the actual worker start command Make sure it really does not include

-q default

/

--work-queue default

. Verified CLI syntax:

bash
prefect worker start --help

shows

--work-queue/-q

is optional, and if set, it limits the queues polled. Check the worker’s pool Make sure the worker is started against the same pool where those queues exist. Inspect the queues You can list queues with:

bash
prefect work-queue ls

and inspect one with:

bash
prefect work-queue inspect <queue-id>

CLI help for work queues: - Work queue CLI docs Most likely explanation Based on your description, the most likely issue is that the worker is effectively polling only

default

— either because: - it was started with a queue filter, or - the environment/config supplying the worker command is injecting that queue filter. If you’re running the worker via Docker, Helm, Kubernetes, systemd, or some wrapper script, double-check the real command being executed. What I’d try next Start a worker explicitly against the queues you want:

bash
prefect worker start --pool my-pool --work-queue default --work-queue queue-a --work-queue queue-b

If that makes those queues become ready, then we’ve confirmed it’s a queue-selection/config issue. If you want, send me: - the exact worker start command - whether you’re using Cloud or Server - whether this is Docker/Kubernetes/Helm/local - output of

prefect version

and I can help narrow down why only

default

is being polled.