<@ULVA73B9P> I'm using the dask task_runner. I alr...
# ask-marvind
David Martin Calalang
07/08/2025, 9:20 PM@Marvin I'm using the dask task_runner. I already ran `prefect server start to show a dashboard (on port 4200). However, when I run my python script with prefect I can't view the flow on the dashboard (port 4200). I believe it's being displayed on a different port/dashboard but I can't find it. I also get this message on my output " INFO | prefect - Starting temporary server on http://127.0.0.1:8444", but when I click on the link it shows a "can't reach this page" error.
m
Marvin
07/08/2025, 9:20 PM✅ thought for 41.1 seconds
Marvin
07/08/2025, 9:20 PMThere are actually two separate dashboards at play here, which explains the confusion:
1. The Prefect UI (port 4200) - This is where you see your flow runs and overall Prefect orchestration
2. The Dask dashboard - This is what's trying to start on port 8444, and it's specifically for monitoring Dask task execution
The "temporary server" message you're seeing is for the Dask dashboard, not the Prefect UI. The reason you can't access it immediately is that the Dask dashboard is only available while your flow is actually running.
Here's how to fix this:
1. Keep your Prefect server running on port 4200
2. When you run your flow, watch for the Dask dashboard URL in the logs
3. While your flow is actively running, try accessing the Dask dashboard URL
4. Your flow runs will still show up in the Prefect UI (port 4200), but the Dask-specific metrics will be on the Dask dashboard
If you want to ensure the Dask dashboard is accessible, you can explicitly configure the DaskTaskRunner. Here's an example:
Copy code
python
from prefect import flow
from prefect_dask import DaskTaskRunner
@flow(task_runner=DaskTaskRunner(
cluster_kwargs={"dashboard_address": ":8787"} # Specify a fixed port
))
def my_flow():
# your flow code here
passn
Nate
07/08/2025, 9:20 PMwhat does
prefect config viewNate
07/08/2025, 9:21 PMi suspect you're running against an ephemeral server instead of the one you intend to use
d
David Martin Calalang
07/08/2025, 9:22 PM Copy code
PREFECT_PROFILE='local'
PREFECT_API_URL='<http://127.0.0.1:4200/api>' (from profile)David Martin Calalang
07/08/2025, 9:24 PMoh nevermind...
Copy code
PREFECT_PROFILE='ephemeral'
PREFECT_SERVER_ALLOW_EPHEMERAL_MODE='true' (from profile)David Martin Calalang
07/08/2025, 9:25 PMso I'm running my python script (with prefect) through a docker container, I'm assuming this is why the context is different
David Martin Calalang
07/08/2025, 9:28 PM@Nate Could I simply set an environment variable within my docker container environment to ensure that it stays on the intended server (port 4200)?
n
Nate
07/08/2025, 9:38 PMim not sure i understand your setup
> through a docker container
where is your script relative to the server? the main point is that
PREFECT_API_URL Copy code
» docker run -p 4200:4200 --rm -d prefecthq/prefect:3.4.7-python3.12 -- prefect server start --host 0.0.0.0 Copy code
» p config view
🚀 you are connected to:
<http://127.0.0.1:4200>
PREFECT_PROFILE='bleeding'
PREFECT_API_URL='<http://127.0.0.1:4200/api>' (from profile)d
David Martin Calalang
07/08/2025, 9:46 PMEssentially, my setup is that I'm mounting some directories
Copy code
docker run --rm -it -v C:/Users/foo/project:/app --workdir //app image:latest bashDavid Martin Calalang
07/09/2025, 5:15 PMIssue resolved