https://prefect.io
Join Slack
In Prefect 2.0, Is there any option via config to change the default port number 4200?
# ask-community
x
In Prefect 2.0, Is there any option via config to change the default port number 4200?
a
Yes, you can set an env variable before starting Orion:
Copy code
export PREFECT_ORION_API_PORT=4242
z
prefect orion start
also takes a 
--port
option. I do not believe the UI supports alternative ports yet though.
upvote 1
x
Thank you Anna and Michael. When I make the port change in .prefect/config.toml the "prefect orion start " command is not taking those changes. Do you know what am I missing.
z
Orion does not respect the 
config.toml
yet.
x
Thank you Michael. If I change the port number from 4200 to 5200, how does the agent know which port the server is running. I deployed and executed few runs I see that status in the agent command line, but I don't see those flows and deployments in the browser UI.
z
You’ll need to set 
PREFECT_ORION_HOST
on the agent to the path that the api is hosted on e.g. 
<http://localhost:5200/api>
I don’t think the UI is going to work with a changed port though.
x
Since the root account using prefect 1.0 with existing default port numbers I am using CONDA virtual env and deployed prefect 2.0, but I can't run it due to port conflicts.
let me try with PREFECT_ORION_HOST
Now the agent is listening to 5200/api and when I deploy and run I see changes in the command line, but I don't see those flows and deployments in the browser UI
how can I see my flows and tasks in UI?
Can I make changes in the settings.py to make it work?
z
As I mentioned, the UI does not support a changed API port yet. It’s hard-coded on this line https://github.com/PrefectHQ/prefect/blob/orion/orion-ui/src/utilities/api.ts#L5
You’d need to install from the source code, modify that line, and run 
prefect dev build-ui
x
ok, let me try. Thanks Michael.
z
Yep! We’re designing a way to configure it at runtime right now but it’s not quite ready yet 🙂
x
which code takes the environment variable or from settings.py while starting prefect orion server? We can apply the same logic to this UI code.
z
The UI code is built statically so it can’t retrieve a runtime variable yet
x
I got the source code and installed. Just checking. after I make change in api.ts every time, do I need to install again, or just I need to restart prefect orion? I believe the former.
by installing the dev code, I can't even see default UI screens like the tabs like flows, tasks, etc. I get {"detail":"Not Found"}, even before I make any changes to the code.
z
Yeah the UI needs to be built into static files for the Python library to serve
prefect dev build-ui
will resolve the “Not found”
You’ll need to rebuild the UI each time you make a change to the type script source
prefect dev start
will host the UI directly from the type script (available at a different port) if you want to test your changes without rebuilding manually
x
thank you, Michael. Let me try.
I can run the UI now. I can deploy and execute the flows and tasks in command line. I do see that activity in Lateness section of the UI, but I don't see the Flows, Deployments in the UI for some reason. I believe when I run prefect orion start, it starts both server and agent (I do see that in the result screen), I don't need to start the agent separately. But for some reason, the deployed tasks are not shown in the UI.
Do I need to build and run any other stuff?
~/.prefect/orion.db is being updated whenever I deploy. I see change in the timestamp.
Do I need to flush and recreate the db?
Also, how can I get the Radar UI integration?
I see the records in orion.db on the required tables e.g. flow. But the UI is not reading it from the db.
a
Do I need to flush and recreate the db?
If you don’t see your flows and flow runs in the UI, then you’re correct, running “prefect orion reset-db” may help to clean the DB state and start from scratch
z
Have you run the deployment?
Your flow/deployment may not appear in the dashboard without any flow runs
Dedicated deployment/flow pages are coming soon.
x
If I run with localhost I can see the my created/executed flows in UI. But if I use different port number or alias host name, the tasks and flows are created in SQLite db when I run it. But, not reflecting in the UI.
Yes I ran the deployment to create and run my flow.
I believe the SQLite URL is not setup correctly for the UI to read when I use different port number or alias host name. Is there any code where I can enforce it similar to api.ts?
Probably I need to make changes in settings.py. Please confirm.
z
The UI hits the API which reads from the DB
It sounds like the UI isn’t hitting the correct API still?
x
if I use 127.0.0.1 with dev build the UI works ok.
but only problem arises when I use different port or alias host
any suggestion to address it?
Also the prefect orion UI is very slow, I am not sure the APIs are taking more time even for simple flows and tasks. Or it is due to DEV build. Can I create a PROD build from the source code and make it go faster.
z
You’re working with development builds / trying a pattern that we don’t support yet. I can’t really speculate as it what’s going on with your connection.
The UI shouldn’t be slow though, I’m not sure what’s going on there.
x
Thanks Michael. QQ, How can I start Prefect Orion using authentication? Also, How to force the user to login while accessing the UI?
z
We don’t have any authentication yet and like Prefect Server I do not think there are plans to add it to the OSS offering due to the complexity involved in deploying secure auth.
x
2 cents. For many firms, security is very important since multiple teams can be part of the same server and all should be separated securely with high level of privacy. I think for now it is a weakest point for Prefect 2.0 at this time. We need to consider adding it. Since Workflow automation involves data and workflow security is a key even within the teams of the same firm.
z
Thanks for the feedback 🙂 I’ve shared this with the team.
110 Views
Open in Slack
PreviousNext
Powered by

Prefect Community

Bring your towel and join one of the fastest growing data communities. Welcome to our second-generation open source orchestration platform, a completely rethought approach to dataflow automation.

Powered by