Hi All, I’ve deployed Prefect Orion to AKS. We’ve...
# prefect-community
m
Hi All, I’ve deployed Prefect Orion to AKS. We’ve connected the UI via ingress with an alias. API calls seem to be working with the alias. However, the UI via the alias seems to be giving this error. I’ve been changing the Orion API URL to see if that resolves it to no avail. Does anyone have any insight to what might be happening here?
r
Hi Mark, I think you are seeing this because you need to set the
PREFECT_ORION_UI_API_URL
environment variable in the container where Orion is running. Set it to whatever hostname and port you use to load the UI in your web browser, and add
/api
to the end of the address. So, for example, if you access the UI at
<http://my-orion-instance>
, you'd set
PREFECT_ORION_UI_API_URL
to
<http://my-orion-instance/api>
.
m
Hi Ryan, for the screenshot, I had replaced the PREFECT_ORION_UI_API_URL for privacy purposes. It was still not working when I changed that environment variable to the hostname and port + /api that I use in the UI of my web browser.
Additionally, to clarify, the actual URL is https.
r
Thanks for the additional information, Mark. Do any errors show up in the console in the browser's developer tools?
m
Yep, These are the following errors (not much detail):
@Ryan Peden - Any ideas on what it might be or any leads you can think of?
r
Not yet, but I am going to try to reproduce it this morning and I will reach out to couple of my colleagues to try to get to the bottom of what's happening. Building tutorials around running Orion on k8s is one of my priorities at the moment, so thank you for taking the time to provide additional detail on this. It is very helpful, and I appreciate it!
🙌 1
m
Sounds good. If you need more information to make your tutorials better, please let me know!
I have a few comments regarding the actual deployment and would be happy to bring them up afterward to not convolute the thread.
I see a new values.yml was committed. Definitely breaking changes from prior (which is fine, I specified version.) Just posting to say I’m glad some of the comments i had regarding the persistence nesting (adding primary) in the postgresql definition was addressed. Also, seems like some new QoL changes were added for clearly defining where to specify resource limits, etc..
So kudos to the prefect team there!
Though, still currently running into this issue with our specific version. Will check if the new chart with clearer ways to specify PREFECT API URL fixes anything.
r
Thanks for the follow-up, Mark - and apologies for not getting back to you sooner; I was working on a feature that went into this release. I'll circle back to this issue and see if I can sort out what's going on.
PREFECT_ORION_UI_API_URL
should have worked here - it works well in Docker Compose, and I'd expect it to work the same way in k8s as well, but in your case it doesn't appear to be.
m
Alright. Thanks Ryan. Please keep me informed. Additionally, I started testing out the chart overhaul released yesterday-> Noticed a fix regarding a typo in the ingress.yaml file has been committed but not yet sent out for a release -> Any idea when the fix on the typo in the ingress.yaml will be released?
Leveraging the 0.3.1 version for Orion. I had set the Prefect_API_URL which is reflected in the error on the Orion UI. Attached is what we are using with our values.yaml file; just without our alias. Are we missing something? - Left it pretty basic just to have a base running api we could further configure. • Keep in mind, we will likely move to the new chart once the ingress typo fix is released.
@Ryan Peden - Turns out, with ingress turned on and our alias, it’s failing because we are specifying the port.
r
Ah! Makes sense. I should have noticed the port. I've encountered the same thing when running Orion behind Nginx with Docker Compose.