[UX] - Setting up self-hosted instance of Kern AI

[Bartmr]

Hello. I'm trying to deploy Kern in a cloud VM.

It is not easy, as the instructions provided alone only work with a regular computer with Kern AI running in localhost.

I tried both cloning the repo and installing the pip dependency (which are basically the same thing as I understand), and here's what I found out:

• accessing the UI as the recommended http://:4455/app/ from the internet returns me {"error":{"code":404,"status":"Not Found","message":"Requested url does not match any rules"}} which is generated from oathkeeper from what I understand. If I open this page from inside the cloud VM, i get {"error":{"code":401,"status":"Unauthorized","message":"Access credentials are invalid"}

• I try to spot where the nginx exposes its port, and its the port 7050. I try to open http://:7050/app and I do get Angular's built index.html, but all further asset requests always return me the contents of index.html, as you can see in the attached screenshots.

Has anybody deployed Kern previously? Could you specify the steps? Thank you 😄

[LeonardPuettmann]

Hello Bartolomeu, thank you for choosing to use refinery! We already offer a hosted version as part of our managed cloud service. If you'd like, we can provide you with access to that! If that is not an option for you, please allow us to dive a little deeper into the problem that you are facing and get back to you soon. Also, could you specify the needs for your server setup in more detail? Thank you! 🙂

[Bartmr]

@LeonardPuettmann I'm currently trying to run an instance of Kern AI as you would normally do locally, but I want to run it in a DigitalOcean Droplet, which is a headless Ubuntu machine that can only be accessed through SSH. I want to run Kern AI in that machine, and access the UI interface through the browser. Unfortunately I do not have any more requirements, as we're still evaluating the possibility of using Kern AI.

[jhoetter]

Hi @Bartmr :)

We currently don't have in the OSS version an option to host refinery on e.g. AWS or DigitalOcean yet. We heard that request multiple times already, so be sure that its evaluation (whether we can include it, and how) is on our roadmap. At the moment, we're however already booked with other features to grow refinery. To give you some estimation, we will have a better answer to that question mid/end of Q1 2023.

We have a hosted version of refinery (https://app.kern.ai), and we're happy to give you enough time to evaluate refinery with your team.

Let us know, we're happy to help :)