Hi - I followed up over DMs, but I. can try to help here.
Typically a Helm Chart is pretty overkill for most people's usecases. I would highly recommend using a single node service on a single EC2 to run Helicone during an MVP stage, or partnering with us to use our enterprise cloud offerings if you are concerned about security/privacy with our basic cloud offerings.
If you still want to go down the HELM route, can you share some more logs of the other containers? I suspect there is a networking error, or the DB or auth service is crashing.
Great, We deploy a lot of stuff into kubernetes, 100s of systems, and have a robust system set up for monitoring, scaling and keeping track of the costs of everything in there. It usually our first choice for running our own systems and third party self hosting.
I have had a look over what I think some of the changes have been that need to be updated in the charts, the worker service seems to have been split into two containers, the oai proxy and helicone api. There is a supabase migration runner that's not in the charts, and we'll need a pass on the env vars to update them. It would be easier if it was super clear what env vars were for what container.
