AUTH-421 Fargate FastAPI container

[y3rsh]

Switch architecture yet again

• We now have 176 seconds to respond to requests.
• Auth is a part of the API.
• Docs https://staging.opentrons.ai/api/doc or https://staging.opentrons.ai/api/redoc

This is all deployed on staging and is responding to requests from #15193

[koji]

left a few comments but they aren't a blocker.
the changes make sense to me.
Thank you for doing this!

[Elyorcv]

I want to understand more on 176s. Sometimes, to generate protocol we may need more than 3 mins, do you think 176s is related to this?

[y3rsh]

I want to understand more on 176s. Sometimes, to generate protocol we may need more than 3 mins, do you think 176s is related to this?

178 seconds is the maximum possible for this architecture and is already not optimal for waiting on a POST request to return a response. There are many other ways to do this we will need to iterate on performance and other methods getting the response.

[Elyorcv]

We should not keep changing the version please. Reproducibility is already difficult due to GPT model itself.

[Elyorcv]

better to stick to 1.25.1, AI projects are updated fast. it may damage things unexpectedly.