Rich mentioned in https://www.youtube.com/watch?v=thpzXjmYyGk that it’s possible to avoid having lambda cold starts when working with API Gateway in a production topology. Can we somewhere vote on this feature or is it a given that it will be implemented soon?
I think this is more than an issue with timeouts.
Currently, on a cold start, the lambda will sometimes return a 500 Internal Server error with the following body: java.io.IOException: Premature EOS, presumed disconnect
We have a clojurescript application hosted on S3 (different domain than the API - so CORS rules apply) that has some trouble with these responses. In trying to implement retry logic for such an error, we have seen that the browser doesn’t allow this retry request because the Access-Control-Allow-Origin header is not set on the error response. We set this header in the ion, because as far as I understand it is not possible to set this header on an AWS API Gateway proxy (you are not allowed to change the integration response for a proxy method).
Does anyone have a work-around for this?