Doesn't work behind Nginx reverse proxy

[cablespaghetti]

Whilst this seems to be reliable when accessed directly, when behind my Kubernetes cluster's Nginx ingress I get a lot of "upstream prematurely closed connection while sending to client" errors and Chrome fails to load a high percentage of the resources. If I request a single file using curl it seems to work. This is using HTTP/1.1.

[thomasmhofmann]

@cablespaghetti Did you find a solution? I am experiencing something similar. Although the ingress does not seem to be a problem a keycloak gatekeeper proxy between the ingress and the s3-proxy does not work with larger files.

[wesmcouch]

~This fork solved my issues https://github.com/daveoxley/aws-s3-proxy~ This fork has been causing me issues and crashing

[cablespaghetti]

@thomasmhofmann @wesmcouch Sorry not found a solution. It would be really good to have a version of this I could use in production.

[Dirrk]

@cablespaghetti I was having this issue when behind an aws alb, and I solved it by setting CONTENT_ENCODING=false in the env. Which is actually enabled by default, but documented incorrectly (see https://github.com/pottava/aws-s3-proxy/issues/21). Hopefully this will help you

[pottava]

Sorry for the inconvenience, everyone! @Dirrk Thanks for the advice. I hope it works.

Since I fixed this issue (https://github.com/pottava/aws-s3-proxy/issues/20) recently as well, you may not have to change any configurations.

[cablespaghetti]

Great thanks. I'll try the new version as soon as I can.

[thomasmhofmann]

Just for the record. My problem was not related to aws-s3-proxy. It was a keycloak gatekeeper in front of the proxy that needed to be configured for longer timeouts.