I have added the HTTP Proxy into the docker.env file and can communicate with the licence server during start up.
However, when I create a REST resource trying to connect to an endpoint in the internet, it doesn't seem to be using the http proxy setting because the error message is saying can't resolve the address. This is an indication the connection is not going through the corporate http proxy server.
FetchError: network timeout at: https://httpbin.org/get?id=1\n at Timeout.<anonymous> (/snapshot/retool_development/node_modules/node-fetch/lib/index.js:1484:13
Version 2.106.4
HTTP_PROXY is set and shown on the Retool Settings page "On Prem Environment". Communication with the license server also works during startup. When checking the communication after running the query, the request doesn't seem to go through the corporate proxy.
From the retool container this test works:
node -e "require('./common/fetch.js')('https://httpbin.org/get?id=1').then((r) => console.log(r));"
Your script in the container sets the proxy, but doesn't seem to be used by the backend.
Thanks for surfacing this. Currently, only a small amount of egress calls respect the HTTP_PROXY environment variable, but the devs are looking to make it so that all egress requests do. The fix should be shipping with our next release (version 2.110)!
Thanks for flagging this here. The HTTP_PROXY_STRICT is specifically for backwards compatibility for use cases like yours where the change causes issues. If you used only the NO_PROXY variable without setting HTTP_PROXY_STRICT=false where specifically are you running into issues?
I'll bring it up with the team to include better documentation on these variables!