Tracing information is misleading when proxying

[fiadliel]

When trying to debug an issue with request proxying (it turned out that the method was being dropped), I initially looked at the trace information as something that might be useful. The output looked like this:

2024-10-02T10:23:49.376042Z DEBUG request:proxy_http_request: starting new connection: https://upstream-host/     method=POST uri=/ version=HTTP/1.1
2024-10-02T10:23:49.376241Z DEBUG request:proxy_http_request:resolve: resolving host="upstream-host" method=POST uri=/ version=HTTP/1.1 host=upstream-host
2024-10-02T10:23:49.380694Z DEBUG request:proxy_http_request: connecting to upstream-ip:443 method=POST uri=/ version=HTTP/1.1
2024-10-02T10:23:49.395309Z DEBUG request:proxy_http_request: connected to upstream-ip:443 method=POST uri=/ version=HTTP/1.1
2024-10-02T10:23:49.575981Z DEBUG request: finished processing request latency=201 ms status=400 method=POST uri=/ version=HTTP/1.1

Information related to the incoming request: URI (rooted at the sub-router) and method.
Information related to the upstream: the response latency, status, host and IP being connected to.

Ideally, information here would be relevant to the stage of processing; there would be a span where the outgoing request is made, and request information in that span would be related to the request after rewriting/processing.

I think it's a low priority issue, but it's useful for people debugging proxy issues to have clear tracing information.