Trace ID configuration experimental feature

[bnjjj]

This is the discussion related to the experimental feature introduced here.

It's all about how to configure the way we can import and export the trace id.
As it's an experimental feature this discussion is opened to gather feedback from the community. Feel free to give your opinions.

Current configuration:

telemetry:
  tracing:
    experimental_response_trace_id:
      enabled: true # default: false
      header_name: "my-trace-id" # default: "apollo-trace-id"
    propagation:
      from_request_header: "x-request-id"
      jaeger: true

[mandajoan]

Hey Apollo team 👋

We are leveraging this feature to expedite debugging by passing the trace id via the response headers to our client who then propagates it accordingly downstream to our various telemetry configurations.

[bnjjj]

Great ! Thanks for your feedback ! Let us know if you're having trouble using this feature :)

[abernix]

Yay! Very much appreciate you engaging and letting us know! 🤘

[panatl]

I see this is in response. Can we make it in request, right on the top of RouterService, so all the subgraphs and GraphOS get it.

[bnjjj]

I'm not sure to understand your request, could you give an example of your usage ?

[panatl]

we are adding the custom header using rhai

let id = uuid_v4(); 
fn process_request(request) { 
  request.headers.routerTraceId = id; 
}
fn process_response(response) { 
  response.headers.routerTraceId = id; 
} 
fn supergraph_service(service) { 
  const request_callback = Fn("process_request"); 
  const response_callback = Fn("process_response"); 
  service.map_request(request_callback); 
  service.map_response(response_callback); 
}

[theJC]

#4422 -- would like to be able to control the format - want base10 representation for our users.

[trifonovms]

How i can disable this header? latest version 1.58.0 has this header as required, event if i'm not mention tracing section, event removing telemetry section from config still add header apollo-trace-id

[trifonovms]

Today tested with 1.58.1, and have the same result

[trifonovms]

nvm, seems it's related to this issue: #4337

[riginoommen]

I am really liking this feature we have an active usecase to support this

[theJC]

Can this feature graduate out from being experimental?