Why does this crate force the use of actix-http/zstd?

It looks like this was introduced by 10bee3e, cc @fourbytes .

I've tested by using the example provided from the main branch. I was trying to refresh the same page (e.g. /metrics) again and again, then I found that, after some period of time, http_server_active_requests got duplicated. I'm not sure whether this is intended behavior or not, but it looks strange for me.

I just tried the server middleware example in from the doc and jaeger doesn't find the service and requests.

The client example on this repo work :)

Opentelemetry v1.19 was released and It would be really lovely to update all crates, not just the Opentelemetry

Hello,

I am not seeing the status code that was returned from the actix-web server span. Is this expected or maybe I have missed something?

Thank you for your work.

I wanna capture these kinds of log (ERROR, INFO, WARN) from the running console, and export them to something like Logstash, but I only find Metrics and Tracing features in the docs.

Hi,
Just wanted to ask if it's possible to add an example of using actix-web-opentelemetry with a pipeline using OTLP.
That will allow people to start quickly and send traces to Observability tools.

Could you release the v3 of this module on cargo please ? 😊

The dependency in Cargo.toml should ideally be:

- actix-web = "3.0.2"
+ actix-web = { version = "3.0.2", default-features = false }

It seems currently the client span won't set span status to error when the http request fails in some ways (5xx, 4xx, etc), so the resulting spans' status are always UNSET. Do you think its reasonable to automatically set the span status according to the http status similar to what golang's contrib net/http do? (https://github.com/open-telemetry/opentelemetry-go-contrib/blob/6c86403/instrumentation/net/http/otelhttp/handler.go#L219)

The spec have this included so at least we have a standard to follow.

open-telemetry/opentelemetry-specification#394 can be determined at build time via CARGO_PKG_VERSION

Hello,

Thanks for this library! We would like to filter requests being traced to exclude healthchecks ones, is there a way to implement a filter determining if a request needs to be traced?

Now that actix-web 3.0 is released, is it possible to upgrade to the dependency to actix-web 3.0?

If i try to use zipkin pipeline instead of jaeger it fail with: Cannot drop a runtime in a context where blocking is not allowed. This happens when a runtime is dropped from within an asynchronous context.

Code:

[dependencies]
actix-web = "3.2"
actix-web-opentelemetry = "0.7"
opentelemetry-zipkin = "0.7"

use actix_web::{get, web, App, HttpServer, Responder};
use actix_web_opentelemetry::RequestTracing;

#[get("/{id}/{name}/index.html")]
async fn index(web::Path((id, name)): web::Path<(u32, String)>) -> impl Responder {
    format!("Hello {}! id:{}", name, id)
}

#[actix_web::main]
async fn main() -> std::io::Result<()> {
    opentelemetry_zipkin::new_pipeline()
            .install()
            .expect("pipeline install error");
    HttpServer::new(|| App::new().service(index).wrap(RequestTracing::new()))
        .bind("127.0.0.1:8080")?
        .run()
        .await
}

This may not be possible but as of today I haven't found a way to access the context created by this middleware in my own middleware. Is this possible or am I missing something. I have tried ordering my middleware registration in all the combinations I could think of.

I would love some guidance on what might be possible today.

Thank you.

Can you enable "process" feature for Prometheus metrics like this?

https://docs.rs/actix-web-prometheus/0.1.2/actix_web_prometheus/

hey 👋 thanks for putting together this crate!

I see you've defined the type RequestTracingMiddleware to be generic in terms of an HttpTextFormat, and was hoping to use this type like RequestTracing does, but with TraceContextPropagator instead of B3Propagator.

In trying to do the above myself, I found that RequestTracingMiddleware is public but not exposed in lib.rs.

Have you given thought to how library users can do this, and what the desired api would be? Ideally I could write something as simple as:

use actix_web_opentelemetry::RequestTracing
...
App::new().wrap(RequestTracing::new(TextContextPropagator::new()))

thanks for everything 🙏

This code (from your example, with 1 small change)

let (_tracer, _uninstall) = opentelemetry_jaeger::new_pipeline()
      .with_agent_endpoint("localhost:6831")
      .install()
      .expect("pipeline install error");

produces this error:

thread 'main' panicked at 'there is no reactor running, must be called from the context of a Tokio 1.x runtime'

In your example, there is no special Tokio runtime created to run this code, and I'd like to avoid it too :)
Could you please give me a hint on how to make it work?

open-telemetry/opentelemetry-specification#393