We use the open telemetry exporter in Traefik to Signoz. but it adds a lot of internal spans so it is hard to track the actual flows from the entrypoint to the backend machine. Does anyone have any experience with this? To make this look useable? Ive tried filtering out internal, but it adds a lot of mess and unrelated span id's

Suggest posting an screenshot of the trace, highlighting the noise, and provide snippets of what filtering you've already enabled (and if you're doing it in the traefik config or the Signoz collector). For a particular trace, you can also use a query with

group by name

and the Table View to quantify the noise. Is Traefik the correct place to enable an exporter if you're more concerned with application than proxy performance? In my case there's lots of redundant spans from Node's HTTP layer, Express middleware, then the NestJS framework before it reaches our controllers so I disable some of these auto-instrumentation.