Using OTLP Tracing with DataDog

Traefik Traefik v3 (latest)
1

I'm in the process of testing out Traefik v3 so we can migrate over to it. We currently use the DataDog tracing agent in Traefik 2.11 without any issues although since that's gone in v3, we are migrating it over to OpenTelemetry and utilizing DataDog's OTLP Ingestion. The default endpoint that Traefik uses for OTLP appears to be "http://localhost" although due to our DataDog setup, localhost connectivity doesn't/won't work. Is there a way to have Traefik OTLP to point to the HostIP of the node? We're already passing in an environment variable for the HostIP although since the configuration is passed in at runtime, I'm assuming variable substitution doesn't work.

Our setup details:
K8s: 1.32.1
Traefik: 3.3.4
Traefik Helm Chart: 34.4.1
DataDog Deployment Method: DataDog Operator 2.7.0 via Helm Chart 2.7.5

2

You can define the endpoint for OTLP (doc).

We are using Traefik in Swarm, there some templating can happen with compose before Traefik is started. Check if helm can do the same.

3

Do you have some documentation you could reference on the way you're templating with compose before Traefik starts? Wondering if I can find a way to do the same in Kubernetes since helm templating is only at manifest install time and I'd need something to template it at pod start.

4

My mistake, seems templating is supported only in a few places. Docker compose can set the hostname dynamically, Traefik sets X-Forwarded-Server to the hostname, so this is how we identify the used reverse proxy.