Monitor Traefik v2 Ingress controller with Prometheus by host

Traefik Traefik v2
,
1

Hi! I have a k3s cluster with Traefik v2 as Ingress Controller, I successfully configured Prometheus to scrape Traefik's metrics but the exported metrics are labeled by Ingress resource.

This is a problem because a single Ingress resource can be configured to serve multiple domains, and the name of an Ingress resource isn't ideal to categorize the requests.

As an example, in v1 I got metrics like:

traefik_backend_requests_total{backend="url.of.ingress.com", code="200", container="traefik", endpoint="...", instance="...", job="traefik-prometheus", method="DELETE", namespace="...", pod="...", protocol="http", service="traefik-prometheus"}

And from here I could filter/aggregate metrics in Grafana by host thanks to the backend field.

But in v2, there are no more backends and I get metrics like:

traefik_service_requests_total{code="200", container="traefik", endpoint="...", exported_service="name-of-ingress-resource-80@kubernetes", instance="...", job="traefik-metrics", method="GET", namespace="...", pod="...", protocol="http", service="traefik-metrics"}

That are not useful at all, since I cannot see the destination host, the only field useful for identification is exported_service but again, is no host.

Is there a way to export what I'm looking for?

1 Like
2

I would also like to be able to aggregate in grafana the metrics by the targeted url.
Is there any possibility?

3

When enabling routers label, you may have more information, depending of your dynamic configuration.

It's also possible with data in access.logs to aggregate per target url or path.