Add Traefik Dashboard #797
Conversation
Adam-D-Lewis
commented
Aug 31, 2021
•
Adam-D-Lewis
commented
- Configures Traefik to produce data for Prometheus to Scrape
- Configures Prometheus to scrape data from Traefik
- Adds a grafana dashboard displaying relevant traefik data
…rometheus_grafana
| # This job will scrape from any service with the label app.kubernetes.io/component=traefik-internal-service | ||
| # and the annotation app.kubernetes.io/scrape=true | ||
| - job_name: 'traefik' | ||
|
|
||
| kubernetes_sd_configs: | ||
| - role: service | ||
|
|
||
| relabel_configs: | ||
| - source_labels: [__meta_kubernetes_service_label_app_kubernetes_io_component] | ||
| action: keep | ||
| regex: traefik-internal-service | ||
| - source_labels: [__meta_kubernetes_service_annotation_prometheus_io_scrape] | ||
| action: keep | ||
| regex: true |
There was a problem hiding this comment.
As set up now, this will scrape only k8s services with both labels app.kubernetes.io/component: traefik-internal-service and prometheus.io/scrape: true (see relabel configs).
It is possible to instead scrape all services with the label prometheus.io/scrape: true, but they'll all be included under the same prometheus job. I'm not sure if that's the "right way" to do things in Prometheus given their definition of job which I've pasted below (from here):
In Prometheus terms, an endpoint you can scrape is called an instance, usually corresponding to a single process. A collection of instances with the same purpose, a process replicated for scalability or reliability for example, is called a job.
There was a problem hiding this comment.
@costrouc, your thoughts?
There was a problem hiding this comment.
As set up now, this will scrape only k8s services with both labels app.kubernetes.io/component: traefik-internal-service and prometheus.io/scrape: true (see relabel configs).
I'm not a fan of this since for each new service we'll have to add a new job for scraping.
It is possible to instead scrape all services with the label prometheus.io/scrape: true, but they'll all be included under the same prometheus job.
This is exactly what I'd like to have. I'd like to not have to require development to qhub to require additional modifications to the monitoring configuration.
|
I also had to edit the traefik dashboard a bit since our traefik container (not pod) is called |
|
Asside from requiring two labels this PR looks good to me. I'd like to have anything with |
|
Combined the 4 return code graphs into one
|
|
|
