search

carlosedp / cluster-monitoring

Cluster monitoring stack for clusters based on Prometheus Operator
MIT License
740 stars 200 forks source link

compatible with HA k3s #65

Closed tjengbudi closed 3 years ago

tjengbudi commented 4 years ago

hello, i run this with HA k3s but the dashboard always show NONE instead of the graph

can this compatible with 2 or more master cluster?

carlosedp commented 4 years ago

I never tested this on K3s HA mode. Improvements are welcome :)

magikmw commented 3 years ago

I'm running cluster-monitoring on an etcd-enabled k3s cluster with 3 control plane nodes without issues atm.

@tjengbudi I had some trouble finding the custom dashboards (they are in a folder called Default in dashboard management, see picture below), but that's just my lack of familiarity with Grafana itself.

obraz

tillik commented 3 years ago

I too am trying this out on k3s and noticed after deployment all dashboards are empty in Grafana. I also get an error saying "Template variables could not be initialized: undefined" and there seems to be only a default datasource for Prometheus on http://prometheus-k8s.monitoring.svc:9090

Any ideas, what could lead to this?

I was basically following this guide (using nginx and metallb instead of traefik and servicelb): https://kauri.io/#deploy-prometheus-and-grafana-to-monitor-a-kube/186a71b189864b9ebc4ef7c8a9f0a6b5/a

magikmw commented 3 years ago

@tillik That guide ommits running plain make between make vendor and make deploy.

No idea otherwise.

carlosedp commented 3 years ago

Good to know it works on K3s with HA. Closing.