monitoring-and-troubleshooting.md

Monitoring and troubleshooting

Warning

This documentation covers the legacy mode of ARC (resources in the actions.summerwind.net namespace). If you're looking for documentation on the newer autoscaling runner scale sets, it is available in GitHub Docs. To understand why these resources are considered legacy (and the benefits of using the newer autoscaling runner scale sets), read this discussion (#2775).

Metrics

The controller also exposes Prometheus metrics on a /metrics endpoint. By default this is on port 8443 behind an RBAC proxy.

If needed, the proxy can be disabled in the values.yml file:

metrics:
  serviceAnnotations: {}
  serviceMonitor: false
  serviceMonitorLabels: {}
+ port: 8080
  proxy:
+   enabled: false

If Prometheus is available inside the cluster, then add some podAnnotations to begin scraping the metrics:

podAnnotations:
+ prometheus.io/scrape: "true"
+ prometheus.io/path: /metrics
+ prometheus.io/port: "8080"

Troubleshooting

See troubleshooting guide for solutions to various problems people have run into consistently.