kube-prometheus-stack.yaml

# runbooks for alerts in kube-prometheus-stack
# the AI will follow the instructions inside these runbooks to investigate alerts!
# please feel free to open PRs adding your own runboks
runbooks:
  - match:
      issue_name: "(KubeSchedulerDown)|(KubeControllerManagerDown)"
    instructions: >
      Check if the cluster is a managed cluster like EKS by fetching nodes and looking at their labels.
      If so, tell the user this is likely a known false positive in the kube-prometheus-stack alert because Prometheus can't scrape the scheduler which is managed by the cloud provider.
      On the other hand, if this is a self-managed Kubernetes, either the scheduler is really down (unlikely) or it is running but Prometheus can't scrape it.