coreos/container-linux-update-operator

agent: expose metrics

lucab opened this issue · 0 comments

lucab commented

update-agent is typically running on all CL nodes, watching and reacting to update-engine events and annotation. It should expose its status as metrics that can be analyzed and alerted upon. Access to those should be governed by kubernetes RBAC policies.

This is a preliminary list of interesting metrics:

  • go runtime stats
  • update-engine status
  • update-engine last checked time
  • reboot-needed state
  • reboot-in-progress state
  • CL version and channel