Prometheus dynamic alert rule.

[akshay sharma]

In my setup, Prometheus is monitoring multiple nodes simultaneously, say, x,y z.

I want to raise alerts, once cpu utilization exceeds "a" value for each of the nodes.

Below is the alert rule.

alert: cpu_utilization
expr: 100 - (avg by(instance) (irate(node_cpu_seconds_total{job="--",mode="idle"}[5m])) * 100) > a
labels:
  severity: critical
annotations:
  summary: CPU utilization has crossed a%

QUERY: 

1) How can I use the same rule for multiple nodes/jobs? Is there any way to update job names dynamically? As I want to avoid multiple alert rules for each job. 

Thanks,

[Stuart Clark]

Just remove the job selector from your query and it will alert for all jobs

--
Sent from my Android device with K-9 Mail. Please excuse my brevity.