Prometheus and AlertManager

The first step is to configure AlertManager to send alerts to Robusta. see Sending Alerts to Robusta

Robusta can run actions in response to any Prometheus alert. For example:

- triggers:
  - on_prometheus_alert:
      alert_name: HostHighCpuLoad
  - node_bash_enricher:
     bash_command: ps aux

This will run the ps aux on the relevant node whenever a HostHighCpuLoad alert fires. The output will be sent to the default sinks.

How it works

Relevant Kubernetes resources are loaded from the alert's metadata. Then Robusta actions are run with those resources as input.

In the example above, the node_cpu_enricher receives the node on which the alert fired.

Limiting when on_prometheus_alert fires

You can limit when the automation runs by applying the following filters to on_prometheus_alert:

alert_name (str)
status (str) = firing

one of "firing", "resolved", or "all"

pod_name_prefix (str)
namespace_prefix (str)
instance_name_prefix (str)

Developing actions

Here is a custom playbook action that runs on Prometheus alerts:

def my_action(alert: PrometheusKubernetesAlert):
    print(f"The alert {alert.alert_name} fired on pod {}")
    print(f"The pod has these processes:", alert.pod.exec("ps aux"))
    print(f"The pod has {len(alert.pod.spec.containers)} containers")

alert.pod is a Kubernetes pod object. It will exist if the Prometheus alert had a pod label and the pod is alive when the playbook runs. There are also node, deployment, and daemonset fields.