Aller au contenu

PrometheusNotConnectedToAlertmanagers#

This alert is triggered when Prometheus instances have no (or not enough) connections to Alertmanager.

There are two types of alerts depending on the platforms : Caascad or NGOT .

  • Custom alert only
  • Firing when Prometheus is not connected to all members of the Alertmanager cluster
  • For Prometheus cloud-app, we also take into account the Alertmanagers of the customers
  • Custom and upstream alert
  • Custom alert evaluated by the Prometheus central : firing when Prometheus cluster or central is not connected to all members of the Alertmanager cluster
  • Upstream alert evaluated by the Prometheus cluster : firing when Prometheus cluster or central is not connected to any alertmanager

Troubleshooting#

Check Prometheus configuration#

kubectl get prometheus -n <prometheus_namespace> # Notice the namespace label in alert

Check that this section is present : 

additionalAlertManagerConfigs:
    key: <secret_key>
    name: <secret_name>

Chech the presence of secret : 

kubectl get secret -n <prometheus_namespace> <secret_name> -o yaml