Files
helm-charts/charts/postgres-cluster/prometheus_rules/cluster-zone_spread-warning.yaml
Alex Lebens 1ca985edc7
All checks were successful
release-charts-postgres-cluster / release (push) Successful in 18s
rebase this chart on cnpg provided chart
2025-05-13 00:14:16 -05:00

19 lines
894 B
YAML

{{- $alert := "CNPGClusterZoneSpreadWarning" -}}
{{- if not (has $alert .excludeRules) -}}
alert: {{ $alert }}
annotations:
summary: CNPG Cluster instances in the same zone.
description: |-
CloudNativePG Cluster "{{ .namespace }}/{{ .cluster }}" has instances in the same availability zone.
A disaster in one availability zone will lead to a potential service disruption and/or data loss.
runbook_url: https://github.com/cloudnative-pg/charts/blob/main/charts/cluster/docs/runbooks/CNPGClusterZoneSpreadWarning.md
expr: |
{{ .Values.cluster.instances }} > count(count by (label_topology_kubernetes_io_zone) (kube_pod_info{namespace="{{ .namespace }}", pod=~"{{ .podSelector }}"} * on(node,instance) group_left(label_topology_kubernetes_io_zone) kube_node_labels)) < 3
for: 5m
labels:
severity: warning
namespace: {{ .namespace }}
cnpg_cluster: {{ .cluster }}
{{- end -}}