[cluster]: don't fire HA alerts on replica clusters

Author: Allex1

charts/cluster/Chart.yaml

@@ -22,7 +22,7 @@ description: Deploys and manages a CloudNativePG cluster and its associated reso
 kubeVersion: ">=1.29.0-0"
 icon: https://raw.githubusercontent.com/cloudnative-pg/artwork/main/cloudnativepg-logo.svg
 type: application
-version: 0.7.0
+version: 0.7.1
 sources:
   - https://github.com/cloudnative-pg/charts
 keywords:

charts/cluster/README.md

@@ -1,6 +1,6 @@
 # cluster
 
-![Version: 0.7.0](https://img.shields.io/badge/Version-0.7.0-informational?style=flat-square) ![Type: application](https://img.shields.io/badge/Type-application-informational?style=flat-square)
+![Version: 0.7.1](https://img.shields.io/badge/Version-0.7.1-informational?style=flat-square) ![Type: application](https://img.shields.io/badge/Type-application-informational?style=flat-square)
 
 > **Warning**
 > ### This chart is under active development.

charts/cluster/docs/runbooks/CNPGClusterHACritical.md

@@ -12,6 +12,11 @@ instances. The replaced instance may need some time to catch-up with the cluster
 This alarm will be always triggered if your cluster is configured to run with only 1 instance. In this case you
 may want to silence it.
 
+This alert does not fire for replica clusters. On a replica cluster the designated primary runs a WAL receiver
+(streaming from the source cluster), which lowers the computed standby count by one. The rule is gated on
+`cnpg_collector_replica_mode == 0`, which yields an empty result on replica clusters, so the alert only fires on
+primary clusters.
+
 Impact
 ------
 

charts/cluster/docs/runbooks/CNPGClusterHAWarning.md

@@ -9,6 +9,11 @@ The `CNPGClusterHAWarning` alert is triggered when the CloudNativePG cluster rea
 This alarm will be always triggered if your cluster is configured to run with less than `3` instances. In this case you
 may want to silence it.
 
+This alert does not fire for replica clusters. On a replica cluster the designated primary runs a WAL receiver
+(streaming from the source cluster), which lowers the computed standby count by one and would otherwise fire this alert
+permanently. The rule is gated on `cnpg_collector_replica_mode == 0`, which yields an empty result on replica clusters,
+so the alert only fires on primary clusters.
+
 Impact
 ------
 

charts/cluster/prometheus_rules/cluster-ha-critical.yaml

@@ -18,6 +18,8 @@ annotations:
   runbook_url: https://github.com/cloudnative-pg/charts/blob/main/charts/cluster/docs/runbooks/CNPGClusterHACritical.md
 expr: |
   max by (job) (cnpg_pg_replication_streaming_replicas{namespace="{{ .namespace }}"} - cnpg_pg_replication_is_wal_receiver_up{namespace="{{ .namespace }}"}) < 1
+    and on (job)
+  (max by (job) (cnpg_collector_replica_mode{namespace="{{ .namespace }}"}) == 0)
 for: 5m
 labels:
   severity: critical

charts/cluster/prometheus_rules/cluster-ha-warning.yaml

@@ -16,6 +16,8 @@ annotations:
   runbook_url: https://github.com/cloudnative-pg/charts/blob/main/charts/cluster/docs/runbooks/CNPGClusterHAWarning.md
 expr: |
   max by (job) (cnpg_pg_replication_streaming_replicas{namespace="{{ .namespace }}"} - cnpg_pg_replication_is_wal_receiver_up{namespace="{{ .namespace }}"}) < 2
+    and on (job)
+  (max by (job) (cnpg_collector_replica_mode{namespace="{{ .namespace }}"}) == 0)
 for: 5m
 labels:
   severity: warning