feat(monitoring): align storage to live pvc, add loki datasource and alerts
Match grafana/alertmanager storageClass to live PVCs to avoid immutable-field failures. Add Loki datasource and LokiDown/AlloyDown alerts.
This commit is contained in:
1 parent
34c33697bb
commit
c98ea8b957
3 files changed
+34
-6
No files matched your search
@@ -66,3 +66,21 @@ spec:
|
||||
annotations:
|
||||
summary: "Node memory pressure"
|
||||
description: "Node {{ $labels.instance }} has used more than 90% of memory for 15 minutes."
|
||||
|
||||
- alert: LokiDown
|
||||
expr: kube_statefulset_status_replicas_unavailable{statefulset="loki"} > 0 or kube_deployment_status_replicas_unavailable{deployment="loki-gateway"} > 0
|
||||
for: 10m
|
||||
labels:
|
||||
severity: warning
|
||||
annotations:
|
||||
summary: "Loki is down"
|
||||
description: "Loki in namespace {{ $labels.namespace }} has unavailable replicas for more than 10 minutes. Logs are not queryable."
|
||||
|
||||
- alert: AlloyDown
|
||||
expr: kube_daemonset_status_number_unavailable{daemonset="alloy"} > 0
|
||||
for: 10m
|
||||
labels:
|
||||
severity: warning
|
||||
annotations:
|
||||
summary: "Alloy is down"
|
||||
description: "Alloy DaemonSet in namespace {{ $labels.namespace }} has {{ $value }} unavailable pods for more than 10 minutes. Pod logs are not being shipped to Loki."
|
||||
Reference in new issue
Block a user