refactor(monitoring-ops): disable alertmanager

Some downtime is OK for Grafana, since I am using PVC with access mode `ReadWriteOnce` which is a blocker for a rolling update deployment strategy

clusters/bee/argocd-apps/grafana-ops.yaml

@@ -13,10 +13,13 @@ spec:
   source:
     repoURL: https://grafana.github.io/helm-charts
     chart: grafana
-    targetRevision: "6.50.0"
+    targetRevision: "6.50.6"
     helm:
       releaseName: grafana-ops
       values: |
+        deploymentStrategy:
+          type: Recreate
+
         ingress:
           enabled: true
 
@@ -49,12 +52,16 @@ spec:
           datasources.yaml:
             apiVersion: 1
             datasources:
-            - name: Cluster Prometheus
+            - name: Prometheus
+              type: prometheus
+              url: http://ops-prometheus.monitoring-ops:9090
+              access: proxy
+              isDefault: true
+            - name: In-Cluster Prometheus
               type: prometheus
               url: http://monitoring-prometheus.monitoring:9090
               access: proxy
-              isDefault: true
-            - name: Cluster Loki
+            - name: In-Cluster Loki
               type: loki
               url: http://loki-gateway.logging:80
               access: proxy

clusters/bee/argocd-apps/monitoring-ops.yaml

@@ -0,0 +1,68 @@
+apiVersion: argoproj.io/v1alpha1
+kind: Application
+metadata:
+  name: monitoring-ops
+  namespace: argocd
+  finalizers:
+    - resources-finalizer.argocd.argoproj.io
+spec:
+  destination:
+    namespace: monitoring-ops
+    server: https://kubernetes.default.svc
+  project: default
+  source:
+    repoURL: https://prometheus-community.github.io/helm-charts
+    chart: kube-prometheus-stack
+    targetRevision: "44.3.1"
+    helm:
+      releaseName: monitoring-ops
+      skipCrds: true
+      values: |
+        cleanPrometheusOperatorObjectNames: true
+        fullnameOverride: ops
+
+        defaultRules:
+          create: false
+
+        prometheus:
+          podDisruptionBudget:
+            enabled: true
+            minAvailable: 1
+            maxUnavailable: ""
+
+          prometheusSpec:
+            logFormat: json
+            retention: 5d
+            replicas: 2
+
+            storageSpec:
+              volumeClaimTemplate:
+                spec:
+                  storageClassName: hcloud-volumes
+                  accessModes: ["ReadWriteOnce"]
+                  resources:
+                    requests:
+                      storage: 50Gi
+
+        alertmanager:
+          enabled: false
+
+        prometheusOperator:
+          enabled: false
+
+        grafana:
+          enabled: false
+
+        kubeStateMetrics:
+          enabled: false
+
+        nodeExporter:
+          enabled: false
+
+  syncPolicy:
+    automated:
+      prune: true
+      selfHeal: true
+
+    syncOptions:
+    - CreateNamespace=true