Add Thanos to monitoring. just for historical data storage currently, alerting still happens the old fashion way

k8s/monitoring/kustomization.yaml

@@ -6,10 +6,13 @@ resources:
   - ingresses.yaml
   - secrets.yaml
   - grafana-database.yaml
+  - thanos.yaml
+images:
+  - name: quay.io/thanos/thanos
+    newTag: v0.36.0
 secretGenerator:
   - name: additional-scrape-configs
     options:
       disableNameSuffixHash: true
     files:
       - prom-scrape-configs/scrape-configs.yaml
-

k8s/monitoring/thanos.yaml

@@ -0,0 +1,306 @@
+# This file contains all the components for Thanos that aren't configured by kube-prometheus
+---
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: thanos-querier
+spec:
+  selector:
+    matchLabels:
+      app: thanos-querier
+  template:
+    metadata:
+      labels:
+        app: thanos-querier
+      annotations:
+        prometheus.io/scrape: "true"
+        prometheus.io/port: "9090"
+    spec:
+      containers:
+      - name: thanos-querier
+        image: quay.io/thanos/thanos:latest
+        args:
+          - query
+          - --http-address
+          - 0.0.0.0:9090
+          - --endpoint
+          - dnssrv+_grpc._tcp.prometheus-k8s-headless.monitoring.svc
+          - --endpoint
+          - dnssrv+_grpc._tcp.thanos-store.monitoring.svc
+        resources:
+          limits:
+            memory: "128Mi"
+            cpu: "500m"
+        ports:
+        - containerPort: 9090
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: thanos-querier
+spec:
+  selector:
+    app: thanos-querier
+  ports:
+  - port: 9090
+    targetPort: 9090
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: prometheus-k8s-headless
+spec:
+  selector:
+    app.kubernetes.io/component: prometheus
+    app.kubernetes.io/instance: k8s
+    app.kubernetes.io/name: prometheus
+    app.kubernetes.io/part-of: kube-prometheus
+  clusterIP: None
+  ports:
+  - name: grpc
+    port: 10901
+---
+apiVersion: apps/v1
+kind: StatefulSet
+metadata:
+  labels:
+    app: thanos-store
+  name: thanos-store
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: thanos-store
+  serviceName: thanos-store
+  template:
+    metadata:
+      labels:
+        app: thanos-store
+      annotations:
+        prometheus.io/scrape: "true"
+        prometheus.io/port: "10902"
+    spec:
+      affinity:
+        podAntiAffinity:
+          preferredDuringSchedulingIgnoredDuringExecution:
+          - podAffinityTerm:
+              labelSelector:
+                matchExpressions:
+                - key: app.kubernetes.io/name
+                  operator: In
+                  values:
+                  - thanos-store
+                - key: app.kubernetes.io/instance
+                  operator: In
+                  values:
+                  - thanos-store
+              namespaces:
+              - monitoring
+              topologyKey: kubernetes.io/hostname
+            weight: 100
+      containers:
+      - args:
+        - store
+        - --log.level=info
+        - --log.format=logfmt
+        - --data-dir=/var/thanos/store
+        - --grpc-address=0.0.0.0:10901
+        - --http-address=0.0.0.0:10902
+        - --objstore.config=$(OBJSTORE_CONFIG)
+        - --ignore-deletion-marks-delay=24h
+        env:
+        - name: OBJSTORE_CONFIG
+          valueFrom:
+            secretKeyRef:
+              key: thanos.yaml
+              name: thanos-objstore
+        - name: HOST_IP_ADDRESS
+          valueFrom:
+            fieldRef:
+              fieldPath: status.hostIP
+        image: quay.io/thanos/thanos:latest
+        livenessProbe:
+          failureThreshold: 8
+          httpGet:
+            path: /-/healthy
+            port: 10902
+            scheme: HTTP
+          periodSeconds: 30
+          timeoutSeconds: 1
+        name: thanos-store
+        ports:
+        - containerPort: 10901
+          name: grpc
+        - containerPort: 10902
+          name: http
+        readinessProbe:
+          failureThreshold: 20
+          httpGet:
+            path: /-/ready
+            port: 10902
+            scheme: HTTP
+          periodSeconds: 5
+        resources:
+          limits:
+            memory: "128Mi"
+            cpu: "500m"
+        securityContext:
+          allowPrivilegeEscalation: false
+          capabilities:
+            drop:
+            - ALL
+          readOnlyRootFilesystem: true
+          runAsGroup: 65532
+          runAsNonRoot: true
+          runAsUser: 65534
+          seccompProfile:
+            type: RuntimeDefault
+        terminationMessagePolicy: FallbackToLogsOnError
+        volumeMounts:
+        - mountPath: /var/thanos/store
+          name: data
+          readOnly: false
+      securityContext:
+        fsGroup: 65534
+        runAsGroup: 65532
+        runAsNonRoot: true
+        runAsUser: 65534
+        seccompProfile:
+          type: RuntimeDefault
+      terminationGracePeriodSeconds: 120
+      volumes: []
+  volumeClaimTemplates:
+  - metadata:
+      labels:
+        app: thanos-store
+      name: data
+    spec:
+      accessModes:
+      - ReadWriteOnce
+      resources:
+        requests:
+          storage: 10Gi
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: thanos-store
+spec:
+  selector:
+    app: thanos-store
+  ports:
+  - name: grpc
+    port: 10901
+  - name: http
+    port: 10902
+---
+apiVersion: apps/v1
+kind: StatefulSet
+metadata:
+  labels:
+    app: thanos-compact
+  name: thanos-compact
+  namespace: thanos
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: thanos-compact
+  serviceName: thanos-compact
+  template:
+    metadata:
+      labels:
+        app: thanos-compact
+      annotations:
+        prometheus.io/scrape: "true"
+        prometheus.io/port: "10902"
+    spec:
+      affinity:
+        podAntiAffinity:
+          preferredDuringSchedulingIgnoredDuringExecution:
+          - podAffinityTerm:
+              labelSelector:
+                matchExpressions:
+                - key: app.kubernetes.io/name
+                  operator: In
+                  values:
+                  - thanos-compact
+                - key: app.kubernetes.io/instance
+                  operator: In
+                  values:
+                  - thanos-compact
+              namespaces:
+              - thanos
+              topologyKey: kubernetes.io/hostname
+            weight: 100
+      containers:
+      - args:
+        - compact
+        - --wait
+        - --log.level=info
+        - --log.format=logfmt
+        - --objstore.config=$(OBJSTORE_CONFIG)
+        - --data-dir=/var/thanos/compact
+        - --retention.resolution-1h=365d
+        env:
+        - name: OBJSTORE_CONFIG
+          valueFrom:
+            secretKeyRef:
+              key: thanos.yaml
+              name: thanos-objstore
+        - name: HOST_IP_ADDRESS
+          valueFrom:
+            fieldRef:
+              fieldPath: status.hostIP
+        image: quay.io/thanos/thanos:v0.30.2
+        imagePullPolicy: IfNotPresent
+        livenessProbe:
+          failureThreshold: 4
+          httpGet:
+            path: /-/healthy
+            port: 10902
+            scheme: HTTP
+          periodSeconds: 30
+        name: thanos-compact
+        ports:
+        - containerPort: 10902
+          name: http
+        readinessProbe:
+          failureThreshold: 20
+          httpGet:
+            path: /-/ready
+            port: 10902
+            scheme: HTTP
+          periodSeconds: 5
+        resources:
+          limits:
+            cpu: 500m
+            memory: 500Mi
+          requests:
+            cpu: 200m
+            memory: 100Mi
+        terminationMessagePolicy: FallbackToLogsOnError
+        volumeMounts:
+        - mountPath: /var/thanos/compact
+          name: data
+          readOnly: false
+      nodeSelector:
+        kubernetes.io/os: linux
+      securityContext:
+        fsGroup: 65534
+        runAsGroup: 65532
+        runAsNonRoot: true
+        runAsUser: 65534
+        seccompProfile:
+          type: RuntimeDefault
+      terminationGracePeriodSeconds: 120
+      volumes: []
+  volumeClaimTemplates:
+  - metadata:
+      name: data
+    spec:
+      accessModes:
+      - ReadWriteOnce
+      resources:
+        requests:
+          storage: 10Gi

k8s/operators/kube-prometheus/alertmanager-patches.yaml

@@ -1,3 +1,14 @@
 - op: add
   path: /spec/externalUrl
   value: https://alertmanager.k8s.home.finn.io
+- op: replace
+  path: /spec/resources/requests/memory
+  value: 40Mi
+- op: add
+  path: /spec/storage
+  value:
+    volumeClaimTemplate:
+      spec:
+        resources:
+          requests:
+            storage: 1Gi

k8s/operators/kube-prometheus/grafana/datasources.yaml

@@ -7,8 +7,8 @@
       "name": "prometheus",
       "orgId": 1,
       "type": "prometheus",
-      "url": "http://prometheus-k8s.monitoring.svc:9090",
+      "url": "http://thanos-querier.monitoring.svc:9090",
-      "version": 1
+      "version": 2
     },
     {
       "access": "proxy",

k8s/operators/kube-prometheus/kustomization.yaml

@@ -14,11 +14,11 @@ patches:
   - path: network-policies-patch.yaml
     target:
       kind: NetworkPolicy
-      name: prometheus-k8s
+      name: alertmanager-main
-  - path: network-policies-patch.yaml
+  - path: prometheus-network-policy-patch.yaml
     target:
       kind: NetworkPolicy
-      name: alertmanager-main
+      name: prometheus-k8s
   - path: prometheus-patch.yaml
     target:
       kind: Prometheus
@@ -43,6 +43,10 @@ patches:
     target:
       kind: PrometheusRule
       name: node-exporter-rules
+  - path: prometheus-operator-deployment-patches.yaml
+    target:
+      kind: Deployment
+      name: prometheus-operator
 secretGenerator:
   - name: grafana-config
     namespace: monitoring

k8s/operators/kube-prometheus/prometheus-network-policy-patch.yaml

@@ -0,0 +1,22 @@
+- op: add
+  path: /spec/ingress/0/from/-
+  value:
+    podSelector:
+      matchLabels:
+        app.kubernetes.io/name: traefik
+    namespaceSelector:
+      matchLabels:
+        kubernetes.io/metadata.name: kube-system
+- op: add
+  path: /spec/ingress/-
+  value:
+    from:
+      - podSelector:
+          matchLabels:
+            app: thanos-querier
+        namespaceSelector:
+          matchLabels:
+            kubernetes.io/metadata.name: monitoring
+    ports:
+      - port: 10901
+        protocol: TCP

k8s/operators/kube-prometheus/prometheus-operator-deployment-patches.yaml

@@ -0,0 +1,3 @@
+- op: add
+  path: /spec/template/spec/containers/0/args/-
+  value: --config-reloader-cpu-limit=500m

k8s/operators/kube-prometheus/prometheus-patch.yaml

@@ -10,3 +10,18 @@
 - op: add
   path: /spec/externalUrl
   value: https://prometheus.k8s.home.finn.io
+- op: add
+  path: /spec/thanos
+  value:
+    image: quay.io/thanos/thanos:v0.36.0
+    objectStorageConfig:
+      key: thanos.yaml
+      name: thanos-objstore
+- op: add
+  path: /spec/storage
+  value:
+    volumeClaimTemplate:
+      spec:
+        resources:
+          requests:
+            storage: 20Gi

k8s/system-upgrade-controller/plan.yaml

@@ -52,9 +52,9 @@ spec:
     - key: node-role.kubernetes.io/control-plane
       operator: DoesNotExist
   tolerations:
-    - key: "rtlsdr"
+    - key: rtlsdr
       value: "true"
-      effect: "NoSchedule"
+      effect: NoSchedule
   prepare:
     args:
     - prepare