add node exporter daemonset (sourcegraph#4197)

Author: ggilmore

base/node-exporter/node-exporter.DaemonSet.yaml

@@ -0,0 +1,108 @@
+apiVersion: apps/v1
+kind: DaemonSet
+metadata:
+  annotations:
+    description: DaemonSet to ensure all nodes run a node-exporter pod.
+    seccomp.security.alpha.kubernetes.io/pod: 'docker/default'
+  labels:
+    deploy: sourcegraph
+    app: node-exporter
+    app.kubernetes.io/component: node-exporter
+  name: node-exporter
+spec:
+  selector:
+    matchLabels:
+      app: node-exporter
+  template:
+    metadata:
+      annotations:
+        description: Collects and exports machine metrics.
+        kubectl.kubernetes.io/default-container: node-exporter
+      labels:
+        deploy: sourcegraph
+        app: node-exporter
+    spec:
+      containers:
+      - name: node-exporter
+        image: index.docker.io/sourcegraph/node-exporter:179720_2022-10-25_4d925e87cfb8@sha256:2d9dcdf0b2226f0c3d550a64d2667710265462350a3ba9ebe37d0302bc64af0f
+        imagePullPolicy: IfNotPresent
+        resources:
+          limits:
+            cpu: '1'
+            memory: 1Gi
+          requests:
+            cpu: 200m
+            memory: 100Mi
+        args:
+          - --web.listen-address=:9100
+          - --path.sysfs=/host/sys
+          - --path.rootfs=/host/root
+          - --path.procfs=/host/proc
+          - --no-collector.wifi
+          - --no-collector.hwmon
+          - --collector.filesystem.ignored-mount-points=^/(dev|proc|sys|var/lib/docker/.+|var/lib/kubelet/pods/.+)($|/)
+          - --collector.netclass.ignored-devices=^(veth.*)$
+          - --collector.netdev.device-exclude=^(veth.*)$
+        env:
+        securityContext:
+          allowPrivilegeEscalation: false
+          readOnlyRootFilesystem: true
+          runAsGroup: 65534
+          runAsUser: 65534
+        volumeMounts:
+        - name: rootfs
+          mountPath: /host/root
+          mountPropagation: HostToContainer
+          readOnly: true
+        - name: sys
+          mountPath: /host/sys
+          mountPropagation: HostToContainer
+          readOnly: true
+        - name: proc
+          mountPath: /host/proc
+          mountPropagation: HostToContainer
+          readOnly: true
+        ports:
+        - name: metrics
+          containerPort: 9100
+          protocol: TCP
+        readinessProbe:
+          failureThreshold: 3
+          httpGet:
+            scheme: HTTP
+            port: metrics
+          initialDelaySeconds: 0
+          periodSeconds: 10
+          successThreshold: 1
+          timeoutSeconds: 1
+        livenessProbe:
+          failureThreshold: 3
+          httpGet:
+            scheme: HTTP
+            port: metrics
+          initialDelaySeconds: 0
+          periodSeconds: 10
+          successThreshold: 1
+          timeoutSeconds: 1
+        terminationMessagePolicy: FallbackToLogsOnError
+      automountServiceAccountToken: false
+      terminationGracePeriodSeconds: 30
+      securityContext:
+        fsGroup: 65534
+        runAsGroup: 65534
+        runAsNonRoot: true
+        runAsUser: 65534
+      nodeSelector:
+      affinity:
+      tolerations:
+      hostPID: true
+      volumes:
+      - name: rootfs
+        hostPath:
+          path: /
+      - name: sys
+        hostPath:
+          path: /sys
+      - name: proc
+        hostPath:
+          path: /proc

base/node-exporter/node-exporter.Service.yaml

@@ -0,0 +1,22 @@
+apiVersion: v1
+kind: Service
+metadata:
+  annotations:
+    description: Prometheus exporter for hardware and OS metrics.
+    url: https://github.com/prometheus/node_exporter
+    prometheus.io/port: "9100"
+    sourcegraph.prometheus/scrape: "true"
+  labels:
+    app.kubernetes.io/component: node-exporter
+    app: node-exporter
+    deploy: sourcegraph
+    sourcegraph-resource-requires: no-cluster-admin
+  name: node-exporter
+spec:
+  ports:
+  - name: metrics
+    port: 9100
+    targetPort: metrics
+  selector:
+    app: node-exporter
+  type: ClusterIP

base/prometheus/prometheus.ConfigMap.yaml

@@ -163,6 +163,11 @@ data:
       - source_labels: [__meta_kubernetes_pod_name]
         action: replace
         target_label: instance
+      # Sourcegraph specific customization. We want to add a label to every 
+      # metric that indicates the node it came from.
+      - source_labels: [__meta_kubernetes_endpoint_node_name]
+        action: replace
+        target_label: nodename
 
     # Example scrape config for probing services via the Blackbox Exporter.
     #
@@ -235,6 +240,11 @@ data:
       - source_labels: [__meta_kubernetes_namespace]
         action: replace
         target_label: ns
+      # Sourcegraph specific customization. We want to add a label to every 
+      # metric that indicates the node it came from.
+      - source_labels: [__meta_kubernetes_endpoint_node_name]
+        action: replace
+        target_label: nodename
 
       metric_relabel_configs:
       # cAdvisor-specific customization. Drop container metrics exported by cAdvisor

overlays/bases/deployments/kustomization.yaml

@@ -54,3 +54,5 @@ resources:
 - base/otel-collector/otel-collector.ConfigMap.yaml
 - base/otel-collector/otel-collector.Deployment.yaml
 - base/otel-collector/otel-collector.Service.yaml
+- base/node-exporter/node-exporter.DaemonSet.yaml
+- base/node-exporter/node-exporter.Service.yaml

overlays/low-resource/kustomization.yaml

@@ -453,5 +453,6 @@ patchesStrategicMerge:
 - cadvisor/delete-ClusterRoleBinding.yaml
 - cadvisor/delete-ServiceAccount.yaml
 - cadvisor/delete-PodSecurityPolicy.yaml
-
+- node-exporter/delete-Daemonset.yaml
+- node-exporter/delete-Service.yaml
 

overlays/low-resource/node-exporter/delete-Daemonset.yaml

@@ -0,0 +1,5 @@
+$patch: delete
+apiVersion: apps/v1
+kind: DaemonSet
+metadata:
+  name: node-exporter

overlays/low-resource/node-exporter/delete-Service.yaml

@@ -0,0 +1,5 @@
+$patch: delete
+apiVersion: v1
+kind: Service
+metadata:
+  name: node-exporter