mg4/k8s/grafana.yaml

# Grafana for the dashboards. The InfluxDB datasource and the three MG4 dashboards are
# provisioned from k8s/grafana/ (ConfigMaps generated by kustomization.yaml); the dashboards
# are read-only in the UI — change the JSON in git instead.
#
# Grafana reads InfluxDB with a token that can only read the mg4 bucket. InfluxDB generates
# every token but the admin one (a chosen value is ignored), so the init container makes it at
# each start: it alone gets the admin token, creates a fresh read-only token, deletes the one
# from the previous start, and leaves only the read-only token in a memory-backed volume for
# Grafana. Nothing to do by hand, and the Grafana process never sees the admin token.
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: grafana-data
  namespace: mg4
  labels:
    app.kubernetes.io/name: grafana
    app.kubernetes.io/part-of: mg4
spec:
  accessModes: ["ReadWriteOnce"]
  # No storageClassName: the cluster's default class is used.
  resources:
    requests:
      storage: 1Gi
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: grafana
  namespace: mg4
  labels:
    app.kubernetes.io/name: grafana
    app.kubernetes.io/part-of: mg4
spec:
  replicas: 1
  strategy:
    type: Recreate # the SQLite database on the volume allows one writer
  selector:
    matchLabels:
      app.kubernetes.io/name: grafana
  template:
    metadata:
      labels:
        app.kubernetes.io/name: grafana
        app.kubernetes.io/part-of: mg4
    spec:
      securityContext:
        runAsNonRoot: true
        runAsUser: 472
        runAsGroup: 472
        fsGroup: 472
        seccompProfile:
          type: RuntimeDefault
      initContainers:
        - name: influx-read-token
          image: docker.io/library/influxdb:2.7.12-alpine
          command:
            - sh
            - -c
            - |
              set -eu
              desc="grafana read-only (made by the Grafana pod at start)"
              export INFLUX_CONFIGS_PATH=/tmp/influx-configs
              bucket=$(influx bucket find --name mg4 --hide-headers | cut -f1)
              old=$(influx auth list --hide-headers | awk -F'\t' -v d="$desc" '$2 == d { print $1 }')
              token=$(influx auth create --description "$desc" --read-bucket "$bucket" --hide-headers | cut -f3)
              [ -n "$token" ] || { echo "no token created" >&2; exit 1; }
              printf '%s' "$token" > /run/influx/token
              for id in $old; do influx auth delete --id "$id" > /dev/null; done
              echo "read-only token for Grafana ready"
          env:
            - name: INFLUX_HOST
              value: http://influxdb.mg4.svc:8086
            - name: INFLUX_ORG
              value: mg4
            - name: INFLUX_TOKEN
              valueFrom:
                secretKeyRef:
                  name: influxdb-auth
                  key: admin-token
          volumeMounts:
            - name: influx-token
              mountPath: /run/influx
          resources:
            requests:
              cpu: 10m
              memory: 32Mi
            limits:
              memory: 128Mi
          securityContext:
            allowPrivilegeEscalation: false
            capabilities:
              drop: ["ALL"]
      containers:
        - name: grafana
          image: docker.io/grafana/grafana:13.2.3
          ports:
            - name: http
              containerPort: 3000
          env:
            - name: GF_SERVER_ROOT_URL
              value: https://car-grafana.k8s.stinnesbeck.com
            - name: GF_SECURITY_ADMIN_USER
              value: admin
            - name: GF_SECURITY_ADMIN_PASSWORD
              valueFrom:
                secretKeyRef:
                  name: grafana-admin
                  key: admin-password
            - name: GF_USERS_ALLOW_SIGN_UP
              value: "false"
            - name: GF_AUTH_ANONYMOUS_ENABLED
              value: "false"
            - name: GF_ANALYTICS_REPORTING_ENABLED
              value: "false"
            - name: GF_ANALYTICS_CHECK_FOR_UPDATES
              value: "false"
            # Read by datasources.yaml; the token is the file the init container wrote.
            - name: INFLUX_URL
              value: http://influxdb.mg4.svc:8086
          volumeMounts:
            - name: data
              mountPath: /var/lib/grafana
            - name: datasources
              mountPath: /etc/grafana/provisioning/datasources
            - name: dashboard-provider
              mountPath: /etc/grafana/provisioning/dashboards
            - name: dashboards
              mountPath: /var/lib/grafana/dashboards/mg4
            - name: influx-token
              mountPath: /run/influx
              readOnly: true
          # The first start creates and migrates Grafana's database: measured 44 s at 0.5 CPU
          # and 3 minutes at 0.1 CPU. The startup probe allows up to 10 minutes for that, and
          # liveness/readiness only begin once it succeeded, so a slow first start is never
          # killed mid-migration (which would restart it slowly again, in a loop).
          startupProbe:
            httpGet:
              path: /api/health
              port: http
            periodSeconds: 10
            timeoutSeconds: 5
            failureThreshold: 60
          readinessProbe:
            httpGet:
              path: /api/health
              port: http
            periodSeconds: 10
            timeoutSeconds: 5
          livenessProbe:
            httpGet:
              path: /api/health
              port: http
            periodSeconds: 30
            timeoutSeconds: 5
            failureThreshold: 3
          # About 406 MiB in steady state in the cluster: request what it uses, and leave room
          # for long-range dashboard queries before the limit OOM-kills it.
          resources:
            requests:
              cpu: 250m
              memory: 512Mi
            limits:
              memory: 1Gi
          securityContext:
            allowPrivilegeEscalation: false
            capabilities:
              drop: ["ALL"]
      volumes:
        - name: data
          persistentVolumeClaim:
            claimName: grafana-data
        - name: datasources
          configMap:
            name: grafana-datasources
        - name: dashboard-provider
          configMap:
            name: grafana-dashboard-provider
        - name: dashboards
          configMap:
            name: grafana-dashboards
        # The read-only InfluxDB token, in memory only; made anew at every start.
        - name: influx-token
          emptyDir:
            medium: Memory
            sizeLimit: 1Mi
---
apiVersion: v1
kind: Service
metadata:
  name: grafana
  namespace: mg4
  labels:
    app.kubernetes.io/name: grafana
    app.kubernetes.io/part-of: mg4
spec:
  selector:
    app.kubernetes.io/name: grafana
  ports:
    - name: http
      port: 80
      targetPort: http