Como ativar métricas personalizadas definidas pelo usuário para escalonamento automático de pod horizontal

Este documento descreve como configurar métricas definidas pelo usuário para escalonamento automático horizontal de pods (HPA) no Google Distributed Cloud.

Esta página é destinada a administradores, arquitetos e operadores que otimizam a arquitetura e os recursos de sistemas para garantir o menor custo total de propriedade para a empresa ou unidade de negócios, além de planejar as necessidades de capacidade e infraestrutura. Para saber mais sobre papéis comuns e tarefas de exemplo referenciados no conteúdo do Google Cloud , consulte Tarefas e funções de usuário comuns do GKE.

Implantar o Prometheus e o Metrics Adapter

Nesta seção, você implanta o Prometheus para extrair métricas definidas pelo usuário e o prometheus-adapter para atender à API Kubernetes Custom Metrics com o Prometheus como no back-end.

Salve os seguintes manifestos de implantação em um arquivo chamado custom-metrics-adapter.yaml.

Conteúdo do arquivo de manifesto para o Prometheus e o Metrics Adapter

# Copyright 2018 Google Inc
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#      http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

apiVersion: v1
kind: ServiceAccount
metadata:
  name: stackdriver-prometheus
  namespace: kube-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: stackdriver-prometheus
  namespace: kube-system
rules:
- apiGroups:
  - ""
  resources:
  - nodes
  - services
  - endpoints
  - pods
  verbs:
  - get
  - list
  - watch
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: stackdriver-prometheus
  namespace: kube-system
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: stackdriver-prometheus
subjects:
- kind: ServiceAccount
  name: stackdriver-prometheus
  namespace: kube-system
---
apiVersion: v1
kind: Service
metadata:
  name: stackdriver-prometheus-app
  namespace: kube-system
  labels:
    app: stackdriver-prometheus-app
spec:
  clusterIP: "None"
  ports:
    - name: http
      port: 9090
      protocol: TCP
      targetPort: 9090
  sessionAffinity: ClientIP
  selector:
    app: stackdriver-prometheus-app
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: stackdriver-prometheus-app
  namespace: kube-system
  labels:
    app: stackdriver-prometheus-app
spec:
  replicas: 1
  selector:
    matchLabels:
      app: stackdriver-prometheus-app
  template:
    metadata:
      labels:
        app: stackdriver-prometheus-app
    spec:
      serviceAccount: stackdriver-prometheus
      containers:
      - name: prometheus-server
        image: prom/prometheus:v2.45.0
        args:
        - "--config.file=/etc/prometheus/config/prometheus.yaml"
        - "--storage.tsdb.path=/data"
        - "--storage.tsdb.retention.time=2h"
        ports:
        - name: prometheus
          containerPort: 9090
        readinessProbe:
          httpGet:
            path: /-/ready
            port: 9090
          periodSeconds: 5
          timeoutSeconds: 3
          # Allow up to 10m on startup for data recovery
          failureThreshold: 120
        livenessProbe:
          httpGet:
            path: /-/healthy
            port: 9090
          periodSeconds: 5
          timeoutSeconds: 3
          failureThreshold: 6
        resources:
          requests:
            cpu: 250m
            memory: 500Mi
        volumeMounts:
        - name: config-volume
          mountPath: /etc/prometheus/config
        - name: stackdriver-prometheus-app-data
          mountPath: /data
      volumes:
      - name: config-volume
        configMap:
          name: stackdriver-prometheus-app
      - name: stackdriver-prometheus-app-data
        emptyDir: {}
      terminationGracePeriodSeconds: 300
      nodeSelector:
        kubernetes.io/os: linux
---
apiVersion: v1
data:
  prometheus.yaml: |
    global:
      scrape_interval: 1m
    rule_files:
    - /etc/config/rules.yaml
    - /etc/config/alerts.yaml
    scrape_configs:
    - job_name: prometheus-io-endpoints
      kubernetes_sd_configs:
      - role: endpoints
      relabel_configs:
      - action: keep
        regex: true
        source_labels:
        - __meta_kubernetes_service_annotation_prometheus_io_scrape
      - action: replace
        regex: (.+)
        source_labels:
        - __meta_kubernetes_service_annotation_prometheus_io_path
        target_label: __metrics_path__
      - action: replace
        regex: (https?)
        source_labels:
        - __meta_kubernetes_service_annotation_prometheus_io_scheme
        target_label: __scheme__
      - action: replace
        regex: ([^:]+)(?::\d+)?;(\d+)
        replacement: $1:$2
        source_labels:
        - __address__
        - __meta_kubernetes_service_annotation_prometheus_io_port
        target_label: __address__
      - action: replace
        source_labels:
        - __meta_kubernetes_namespace
        target_label: namespace
      - action: replace
        source_labels:
        - __meta_kubernetes_pod_name
        target_label: pod
      - action: keep
        regex: (.+)
        source_labels:
        - __meta_kubernetes_endpoint_port_name
    - job_name: prometheus-io-services
      kubernetes_sd_configs:
      - role: service
      metrics_path: /probe
      params:
        module:
        - http_2xx
      relabel_configs:
      - action: replace
        source_labels:
        - __address__
        target_label: __param_target
      - action: replace
        replacement: blackbox
        target_label: __address__
      - action: keep
        regex: true
        source_labels:
        - __meta_kubernetes_service_annotation_prometheus_io_probe
      - action: replace
        source_labels:
        - __meta_kubernetes_namespace
        target_label: namespace
      - action: replace
        source_labels:
        - __meta_kubernetes_pod_name
        target_label: pod
    - job_name: prometheus-io-pods
      kubernetes_sd_configs:
      - role: pod
      relabel_configs:
      - action: keep
        regex: true
        source_labels:
        - __meta_kubernetes_pod_annotation_prometheus_io_scrape
      - action: replace
        regex: (.+)
        source_labels:
        - __meta_kubernetes_pod_annotation_prometheus_io_path
        target_label: __metrics_path__
      - action: replace
        regex: ([^:]+)(?::\d+)?;(\d+)
        replacement: $1:$2
        source_labels:
        - __address__
        - __meta_kubernetes_pod_annotation_prometheus_io_port
        target_label: __address__
      - action: replace
        source_labels:
        - __meta_kubernetes_namespace
        target_label: namespace
      - action: replace
        source_labels:
        - __meta_kubernetes_pod_name
        target_label: pod
kind: ConfigMap
metadata:
  name: stackdriver-prometheus-app
  namespace: kube-system
---

# The main section of custom metrics adapter.
kind: ServiceAccount
apiVersion: v1
metadata:
  name: custom-metrics-apiserver
  namespace: kube-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: custom-metrics:system:auth-delegator
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: system:auth-delegator
subjects:
- kind: ServiceAccount
  name: custom-metrics-apiserver
  namespace: kube-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: custom-metrics-server-resources
rules:
- apiGroups:
  - custom.metrics.k8s.io
  resources: ["*"]
  verbs: ["*"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: custom-metrics-resource-reader
rules:
- apiGroups:
  - ""
  resources:
  - nodes
  - namespaces
  - pods
  - services
  verbs:
  - get
  - watch
  - list
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: custom-metrics-resource-reader
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: custom-metrics-resource-reader
subjects:
- kind: ServiceAccount
  name: custom-metrics-apiserver
  namespace: kube-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: RoleBinding
metadata:
  name: custom-metrics-auth-reader
  namespace: kube-system
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: Role
  name: extension-apiserver-authentication-reader
subjects:
- kind: ServiceAccount
  name: custom-metrics-apiserver
  namespace: kube-system
---
apiVersion: v1
kind: ConfigMap
metadata:
  name: adapter-config
  namespace: kube-system
data:
  config.yaml: |
    rules:
    default: false
      # fliter all metrics
    - seriesQuery: '{pod=~".+"}'
      seriesFilters: []
      resources:
        # resource name is mapped as it is. ex. namespace -> namespace
        template: <<.Resource>>
      name:
        matches: ^(.*)$
        as: ""
      # Aggregate metric on resource level
      metricsQuery: sum(<<.Series>>{<<.LabelMatchers>>}) by (<<.GroupBy>>)
---
apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: custom-metrics-apiserver
  name: custom-metrics-apiserver
  namespace: kube-system
spec:
  replicas: 1
  selector:
    matchLabels:
      app: custom-metrics-apiserver
  template:
    metadata:
      labels:
        app: custom-metrics-apiserver
      name: custom-metrics-apiserver
    spec:
      serviceAccountName: custom-metrics-apiserver
      containers:
      - name: custom-metrics-apiserver
        resources:
          requests:
            cpu: 15m
            memory: 20Mi
          limits:
            cpu: 100m
            memory: 150Mi
        image: registry.k8s.io/prometheus-adapter/prometheus-adapter:v0.11.0
        args:
        - /adapter
        - --cert-dir=/var/run/serving-cert
        - --secure-port=6443
        - --prometheus-url=http://stackdriver-prometheus-app.kube-system.svc:9090/
        - --metrics-relist-interval=1m
        - --config=/etc/adapter/config.yaml
        ports:
        - containerPort: 6443
        volumeMounts:
        - name: serving-cert
          mountPath: /var/run/serving-cert
        - mountPath: /etc/adapter/
          name: config
          readOnly: true
      nodeSelector:
        kubernetes.io/os: linux
      volumes:
      - name: serving-cert
        emptyDir:
          medium: Memory
      - name: config
        configMap:
          name: adapter-config
---
apiVersion: v1
kind: Service
metadata:
  name: custom-metrics-apiserver
  namespace: kube-system
spec:
  ports:
  - port: 443
    targetPort: 6443
  selector:
    app: custom-metrics-apiserver
---
apiVersion: apiregistration.k8s.io/v1
kind: APIService
metadata:
  name: v1beta1.custom.metrics.k8s.io
spec:
  service:
    name: custom-metrics-apiserver
    namespace: kube-system
  group: custom.metrics.k8s.io
  version: v1beta1
  insecureSkipTLSVerify: true
  groupPriorityMinimum: 100
  versionPriority: 100
---
apiVersion: apiregistration.k8s.io/v1
kind: APIService
metadata:
  name: v1beta2.custom.metrics.k8s.io
spec:
  service:
    name: custom-metrics-apiserver
    namespace: kube-system
  group: custom.metrics.k8s.io
  version: v1beta2
  insecureSkipTLSVerify: true
  groupPriorityMinimum: 100
  versionPriority: 100
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: hpa-controller-custom-metrics
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: custom-metrics-server-resources
subjects:
- kind: ServiceAccount
  name: horizontal-pod-autoscaler
  namespace: kube-system

Crie a implantação e o serviço:

kubectl --kubeconfig USER_CLUSTER_KUBECONFIG apply -f custom-metrics-adapter.yaml

A próxima etapa é anotar o aplicativo do usuário para a coleta de métricas.

Anotar um aplicativo de usuário para coletar métricas

Para anotar um aplicativo de usuário a ser extraído e os registros enviados ao Cloud Monitoring, adicione annotations correspondentes aos metadados do serviço, pod e endpoints.

  metadata:
    name: "example-monitoring"
    namespace: "default"
    annotations:
      prometheus.io/scrape: "true"
      prometheus.io/path: "" - Overriding metrics path (default "/metrics")

Implantar um aplicativo de usuário de amostra

Nesta seção, você implantará um aplicativo de amostra com registros e métricas compatíveis com o Prometheus.

  1. Salve os seguintes manifestos de Serviço e Implantação em um arquivo chamado my-app.yaml. Observe que o Serviço tem a anotação prometheus.io/scrape: "true":

    kind: Service
apiVersion: v1
metadata:
  name: "example-monitoring"
  namespace: "default"
  annotations:
    prometheus.io/scrape: "true"
spec:
  selector:
    app: "example-monitoring"
  ports:
    - name: http
      port: 9090
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: "example-monitoring"
  namespace: "default"
  labels:
    app: "example-monitoring"
spec:
  replicas: 1
  selector: 
    matchLabels:
      app: "example-monitoring"
  template: 
    metadata: 
      labels:
        app: "example-monitoring"
    spec:
      containers:
      - image: gcr.io/google-samples/prometheus-dummy-exporter:v0.2.0
        name: prometheus-example-exporter
        command:
        - ./prometheus-dummy-exporter
        args:
        - --metric-name=example_monitoring_up
        - --metric-value=1
        - --port=9090
        resources:
          requests:
            cpu: 100m

  2. Crie a implantação e o serviço:

    kubectl --kubeconfig USER_CLUSTER_KUBECONFIG apply -f my-app.yaml

Usar as métricas personalizadas no HPA

Implante o objeto do HPA para usar a métrica exposta na etapa anterior. Saiba mais sobre os diversos tipos de métricas personalizadas em Como fazer escalonamento automático em métricas diversas e personalizadas.

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: example-monitoring-hpa
  namespace: default
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: example-monitoring
  minReplicas: 1
  maxReplicas: 5
  metrics:
  - type: Pods
    pods:
      metric:
        name: example_monitoring_up
      target:
        type: AverageValue
        averageValue: 20

A métrica do tipo pods tem um seletor de métricas padrão para os rótulos dos pods de destino, e é assim que o kube-controller-manager funciona. Neste exemplo, você consulta a métrica example_monitoring_up com um seletor de {matchLabels: {app: example-monitoring}}, porque eles estão disponíveis nos pods de destino. Qualquer outro seletor especificado será adicionado à lista. Para evitar o seletor padrão, remova todos os rótulos do pod de destino ou use a métrica "Tipo de objeto".

Verificar se as métricas do aplicativo definidas pelo usuário são usadas pelo HPA

Verifique se as métricas do aplicativo definidas pelo usuário são usadas pelo HPA:

kubectl --kubeconfig=USER_CLUSTER_KUBECONFIG describe hpa example-monitoring-hpa

A saída será semelhante ao seguinte:

  Name:               example-monitoring-hpa
  Namespace:          default
  Labels:             
  Annotations:        autoscaling.alpha.kubernetes.io/conditions:
                        [{"type":"AbleToScale","status":"True","lastTransitionTime":"2023-08-23T22:07:24Z","reason":"ReadyForNewScale","message":"recommended size...
                      autoscaling.alpha.kubernetes.io/current-metrics: [{"type":"Pods","pods":{"metricName":"example_monitoring_up","currentAverageValue":"1"}}]
                      autoscaling.alpha.kubernetes.io/metrics: [{"type":"Pods","pods":{"metricName":"example_monitoring_up","targetAverageValue":"20"}}]
  CreationTimestamp:  Wed, 23 Aug 2023 22:07:09 +0000
  Reference:          Deployment/example-monitoring
  Min replicas:       1
  Max replicas:       5
  Deployment pods:    1 current / 1 desired

Custos

O uso de métricas personalizadas para o HPA não gera cobranças extras do Cloud Monitoring. Os pods para ativar métricas personalizadas consomem CPU e memória adicionais com base na quantidade de métricas coletadas.