Configurar a observabilidade do GKE Dataplane V2

Autopilot Padrão

Nesta página, mostramos como configurar clusters do Google Kubernetes Engine (GKE) com a observabilidade do GKE Dataplane V2, a partir das versões 1.28 ou mais recentes do GKE. Para mais informações sobre os benefícios e os requisitos da observabilidade do GKE Dataplane V2, consulte Sobre a observabilidade do GKE Dataplane V2.

Antes de começar

Antes de começar, verifique se você realizou as tarefas a seguir:

Ativar a API Google Kubernetes Engine.

Ativar a API Google Kubernetes Engine

Se você quiser usar a Google Cloud CLI para essa tarefa, instale e, em seguida, inicialize a CLI gcloud. Se você instalou a CLI gcloud anteriormente, instale a versão mais recente executando gcloud components update.
Observação: para instalações atuais da CLI gcloud, defina as propriedades compute/region e compute/zone. Definindo os locais padrão, você evita erros como este na CLI gcloud: One of [--zone, --region] must be supplied: Please specify location.

Configurar métricas do GKE Dataplane V2

Para coletar métricas, configure as métricas do GKE Dataplane V2. É possível configurar métricas do GKE Dataplane V2 ao criar um cluster ou atualizar um cluster em execução com o GKE Dataplane V2. É possível ativar ou desativar as métricas do GKE Dataplane V2 usando a CLI gcloud.

Recomendamos ativar as métricas do GKE Dataplane V2 e o Google Cloud Managed Service para Prometheus no cluster do GKE. Depois que ambos são ativados, as métricas do GKE Dataplane V2 são enviadas para o Google Cloud Managed Service para Prometheus.

Criar um cluster do Autopilot com as métricas do GKE Dataplane V2 ativadas

Quando você cria novos clusters do Autopilot do GKE, o GKE ativa as métricas do GKE Dataplane V2 por padrão no cluster sem exigir uma sinalização específica.

Para usar as métricas do cluster do GKE Autopilot do GKE Dataplane V2 com o Google Cloud Managed Service para Prometheus, configure o recurso ClusterPodMonitoring para extrair as métricas e enviá-las ao Google Cloud Managed Service para Prometheus

Crie um manifesto ClusterPodMonitoring:

# Copyright 2023 Google LLC
#
# 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
#
#     https://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: monitoring.googleapis.com/v1
kind: ClusterPodMonitoring
metadata:
  name: advanced-datapath-observability-metrics
spec:
  selector:
    matchLabels:
      k8s-app: cilium
  endpoints:
  - port: flowmetrics
    interval: 60s
    metricRelabeling:
    # only keep denormalized pod flow metrics
    - sourceLabels: [__name__]
      regex: 'pod_flow_(ingress|egress)_flows_count'
      action: keep
    # extract pod name
    - sourceLabels: [__name__, destination]
      regex: 'pod_flow_ingress_flows_count;([a-zA-Z0-9-\.]+)/([a-zA-Z0-9-\.]+)'
      replacement: '${2}'
      targetLabel: pod_name
      action: replace
    - sourceLabels: [__name__, source]
      regex: 'pod_flow_egress_flows_count;([a-zA-Z0-9-\.]+)/([a-zA-Z0-9-\.]+)'
      replacement: '${2}'
      targetLabel: pod_name
      action: replace
    # extract workload name by removing 2 last "-XXX" parts
    - sourceLabels: [pod_name]
      regex: '([a-zA-Z0-9-\.]+)((-[a-zA-Z0-9\.]+){2})'
      replacement: '${1}'
      targetLabel: workload_name
      action: replace
    # extract workload name by removing one "-XXX" part when pod name has only 2 parts (eg. daemonset)
    - sourceLabels: [pod_name]
      regex: '([a-zA-Z0-9\.]+)((-[a-zA-Z0-9\.]+){1})'
      replacement: '${1}'
      targetLabel: workload_name
      action: replace
    # extract pod namespace
    - sourceLabels: [__name__, destination]
      regex: 'pod_flow_ingress_flows_count;([a-zA-Z0-9-\.]+)/([a-zA-Z0-9-\.]+)'
      replacement: '${1}'
      targetLabel: namespace_name
      action: replace
    - sourceLabels: [__name__, source]
      regex: 'pod_flow_egress_flows_count;([a-zA-Z0-9-\.]+)/([a-zA-Z0-9-\.]+)'
      replacement: '${1}'
      targetLabel: namespace_name
      action: replace
    # extract remote workload name
    - sourceLabels: [__name__, source]
      regex: 'pod_flow_ingress_flows_count;([a-zA-Z0-9-\.]+)/([a-zA-Z0-9-\.]+)'
      replacement: '${2}'
      targetLabel: remote_workload
      action: replace
    - sourceLabels: [__name__, destination]
      regex: 'pod_flow_egress_flows_count;([a-zA-Z0-9-\.]+)/([a-zA-Z0-9-\.]+)'
      replacement: '${2}'
      targetLabel: remote_workload
      action: replace
    # extract remote workload namespace
    - sourceLabels: [__name__, source]
      regex: 'pod_flow_ingress_flows_count;([a-zA-Z0-9-\.]+)/([a-zA-Z0-9-\.]+)'
      replacement: '${1}'
      targetLabel: remote_namespace
      action: replace
    - sourceLabels: [__name__, destination]
      regex: 'pod_flow_egress_flows_count;([a-zA-Z0-9-\.]+)/([a-zA-Z0-9-\.]+)'
      replacement: '${1}'
      targetLabel: remote_namespace
      action: replace
    # default remote workload class to "pod"
    - replacement: 'pod'
      targetLabel: remote_class
      action: replace
    # extract remote workload class from reserved identity
    - sourceLabels: [__name__, source]
      regex: 'pod_flow_ingress_flows_count;reserved:([^/]*)'
      replacement: '${1}'
      targetLabel: remote_class
      action: replace
    - sourceLabels: [__name__, destination]
      regex: 'pod_flow_egress_flows_count;reserved:([^/]*)'
      replacement: '${1}'
      targetLabel: remote_class
      action: replace
  targetLabels:
    metadata: []

Aplique o manifesto ClusterPodMonitoring:

kubectl apply -f ClusterPodMonitoring.yaml

Criar um cluster padrão com métricas do GKE Dataplane V2 ativadas

Para ativar as métricas do GKE Dataplane V2, crie um cluster com a flag --enable-dataplane-v2-metrics:

gcloud container clusters create CLUSTER_NAME \
    --enable-dataplane-v2 \
    --enable-ip-alias \
    --enable-managed-prometheus \
    --enable-dataplane-v2-metrics

Substitua:

CLUSTER_NAME: o nome do cluster.

A flag --enable-managed-prometheus instrui o GKE a usar as métricas com o Google Cloud Managed Service para Prometheus.