咨询：如何在自建Kubernetes集群部署已验证的Elasticsearch镜像

Got it, let's walk through getting your Elasticsearch deployment fully set up on your GKE cluster. You’ve already checked off some key steps—running Elasticsearch locally via Docker, creating your GKE cluster, and starting a ReplicationController YAML. Let’s fill in the missing pieces and cover everything you need for a working deployment.

1. Finalize Your ReplicationController YAML #

Your partial RC is a solid starting point, but we need to add container details, critical Elasticsearch environment variables, and basic resource controls to make it work in Kubernetes. Here’s the full, functional version:

apiVersion: v1
kind: ReplicationController
metadata:
  name: elasticsearch
  labels:
    app: elasticsearch
spec:
  replicas: 2
  selector:
    app: elasticsearch
  template:
    metadata:
      labels:
        app: elasticsearch
    spec:
      containers:
      - name: elasticsearch
        image: elasticsearch:latest # Use the exact tag you pulled locally (e.g., 8.10.0) for consistency
        ports:
        - containerPort: 9200
          name: rest-api
        - containerPort: 9300
          name: inter-node-comm
        env:
        # For multi-node cluster (your 2 replicas), use these discovery settings
        - name: discovery.seed_hosts
          value: "elasticsearch" # Points to the service we'll create next
        - name: cluster.initial_master_nodes
          value: "elasticsearch-0,elasticsearch-1" # Matches pod names the RC will generate
        - name: ES_JAVA_OPTS
          value: "-Xms512m -Xmx512m" # Adjust based on your GKE node's available memory
        resources:
          requests:
            memory: "1Gi"
            cpu: "500m"
          limits:
            memory: "2Gi"
            cpu: "1"

If you want a simpler single-node cluster instead, replace the env block with:

env:
- name: discovery.type
  value: "single-node"
- name: ES_JAVA_OPTS
  value: "-Xms512m -Xmx512m"

2. Create a Service to Expose Elasticsearch #

You need a Kubernetes Service to make Elasticsearch accessible—either within the cluster or externally. Create a file named elasticsearch-service.yaml with this content:

apiVersion: v1
kind: Service
metadata:
  name: elasticsearch
  labels:
    app: elasticsearch
spec:
  ports:
  - port: 9200
    name: rest-api
  - port: 9300
    name: inter-node-comm
  selector:
    app: elasticsearch
  type: ClusterIP # Use NodePort or LoadBalancer if you need external public access

For external access, change type: LoadBalancer—GKE will automatically provision a public IP for you to reach Elasticsearch from outside the cluster.

3. Deploy the Resources to GKE #

Run these commands to apply your manifests to the cluster:

kubectl apply -f elasticsearch.yaml
kubectl apply -f elasticsearch-service.yaml

4. Verify Your Deployment #

Check if your Elasticsearch pods are running successfully:

kubectl get pods -l app=elasticsearch

Check the status of your Service:

kubectl get service elasticsearch

Test connectivity to Elasticsearch (from within the cluster, use a temporary curl pod):

kubectl run -it --rm --image=curlimages/curl curl-test -- curl elasticsearch:9200

You should see a JSON response with Elasticsearch version information if everything is working.

5. Production-Grade Recommendations (Optional) #

• Switch to a Deployment instead of ReplicationController: Deployments support rolling updates and rollbacks, which are far better for managing stateful apps like Elasticsearch. Here’s a quick Deployment replacement for your RC:

  apiVersion: apps/v1
  kind: Deployment
  metadata:
    name: elasticsearch
    labels:
      app: elasticsearch
  spec:
    replicas: 2
    selector:
      matchLabels:
        app: elasticsearch
    template:
      metadata:
        labels:
          app: elasticsearch
      spec:
        containers:
        - name: elasticsearch
          image: elasticsearch:latest
          ports:
          - containerPort: 9200
          - containerPort: 9300
          env:
          - name: discovery.seed_hosts
            value: "elasticsearch"
          - name: cluster.initial_master_nodes
            value: "elasticsearch-0,elasticsearch-1"
          - name: ES_JAVA_OPTS
            value: "-Xms512m -Xmx512m"
          resources:
            requests:
              memory: "1Gi"
              cpu: "500m"
            limits:
              memory: "2Gi"
              cpu: "1"
  

• Add Persistent Storage: Elasticsearch needs persistent volumes to retain data across pod restarts. Add volumeMounts and volumes sections to your pod template using PersistentVolumeClaims.
• Secure Your Cluster: Enable authentication, TLS encryption, and restrict access via NetworkPolicies—your current setup is unsecure for production use.

内容的提问来源于stack exchange，提问作者soundararajan.c