使用 Elastic 技术栈构建 K8S 全栈监控（1）

唐伯虎

在本系列文章中，我将最近学习的如何使用Elastic技术栈来为Kubernetes构建监控环境做一个简单的整理，以便后期查阅及加深理解。

ELK简单介绍
ELK是Elasticsearch、Logstash、Kibana三大开源框架首字母大写简称(后期出现的filebeat(beats中的一种)用来替代logstash的数据收集功能，比较轻量级)。Filebeat是用于转发和集中日志数据的轻量级传送工具。Filebeat监视您指定的日志文件或位置，收集日志事件，并将它们转发到Elasticsearch或Logstash进行索引。Logstash是免费且开放的服务器端数据处理管道，能够从多个来源采集数据，转换数据。Elasticsearch是Elastic Stack核心的分布式搜索和分析引擎,是一个基于Lucene、分布式、通过Restful方式进行交互的近实时搜索平台框架。Elasticsearch为所有类型的数据提供近乎实时的搜索和分析。
创建命名空间
我会用的环境是kubernetes v1.20.4版本的集群。为方便管理，我将涉及到的所有资源都部署到elastic的名称空间中

# kubectl create ns elastic
namespace/elastic created

配置文件(elasticsearch-configmap.yml)
首先创建一个configmap资源对象生成es的配置文件

需要注意：node.name最好使用完整的能解析到的域名，至少要和discovery.seed_hosts参数保持一致，否则es集群会协商不起来。

---
apiVersion: v1
kind: ConfigMap
metadata:
namespace: elastic
name: es-config
labels:
app: elasticsearch
role: es
data:
elasticsearch.yml: |-
cluster.name: ${CLUSTER_NAME}
node.name: ${NODE_NAME}.es-svc.elastic.svc.cluster.local
path.data: ${PATH_DATA}
path.logs: ${PATH_LOGS}
discovery.seed_hosts: ${NODE_LIST}
cluster.initial_master_nodes: ${NODE_LIST} 
network.host: 0.0.0.0
node:
master: true
data: true
xpack.security.enabled: true
xpack.monitoring.collection.enabled: true
xpack.monitoring.collection.interval: 30s

部署文件(elasticsearch-statefulset.yml)

由于es集群存在多个数据节点，且每个节点上存在的数据是不一致的，数据都需要单独存储，所以采用statefulset控制器进行管理部署，同时需要使用volumeClaimTemplates来分别创建存储卷，我使用nfs作为后端的数据存储。

---
apiVersion: apps/v1
kind: StatefulSet
metadata:
  namespace: elastic
  name: elasticsearch
  labels:
  app: elasticsearch
  role: es
spec:
  serviceName: "es-svc"
  replicas: 3
  selector:
  matchLabels:
    app: elasticsearch
    role: es
  template:
  metadata:
    labels:
    app: elasticsearch
    role: es
  spec:
    # pod反亲和策略，用于将具有role=es标签的pod打散
    affinity:
    podAntiAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
      - labelSelector:
        matchExpressions:
          - key: "role"
          operator: In
          values:
            - "es"
        topologyKey: "kubernetes.io/hostname"
    containers:
    - image: elasticsearch:7.8.0
    imagePullPolicy: IfNotPresent
    name:  elasticsearch
    env:
    - name: CLUSTER_NAME
      value: elasticsearch-cluster
    - name: PATH_DATA
      value: /opt/es/data
    - name: PATH_LOGS
      value: /usr/share/elasticsearch/logs
    - name: NODE_NAME
      valueFrom:
      fieldRef:
        fieldPath: metadata.name
    - name: NODE_LIST
      value: "elasticsearch-0.es-svc.elastic.svc.cluster.local,elasticsearch-1.es-svc.elastic.svc.cluster.local,elasticsearch-2.es-svc.elastic.svc.cluster.local"
    - name: "ES_JAVA_OPTS"
      value: "-Xms1024m -Xmx1024m"    
    ports:
    - containerPort: 9200
      name: http
    - containerPort: 9300
      name: transport
    volumeMounts:
    - mountPath: /usr/share/elasticsearch/config/elasticsearch.yml
      readOnly: true
      subPath: elasticsearch.yml
      name: config
    - mountPath: /opt/es/data
      name: es-data
    - mountPath: /usr/share/elasticsearch/logs
      name: es-logs
    volumes:
    - name: config
    configMap:
      name: es-config
  volumeClaimTemplates:
  - metadata:
    name: es-data
  spec:
    accessModes: [ "ReadWriteOnce" ]
    storageClassName: managed-nfs-storage
    resources:
    requests:
      storage: 10Gi 
  - metadata:
    name: es-logs
  spec:
    accessModes: [ "ReadWriteOnce" ]
    storageClassName: managed-nfs-storage
    resources:
    requests:
      storage: 5Gi

无头服务(elasticsearch-service.yml)

用于集群内部访问

---
kind: Service
apiVersion: v1
metadata:
  namespace: elastic
  name: es-svc
  labels:
  app: elasticsearch
  role: es
spec:
  selector:
  app: elasticsearch
  role: es
  clusterIP: None
  ports:
  - name: http
  port: 9200
  - name: transport
  port: 9300

映射至外网(es-ingress.yml)

配置ingress，提供一个外部访问的域名(elastic.liheng.com)

---
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  namespace: elastic
  name: es-ingress
spec:
  rules:
  - host: elastic.liheng.com
  http: 
    paths:
    - pathType: Prefix
    path: "/"
    backend:
      service:
      name: es-svc
      port: 
        number: 9200

部署
创建上面的4个资源对象即可

# kubectl apply -f elasticsearch-configmap.yml
# kubectl apply -f elasticsearch-statefulset.yml
# kubectl apply -f elasticsearch-service.yml
# kubectl apply -f es-ingress.yml

如果容器启动失败，可以通过kubectl -n elastic logs -f elasticsearch-0查看启动日志。一般会出现"[1] max virtual memory areas vm.max_map_count [65530] is too low,increase to at least [262144]"报错，导致容器启动失败。可通过加入初始化容器将容器系统参数进行初始化调整解决：

···
initContainers:
    - name: init-vm-max-map
    image: busybox
    command: ["sysctl","-w","vm.max_map_count=262144"]
    securityContext:
      privileged: true
    - name: init-fd-ulimit
    image: busybox
    command: ["sh","-c","ulimit -n 65535"]
    securityContext:
      privileged: true
···

查看pod状态

# kubectl -n elastic get pod
NAME        READY   STATUS  RESTARTS   AGE
elasticsearch-0   1/1   Running   0      4m50s
elasticsearch-1   1/1   Running   0      4m6s
elasticsearch-2   1/1   Running   0      117s

生成密码
我们启用了xpack认证模块，我们需要通过命令生成一个初始密码

#kubectl -n elastic exec -it elasticsearch-0 -- bin/elasticsearch-setup-passwords auto -b
Changed password for user apm_system
PASSWORD apm_system = 7IujAS2g2rHLgmNddwqH

Changed password for user kibana_system
PASSWORD kibana_system = qUjxDUNuNGkbqBVyZfs8

Changed password for user kibana
PASSWORD kibana = qUjxDUNuNGkbqBVyZfs8

Changed password for user logstash_system
PASSWORD logstash_system = wToZIb0nafESAlxQJysa

Changed password for user beats_system
PASSWORD beats_system = YUqz4YUojyNxodJFcYb6

Changed password for user remote_monitoring_user
PASSWORD remote_monitoring_user = WQKNWUx2RD0so5WowLL8

Changed password for user elastic
PASSWORD elastic = F31k1cwR8Td1ulNUqnXW

查看es集群状态
浏览器访问`http://elastic.liheng.com

浏览器访问`http://elastic.liheng.com/_cat/nodes?v

浏览器访问http://elastic.liheng.com/_cluster/health?pretty

通过以上访问结果，可以证明ES集群已经正常运行。