快捷導(dǎo)航

elasticsearch?組件基于單機的多實例集群部署方法

更新時間：2024年03月26日 10:34:37 作者：終點就在前方

es 作為搜索引擎，應(yīng)用場景不乏日志分析、網(wǎng)絡(luò)安全、搜索引擎等，有時也會用作日志數(shù)據(jù)庫使用，畢竟其出色的搜索查詢性能，不是同等量級關(guān)系型數(shù)據(jù)庫可以比擬的，這篇文章主要介紹了elasticsearch?組件基于單機的多實例集群,需要的朋友可以參考下

聲明：
本示例主要作為測試用，生產(chǎn)請慎重。

最近公司突發(fā)奇想，想讓我們搞個單機多實例的 es 的集群，看看其性能咋樣。通常來說，es 作為搜索引擎，應(yīng)用場景不乏日志分析、網(wǎng)絡(luò)安全、搜索引擎等，有時也會用作日志數(shù)據(jù)庫使用，畢竟其出色的搜索查詢性能，不是同等量級關(guān)系型數(shù)據(jù)庫可以比擬的，主要還是因為其倒排索引的特殊性，這里不討論倒排索引與 B+ Tree 的性能，我們主要看看這種集群怎么組建的。

環(huán)境準(zhǔn)備：

ubuntu，24核，64G
docker 20.10.2

因為是 es 集群，我們準(zhǔn)備通過 docker 來創(chuàng)建實例，所以之前你還得先 pull es、kibana 的 image:

docker pull elasticsearch:6.8.23
docker pull kibana:6.8.23

如果你的容器有限，可以直接通過腳本運行 docker run，但是如果容器數(shù)量多還有相關(guān)依賴，建議通過容器編排起容器，當(dāng)然數(shù)量更大的情況下，建議通過 k8s 部署。

我們的集群主要包含 3 個 es 節(jié)點，外加一個 kibana 作為觀測，所以通過 docker-compose 作為容器編排，相對合適。

接下來是我們的編排定義：
docker-compose.yml

version: "2.3"
services:
  es-0:
    image: elasticsearch:6.8.23
    hostname: es-0
    container_name: es-0
    environment:
      - bootstrap.memory_lock=true
    ulimits:
      memlock:
        soft: -1
        hard: -1
    cap_add:
      - IPC_LOCK
    volumes:
      - /var/xxx/es_cluster/es-0:/usr/share/elasticsearch/data # 容器數(shù)據(jù)映射
      - ./es_cluster/es-0/elasticsearch:/etc/default/elasticsearch # elasticsearch 文件映射
      - ./es_cluster/es-0/config:/usr/share/elasticsearch/config # 配置映射，主要是 elasticsearch.yaml 和 jvm.options
      - /var/log/es_cluster/es-0/logs:/usr/share/elasticsearch/logs # 日志映射
      - /usr/share/zoneinfo/Asia/Shanghai:/etc/localtime # 時間
      - /etc/timezone:/etc/timezone
      #- ./elasticsearch/jvm.options:/etc/elasticsearch/jvm.options
    ports:
      - "9200:9200" # 端口映射
    command: elasticsearch
    logging:
      options:
        max-size: "200M"
        max-file: "5"
    networks:
      app_net:
        ipv4_address: 172.238.238.219
    healthcheck:
        test: ["CMD", "curl", "-f", "-s", "http://172.238.238.219:9200/_cluster/health?wait_for_status=yellow&timeout=50s"]
        interval: 30s
        timeout: 10s
        retries: 20
    restart: always
  es-1:
    image: elasticsearch:6.8.23
    hostname: es-1
    container_name: es-1
    environment:
      - bootstrap.memory_lock=true
    ulimits:
      memlock:
        soft: -1
        hard: -1
    cap_add:
      - IPC_LOCK
    volumes:
      - /var/xxx/es_cluster/es-1:/usr/share/elasticsearch/data
      - ./es_cluster/es-1/elasticsearch:/etc/default/elasticsearch
      - ./es_cluster/es-1/config:/usr/share/elasticsearch/config
      - /var/log/es_cluster/es-1/logs:/usr/share/elasticsearch/logs
      - /usr/share/zoneinfo/Asia/Shanghai:/etc/localtime
      - /etc/timezone:/etc/timezone
      #- ./elasticsearch/jvm.options:/etc/elasticsearch/jvm.options
    ports:
      - "9201:9201"
    command: elasticsearch
    logging:
      options:
        max-size: "200M"
        max-file: "5"
    networks:
      app_net:
        ipv4_address: 172.238.238.229
    healthcheck:
      test: ["CMD", "curl", "-f", "-s", "http://172.238.238.229:9200/_cluster/health?wait_for_status=yellow&timeout=50s"]
      interval: 30s
      timeout: 10s
      retries: 20
    restart: always
  es-2:
    image: elasticsearch:6.8.23
    hostname: es-2
    container_name: es-2
    environment:
      - bootstrap.memory_lock=true
    ulimits:
      memlock:
        soft: -1
        hard: -1
    cap_add:
      - IPC_LOCK
    volumes:
      - /var/xxx/es_cluster/es-2:/usr/share/elasticsearch/data
      - ./es_cluster/es-2/elasticsearch:/etc/default/elasticsearch
      - ./es_cluster/es-2/config:/usr/share/elasticsearch/config
      - /var/log/es_cluster/es-2/logs:/usr/share/elasticsearch/logs
      - /usr/share/zoneinfo/Asia/Shanghai:/etc/localtime
      - /etc/timezone:/etc/timezone
      #- ./elasticsearch/jvm.options:/etc/elasticsearch/jvm.options
    ports:
      - "9202:9202"
    command: elasticsearch
    logging:
      options:
        max-size: "200M"
        max-file: "5"
    networks:
      app_net:
        ipv4_address: 172.238.238.239
    healthcheck:
      test: ["CMD", "curl", "-f", "-s", "http://172.238.238.239:9200/_cluster/health?wait_for_status=yellow&timeout=50s"]
      interval: 30s
      timeout: 10s
      retries: 20
    restart: always
  kibana:
    image: kibana:6.8.23
    hostname: kibana
    container_name: kibana
    volumes:
      - ./kibana/kibana.yml:/usr/share/kibana/config/kibana.yml
      - /usr/share/zoneinfo/Asia/Shanghai:/etc/localtime
      - /etc/timezone:/etc/timezone
      - ./kibana/kibana.keystore:/usr/share/kibana/data/kibana.keystore
    ports:
      - "5601:5601"
    networks:
      app_net:
        ipv4_address: 172.238.238.242
    restart: always
    logging:
      options:
        max-size: "200M"
        max-file: "5"
networks:
  app_net:
    driver: bridge
    ipam:
      driver: default
      config:
      - subnet: 172.238.238.0/24

這里限定 docker 的網(wǎng)絡(luò)網(wǎng)段。

然后我們要看看對應(yīng)的其他準(zhǔn)備。

主要看我們的對應(yīng)到主機中的 data 目錄，所以參考 yml 中的相關(guān)映射，注意創(chuàng)建相關(guān)目錄。

這里我們主要看看相關(guān)的 elasticsearch.yaml 和 jvm.options。

elasticsearch.yml

cluster:
  name: logserver
#################
node.name: es-0 # 其他節(jié)點類似，修改 node name
discovery.zen.ping.unicast.hosts: ["es-0", "es-1", "es-2"]
network.host: 0.0.0.0
discovery.zen.minimum_master_nodes: 2
gateway.recover_after_nodes: 3
http.port: 9200
transport.tcp.port: 9300
node.master: true
node.data: true
http.host: 0.0.0.0
http:
  enabled: true
  compression: true
  cors:
    enabled: true
    allow-origin: "*"
bootstrap.memory_lock: true
bootstrap.system_call_filter: false
path.data: /usr/share/elasticsearch/data
#cluster.routing.allocation.disk.threshold_enabled: true
#cluster.routing.allocation.disk.watermark.flood_stage: 80gb
#cluster.routing.allocation.disk.watermark.high: 100gb
#cluster.routing.allocation.disk.watermark.low: 120gb
##Lock memory, do not write swap
#bootstrap.mlockall: true
##The cache type is set to Soft Reference,
##and is reclaimed only if there is not enough memory
#index.cache.field.max_size: 50000
#index.cache.field.expire: 10m
#index.cache.field.type: soft
##Weighing the performance of the index and the timeliness of retrieval
#index.translog.flush_threshold_ops: 10000
#index.refresh_interval: 1
#number_of_replicas: 0
#indices.lifecycle.poll_interval: 5m
xpack.ml.enabled: false

jvm.options

...
# Xms represents the initial size of total heap space
# Xmx represents the maximum size of total heap space
-Xms16g # 主機內(nèi)存64G，每個實例分配16G
-Xmx16g
################################################################
## Expert settings
################################################################
##
...

這兩個文件按照上面內(nèi)容修改。

接下來是 kibana.yml:

...
elasticsearch.url: "http://172.xxx.xxx.xxx:9200" # 根據(jù)實際情況，填入自己的主機ip
...

接下來通過 docker-compose 命令就可以起容器了。

docker-compsoe up -d # 后臺運行容器
docker-compose ps # 查看容器運行狀態(tài)
docker-compose down # 停掉容器

接下來看看容器狀態(tài)：

# docker-compose ps
 Name               Command                  State                         Ports
---------------------------------------------------------------------------------------------------
es-0     /usr/local/bin/docker-entr ...   Up (healthy)   0.0.0.0:9200->9200/tcp, 9300/tcp
es-1     /usr/local/bin/docker-entr ...   Up (healthy)   9200/tcp, 0.0.0.0:9201->9201/tcp, 9300/tcp
es-2     /usr/local/bin/docker-entr ...   Up (healthy)   9200/tcp, 0.0.0.0:9202->9202/tcp, 9300/tcp
kibana   /usr/local/bin/kibana-docker     Up             0.0.0.0:5601->5601/tcp

通過 kibana 查看容器狀態(tài)：