【ELK實(shí)踐&開(kāi)發(fā)★壹】基于Docker純手工構(gòu)建Elasticsearch高可用集群環(huán)境

jasonruan 2018.12.03

[TOC]

0 前言

Elasticsearch——You Know弦疮，F(xiàn)or Search：）诵竭，既然如此，筆者也不過(guò)多介紹谤草，直接進(jìn)入正題——基于Docker純手工構(gòu)建Elasticsearch高可用集群衙傀。

1 Elasticsearch環(huán)境搭建

1.1 構(gòu)建開(kāi)發(fā)環(huán)境基礎(chǔ)鏡像

關(guān)于開(kāi)發(fā)基礎(chǔ)鏡像構(gòu)建抬吟，請(qǐng)參看筆者另外一篇文章：《基于以太坊為女兒發(fā)一枚數(shù)字貨幣》(http://www.reibang.com/p/455f17f4143b)

1.2 構(gòu)建Elasticsearch基礎(chǔ)鏡像

基于開(kāi)發(fā)環(huán)境基礎(chǔ)鏡像安裝Elasticsearch當(dāng)前較新版本6.5.1(November 20, 2018)

PS：之所以用較新版本，因筆者準(zhǔn)備發(fā)文前一看统抬，才發(fā)現(xiàn)ES已經(jīng)更新到6.5.2（December 05, 2018）火本，ES的版本發(fā)布速度也是太快了：（

1.2.1 編寫(xiě)Dockerfile

# js.es.Dockerfile 
FROM js/ubuntu:18.04
ENV ES_VERSION "6.5.1"
ENV JAVA_HOME /opt/java/jdk1.8.0_181
ENV PATH $PATH:$JAVA_HOME/bin
RUN mkdir -p /rzexin/soft && \
        wget https://artifacts.elastic.co/downloads/Elasticsearch/Elasticsearch-$ES_VERSION.tar.gz -P /rzexin/soft && \
        tar -zxvf /rzexin/soft/Elasticsearch-$ES_VERSION.tar.gz -C /rzexin/soft 
WORKDIR /rzexin/soft/
RUN ln -s Elasticsearch-$ES_VERSION es
WORKDIR /rzexin/soft/es
 
RUN groupadd jses && useradd jses -g jses && chown -R jses:jses /rzexin/soft
USER jses 
 
RUN sed -i 's%#network.host.*$%network.host: 0.0.0.0%g' config/Elasticsearch.yml 
RUN mkdir /rzexin/soft/es/data
 
ENTRYPOINT ["./bin/Elasticsearch"] 

1.2.2 構(gòu)建鏡像

$ sudo docker build -t rzexin/es:6.5.1 -f js.es.Dockerfile .

1.3 單節(jié)點(diǎn)環(huán)境搭建

1.3.1 啟動(dòng)鏡像

$ sudo docker run -e ES_JAVA_OPTS="-Xms256m -Xmx256m" -d -p 9200:9200 -p 9300:9300 --name JS01 rzexin/es:6.5.1
a4941ef67aae56e394a7d6ada511d026fad4cba29879c3fe86f37e8120c7fe5a

1.3.2 測(cè)試驗(yàn)證

在宿主機(jī)執(zhí)行命令，如輸出如下信息聪建，表明單節(jié)點(diǎn)的Elasticsearch已經(jīng)啟動(dòng)成功：）

$ http http://localhost:9200
HTTP/1.1 200 OK
content-encoding: gzip
content-length: 297
content-type: application/json; charset=UTF-8

{
    "cluster_name": "Elasticsearch",
    "cluster_uuid": "V5-76-zfQmKv-JgQkjlK4Q",
    "name": "RDwpJOV",
    "tagline": "You Know, for Search",
    "version": {
        "build_date": "2018-11-16T02:22:42.182257Z",
        "build_flavor": "default",
        "build_hash": "8c58350",
        "build_snapshot": false,
        "build_type": "tar",
        "lucene_version": "7.5.0",
        "minimum_index_compatibility_version": "5.0.0",
        "minimum_wire_compatibility_version": "5.6.0",
        "number": "6.5.1"
    }
}

1.4 高可用集群環(huán)境搭建

以下以搭建3個(gè)Elasticsearch節(jié)點(diǎn)的集群環(huán)境為例介紹

1.4.1 構(gòu)建Elasticsearch集群鏡像

1.4.1.1 編寫(xiě)Dockerfile

在Elasticsearch基礎(chǔ)鏡像之上構(gòu)建集群鏡像

主要是設(shè)置集群名稱钙畔、節(jié)點(diǎn)名稱、集群節(jié)點(diǎn)列表等

FROM rzexin/es:6.5.1          
 
RUN sed -i 's%#cluster.name.*$%cluster.name: jses%g' config/elasticsearch.yml && \
        sed -i 's%#discovery.zen.minimum_master_nodes:.*$%discovery.zen.minimum_master_nodes: 1%g' config/elasticsearch.yml
 
RUN echo "[ ! -z $""NODE_NAME"" ] && sed -i 's%#node.name.*$%node.name: '$""NODE_NAME""'%g' config/elasticsearch.yml" > /rzexin/soft/es/start.sh && \
    echo "[ ! -z $""DISCOVERY_HOSTS"" ] && sed -i 's%#discovery.zen.ping.unicast.hosts:.*$%discovery.zen.ping.unicast.hosts:                     ['$""DISCOVERY_HOSTS""']%g' config/elasticsearch.yml" >> /rzexin/soft/es/start.sh && \
    echo "./bin/elasticsearch" >> /rzexin/soft/es/start.sh && \
    chmod a+x /rzexin/soft/es/start.sh
 
ENTRYPOINT ["sh", "/rzexin/soft/es/start.sh"]

1.4.1.2 構(gòu)建鏡像

$ sudo docker build -t rzexin/escluster:6.5.1 -f js.es.cluster.Dockerfile .

1.4.2 創(chuàng)建數(shù)據(jù)文件掛載目錄

分別為3個(gè)節(jié)點(diǎn)創(chuàng)建數(shù)據(jù)文件掛載目錄

$ sudo docker volume create --name=data-js-es-1
data-js-es-1
$ sudo docker volume create --name=data-js-es-2
data-js-es-2
$ sudo docker volume create --name=data-js-es-3
data-js-es-3

1.4.3 編寫(xiě)docker-compose編排文件

# cat js.es.docker-compose.yml 
version: "3"
services:
  js01:
    build:
      context: .
      dockerfile: js.es.cluster.Dockerfile
    image: rzexin/escluster:6.5.1 
    container_name: js01
    hostname: js01
    privileged: true
    ports:
      - "9200:9200"
      - "9300:9300"
    expose:
      - 9200
      - 9300
    environment:
      NODE_NAME: "js-es-1"
      DISCOVERY_HOSTS: "\"js01\",\"js02\",\"js03\""
      ES_JAVA_OPTS: "-Xms256m -Xmx256m"
    volumes:
        - data-js-es-1:/rzexin/soft/es/data
  js02:
    build:
      context: .
      dockerfile: js.es.cluster.Dockerfile
    image: rzexin/escluster:6.5.1 
    container_name: js02
    hostname: js02
    ports:
      - "29200:9200"
      - "29300:9300"
    expose:
      - 29200
      - 29300
    environment:
      NODE_NAME: "js-es-2"
      DISCOVERY_HOSTS: "\"js01\",\"js02\",\"js03\""
      ES_JAVA_OPTS: "-Xms256m -Xmx256m"
    volumes:
      - data-js-es-2:/rzexin/soft/es/data
  js03:
    build:
      context: .
      dockerfile: js.es.cluster.Dockerfile
    image: rzexin/escluster:6.5.1 
    container_name: js03
    hostname: js03
    ports:
      - "39200:9200"
      - "39300:9300"
    expose:
      - 39200
      - 39300
    environment:
      NODE_NAME: "js-es-3"
      DISCOVERY_HOSTS: "\"js01\",\"js02\",\"js03\""
      ES_JAVA_OPTS: "-Xms256m -Xmx256m"
    volumes:
      - data-js-es-3:/rzexin/soft/es/data
volumes:
  data-js-es-1:
    external: true
  data-js-es-2:
    external: true
  data-js-es-3:
    external: true

1.4.4 啟動(dòng)集群節(jié)點(diǎn)

$ sudo docker-compose -f js.es.docker-compose.yml up -d
Creating network "dockerescluster_default" with the default driver
Creating js01 ... done
Creating js03 ... done
Creating js02 ... done

1.4.5 查看集群節(jié)點(diǎn)狀態(tài)

$ sudo docker-compose -f js.es.docker-compose.yml ps
Name             Command             State                                   Ports     
-------------------------------------------------------------------------------------------------------------------
js01   sh /rzexin/soft/es/start.sh   Up      0.0.0.0:9200->9200/tcp, 0.0.0.0:9300->9300/tcp                        
js02   sh /rzexin/soft/es/start.sh   Up      29200/tcp, 29300/tcp, 0.0.0.0:29200->9200/tcp, 0.0.0.0:29300->9300/tcp
js03   sh /rzexin/soft/es/start.sh   Up      39200/tcp, 39300/tcp, 0.0.0.0:39200->9200/tcp, 0.0.0.0:39300->9300/tcp

1.4.6 測(cè)試驗(yàn)證

在宿主機(jī)執(zhí)行命令金麸，查看集群節(jié)點(diǎn)情況擎析，打星號(hào)*的是自動(dòng)選舉出的主節(jié)點(diǎn)

$ curl http://localhost:9200/_cat/nodes?pretty
172.19.0.4 58 98 32 2.31 3.63 2.19 mdi - js-es-1
172.19.0.3 60 98 33 2.31 3.63 2.19 mdi * js-es-2
172.19.0.2 57 98 33 2.31 3.63 2.19 mdi - js-es-3

1.5 官方docker鏡像搭建

要是不愿意折騰手工編譯安裝搭建，也可考慮直接使用Elasticsearch官方的docker鏡像挥下，具體步驟如下：

1.5.1 獲取官方鏡像

• 搜索鏡像

$ sudo docker search Elasticsearch:6.5.1
NAME             DESCRIPTION                                     STARS OFFICIAL  
Elasticsearch    Elasticsearch is a powerful open source sear…   3270  [OK]      
kibana           Kibana gives shape to any kind of data — str…   1332  [OK]    

• 下載鏡像

$ sudo docker pull Elasticsearch:6.5.1
6.5.1: Pulling from library/Elasticsearch
Digest: sha256:0db7521b9cd9f3906aae70fd146b78ce204b1c7c13fc32e4a2ed449e6576af53
Status: Image is up to date for Elasticsearch:6.5.1

• 查看鏡像

$ sudo docker images elasticsearch
REPOSITORY          TAG                 IMAGE ID            CREATED             SIZE
elasticsearch       6.5.1               32f93c89076d        2 weeks ago         773MB

1.5.2 運(yùn)行官方鏡像

• 運(yùn)行鏡像

運(yùn)行該鏡像默認(rèn)會(huì)占用2G內(nèi)存揍魂，受限于機(jī)器性能，故通過(guò)ES_JAVA_OPTS參數(shù)限制內(nèi)存使用

$ sudo docker run -e ES_JAVA_OPTS="-Xms256m -Xmx256m" -d -p 9200:9200 -p 9300:9300 --name ES01 32f93c89076d
29ff9574492ce0f3c008664fc1044851402e3594fbada62d91b4da725615fbbf

• 檢查鏡像

$ sudo docker ps 
CONTAINER ID        IMAGE               COMMAND                  CREATED             STATUS              PORTS                                            NAMES
29ff9574492c        32f93c89076d        "/usr/local/bin/dock…"   23 seconds ago      Up 22 seconds       0.0.0.0:9200->9200/tcp, 0.0.0.0:9300->9300/tcp   ES01

1.5.3 測(cè)試驗(yàn)證

看到如下輸出棚瘟，表示Elasticsearch啟動(dòng)成功

$ http 'http://localhost:9200' 
HTTP/1.1 200 OK
content-encoding: gzip
content-length: 298
content-type: application/json; charset=UTF-8

{
    "cluster_name": "docker-cluster",
    "cluster_uuid": "djDC-rsqQ9KD2hOXRP3Rew",
    "name": "jzqf9Oy",
    "tagline": "You Know, for Search",
    "version": {
        "build_date": "2018-11-16T02:22:42.182257Z",
        "build_flavor": "default",
        "build_hash": "8c58350",
        "build_snapshot": false,
        "build_type": "tar",
        "lucene_version": "7.5.0",
        "minimum_index_compatibility_version": "5.0.0",
        "minimum_wire_compatibility_version": "5.6.0",
        "number": "6.5.1"
    }
}

2 容災(zāi)驗(yàn)證

• 查看當(dāng)前集群節(jié)點(diǎn)情況

  打星號(hào)*的是自動(dòng)選舉出的主節(jié)點(diǎn)

$ curl http://localhost:9200/_cat/nodes?pretty
172.19.0.4 58 98 32 2.31 3.63 2.19 mdi - js-es-1
172.19.0.3 60 98 33 2.31 3.63 2.19 mdi * js-es-2
172.19.0.2 57 98 33 2.31 3.63 2.19 mdi - js-es-3

• 停掉2號(hào)主節(jié)點(diǎn)（js-es-2）

  $ sudo docker-compose -f js.es.docker-compose.yml stop js02
  Stopping js02 ... done
  

• 查看集群節(jié)點(diǎn)情況

  可見(jiàn)2號(hào)主節(jié)點(diǎn)（js-es-2）已被剔掉现斋，并重新選舉出了新的主節(jié)點(diǎn)

  $ curl http://localhost:9200/_cat/nodes?pretty
  172.23.0.4 48 90 22 1.01 2.86 2.64 mdi * js-es-3
  172.23.0.3 56 90 22 1.01 2.86 2.64 mdi - js-es-1
  

• 啟動(dòng)2號(hào)節(jié)點(diǎn)（js-es-2）

  $ sudo docker-compose -f js.es.docker-compose.yml start js02
  Starting js02 ... done
  

• 再次查看集群節(jié)點(diǎn)情況

  可見(jiàn)2號(hào)節(jié)點(diǎn)又重新加入集群，但這時(shí)已經(jīng)不是主節(jié)點(diǎn)了

  $ curl http://localhost:9200/_cat/nodes?pretty
  172.23.0.2 43 98 31 1.58 2.25 2.43 mdi - js-es-2
  172.23.0.4 51 98 31 1.58 2.25 2.43 mdi * js-es-3
  172.23.0.3 58 98 31 1.58 2.25 2.43 mdi - js-es-1
  

3 常見(jiàn)錯(cuò)誤

3.1 錯(cuò)誤：can not run elasticsearch as root

• 錯(cuò)誤描述

  執(zhí)行Elasticsearch啟動(dòng)命令報(bào)錯(cuò)

org.elasticsearch.bootstrap.StartupException: java.lang.RuntimeException: can not run elasticsearch as root

• 錯(cuò)誤原因

  出于系統(tǒng)安全考慮偎蘸，不可使用root用戶啟動(dòng)庄蹋，因Elasticsearch可以接收用戶輸入的腳本并且執(zhí)行瞬内，為了系統(tǒng)安全考慮，建議創(chuàng)建一個(gè)單獨(dú)的用戶用來(lái)運(yùn)行Elasticsearch蔓肯。

• 解決方案

  創(chuàng)建用戶和分組遂鹊，并使用該用戶啟動(dòng)es

groupadd jses
useradd jses -g jses
chown -R jses:jses /rzexin/soft/es
su jses
./bin/elasticsearch

3.2 錯(cuò)誤：vm.max_map_count is too low

[1]: max virtual memory areas vm.max_map_count [65530] is too low, increase to at least [262144]

（1）臨時(shí)修改

• 在宿主機(jī)上執(zhí)行命令

# 查看當(dāng)前值
$ sudo sysctl -a | grep "vm.max_map_count"
vm.max_map_count = 65530

# 修改成最低要求
$ sudo sysctl -w vm.max_map_count=262144
vm.max_map_count = 262144

# 查看修改后的值
$ sudo sysctl -a | grep "vm.max_map_count"
vm.max_map_count = 262144

• 重啟docker鏡像

# 先刪除容器振乏，否則報(bào)錯(cuò)
$ sudo docker run -e ES_JAVA_OPTS="-Xms512m -Xmx512m" -d -p 9200:9200 -p 9300:9300 --name ES01 32f93c89076d
docker: Error response from daemon: Conflict. The container name "/ES01" is already in use by container "a9e2daa0227bb60a7ce3fffd0881eb7d4caa2b032c3933f52fb6a5bcd55def12". You have to remove (or rename) that container to be able to reuse that name.
See 'docker run --help'.

$ sudo docker ps -l
CONTAINER ID        IMAGE               COMMAND                  CREATED             STATUS                      PORTS               NAMES
a9e2daa0227b        32f93c89076d        "/usr/local/bin/dock…"   6 minutes ago       Exited (78) 6 minutes ago                       ES01

$ sudo docker rm a9e2daa0227b
a9e2daa0227b

# 重啟鏡像
$ sudo docker run -e ES_JAVA_OPTS="-Xms512m -Xmx512m" -d -p 9200:9200 -p 9300:9300 --name ES01 32f93c89076d
6749988a01723a1ba9079066274676fbdd302866011e8158a70729b0b98de11b

（2）永久修改

修改配置文件：sudo vi /etc/sysctl.conf

加入：

vm.max_map_count=262144

保存即可蔗包。

4 后記

完成了Elasticsearch集群環(huán)境搭建，下一篇將介紹Elasticsearch與SpringBoot和Golang的整合慧邮。