EFK教程-ElasticSearch角色分离

目前创新互联已为上千的企业提供了网站建设、域名、虚拟主机、网站托管、企业网站设计、望江网站维护等服务，公司将坚持客户导向、应用为本的策略，正道将秉承"和谐、参与、激情"的文化，与客户和合作伙伴齐心协力一起成长，共同发展。

通过将elasticsearch的data、ingest、master角色进行分离，搭建起高性能+高可用的ES架构

作者：“发颠的小狼”，欢迎转载与投稿

目录

▪ 用途
▪ 架构
▪ 步骤说明
▪ elasticsearch-data部署
▪ elasticsearch-ingest部署
▪ elasticsearch-master部署

用途

在第一篇《EFK教程 - 快速入门指南》中，阐述了EFK的安装部署，其中ES的架构为三节点，即master、ingest、data角色同时部署在三台服务器上。

在本文中，将进行角色分离部署，并且每个角色分别部署三节点，在实现性能最大化的同时保障高可用。

▷ elasticsearch的master节点：用于调度，采用普通性能服务器来部署
▷ elasticsearch的ingest节点：用于数据预处理，采用性能好的服务器来部署
▷ elasticsearch的data节点：用于数据落地存储，采用存储性能好的服务器来部署

若不知道去哪找《EFK教程 - 快速入门指南》，可在主流搜索引擎里搜索：
小慢哥 EFK教程 快速入门指南
或者
小慢哥 EFK教程 基于多节点ES的EFK安装部署配置

架构

服务器配置

注意：此处的架构是之前的文章《EFK教程 - 快速入门指南》的拓展，因此请先按照《EFK教程 - 快速入门指南》完成部署

步骤说明

1️⃣ 部署3台data节点，加入原集群
2️⃣ 部署3台ingest节点，加入原集群
3️⃣ 将原有的es索引迁移到data节点
4️⃣ 将原有的es节点改造成master节点

elasticsearch-data部署

之前已完成了基础的elasticsearch架构，现需要新增三台存储节点加入集群，同时关闭master和ingest功能

elasticsearch-data安装：3台均执行相同的安装步骤

tar -zxvf elasticsearch-7.3.2-linux-x86_64.tar.gz
mv elasticsearch-7.3.2 /opt/elasticsearch
useradd elasticsearch -d /opt/elasticsearch -s /sbin/nologin
mkdir -p /opt/logs/elasticsearch
chown elasticsearch.elasticsearch /opt/elasticsearch -R
chown elasticsearch.elasticsearch /opt/logs/elasticsearch -R
# 数据盘需要elasticsearch写权限
chown elasticsearch.elasticsearch /data/SAS -R

# 限制一个进程可以拥有的VMA(虚拟内存区域)的数量要超过262144，不然elasticsearch会报max virtual memory areas vm.max_map_count [65535] is too low, increase to at least [262144]
echo "vm.max_map_count = 655350" >> /etc/sysctl.conf
sysctl -p

elasticsearch-data配置

▷ 192.168.1.51 /opt/elasticsearch/config/elasticsearch.yml

cluster.name: my-application
node.name: 192.168.1.51
# 数据盘位置，如果有多个硬盘位置，用","隔开
path.data: /data/SAS
path.logs: /opt/logs/elasticsearch
network.host: 192.168.1.51

discovery.seed_hosts: ["192.168.1.31","192.168.1.32","192.168.1.33"]
cluster.initial_master_nodes: ["192.168.1.31","192.168.1.32","192.168.1.33"]
http.cors.enabled: true
http.cors.allow-origin: "*"

# 关闭master功能
node.master: false
# 关闭ingest功能
node.ingest: false
# 开启data功能
node.data: true

▷ 192.168.1.52 /opt/elasticsearch/config/elasticsearch.yml

cluster.name: my-application
node.name: 192.168.1.52
# 数据盘位置，如果有多个硬盘位置，用","隔开
path.data: /data/SAS
path.logs: /opt/logs/elasticsearch
network.host: 192.168.1.52

discovery.seed_hosts: ["192.168.1.31","192.168.1.32","192.168.1.33"]
cluster.initial_master_nodes: ["192.168.1.31","192.168.1.32","192.168.1.33"]
http.cors.enabled: true
http.cors.allow-origin: "*"

# 关闭master功能
node.master: false
# 关闭ingest功能
node.ingest: false
# 开启data功能
node.data: true

▷ 192.168.1.53 /opt/elasticsearch/config/elasticsearch.yml

cluster.name: my-application
node.name: 192.168.1.53
# 数据盘位置，如果有多个硬盘位置，用","隔开
path.data: /data/SAS
path.logs: /opt/logs/elasticsearch
network.host: 192.168.1.53

discovery.seed_hosts: ["192.168.1.31","192.168.1.32","192.168.1.33"]
cluster.initial_master_nodes: ["192.168.1.31","192.168.1.32","192.168.1.33"]
http.cors.enabled: true
http.cors.allow-origin: "*"

# 关闭master功能
node.master: false
# 关闭ingest功能
node.ingest: false
# 开启data功能
node.data: true

elasticsearch-data启动

sudo -u elasticsearch /opt/elasticsearch/bin/elasticsearch

elasticsearch集群状态

curl "http://192.168.1.31:9200/_cat/health?v"

elasticsearch-data状态

curl "http://192.168.1.31:9200/_cat/nodes?v"

elasticsearch-data参数说明

status: green  # 集群健康状态
node.total: 6  # 有6台机子组成集群
node.data: 6  # 有6个节点的存储
node.role: d  # 只拥有data角色
node.role: i  # 只拥有ingest角色
node.role: m  # 只拥有master角色
node.role: mid  # 拥master、ingest、data角色

elasticsearch-ingest部署

现需要新增三台ingest节点加入集群，同时关闭master和data功能

elasticsearch-ingest安装：3台es均执行相同的安装步骤

tar -zxvf elasticsearch-7.3.2-linux-x86_64.tar.gz
mv elasticsearch-7.3.2 /opt/elasticsearch
useradd elasticsearch -d /opt/elasticsearch -s /sbin/nologin
mkdir -p /opt/logs/elasticsearch
chown elasticsearch.elasticsearch /opt/elasticsearch -R
chown elasticsearch.elasticsearch /opt/logs/elasticsearch -R

# 限制一个进程可以拥有的VMA(虚拟内存区域)的数量要超过262144，不然elasticsearch会报max virtual memory areas vm.max_map_count [65535] is too low, increase to at least [262144]
echo "vm.max_map_count = 655350" >> /etc/sysctl.conf
sysctl -p

elasticsearch-ingest配置

▷ 192.168.1.41 /opt/elasticsearch/config/elasticsearch.yml

cluster.name: my-application
node.name: 192.168.1.41
path.logs: /opt/logs/elasticsearch
network.host: 192.168.1.41

discovery.seed_hosts: ["192.168.1.31","192.168.1.32","192.168.1.33"]
cluster.initial_master_nodes: ["192.168.1.31","192.168.1.32","192.168.1.33"]
http.cors.enabled: true
http.cors.allow-origin: "*"

# 关闭master功能
node.master: false
# 开启ingest功能
node.ingest: true
# 关闭data功能
node.data: false

▷ 192.168.1.42 /opt/elasticsearch/config/elasticsearch.yml

cluster.name: my-application
node.name: 192.168.1.42
path.logs: /opt/logs/elasticsearch
network.host: 192.168.1.42

discovery.seed_hosts: ["192.168.1.31","192.168.1.32","192.168.1.33"]
cluster.initial_master_nodes: ["192.168.1.31","192.168.1.32","192.168.1.33"]
http.cors.enabled: true
http.cors.allow-origin: "*"

# 关闭master功能
node.master: false
# 开启ingest功能
node.ingest: true
# 关闭data功能
node.data: false

▷ 192.168.1.43 /opt/elasticsearch/config/elasticsearch.yml

cluster.name: my-application
node.name: 192.168.1.43
path.logs: /opt/logs/elasticsearch
network.host: 192.168.1.43

discovery.seed_hosts: ["192.168.1.31","192.168.1.32","192.168.1.33"]
cluster.initial_master_nodes: ["192.168.1.31","192.168.1.32","192.168.1.33"]
http.cors.enabled: true
http.cors.allow-origin: "*"

# 关闭master功能
node.master: false
# 开启ingest功能
node.ingest: true
# 关闭data功能
node.data: false

elasticsearch-ingest启动

sudo -u elasticsearch /opt/elasticsearch/bin/elasticsearch

elasticsearch集群状态

curl "http://192.168.1.31:9200/_cat/health?v"

elasticsearch-ingest状态

curl "http://192.168.1.31:9200/_cat/nodes?v"

elasticsearch-ingest参数说明

status: green  # 集群健康状态
node.total: 9  # 有9台机子组成集群
node.data: 6  # 有6个节点的存储
node.role: d  # 只拥有data角色
node.role: i  # 只拥有ingest角色
node.role: m  # 只拥有master角色
node.role: mid  # 拥master、ingest、data角色

elasticsearch-master部署

首先，将上一篇《EFK教程 - 快速入门指南》中部署的3台es（192.168.1.31、192.168.1.32、192.168.1.33）改成只有master的功能， 因此需要先将这3台上的索引数据迁移到本次所做的data节点中

1️⃣ 索引迁移：一定要做这步，将之前的索引放到data节点上

curl -X PUT "192.168.1.31:9200/*/_settings?pretty" -H 'Content-Type: application/json' -d'
{
  "index.routing.allocation.include._ip": "192.168.1.51,192.168.1.52,192.168.1.53"
}'

2️⃣ 确认当前索引存储位置：确认所有索引不在192.168.1.31、192.168.1.32、192.168.1.33节点上

curl "http://192.168.1.31:9200/_cat/shards?h=n"

elasticsearch-master配置

注意事项：修改配置，重启进程，需要一台一台执行，要确保第一台成功后，再执行下一台。重启进程的方法：由于上一篇文章《EFK教程 - 快速入门指南》里，是执行命令跑在前台，因此直接ctrl - c退出再启动即可，启动命令如下

sudo -u elasticsearch /opt/elasticsearch/bin/elasticsearch

▷ 192.168.1.31 /opt/elasticsearch/config/elasticsearch.yml

cluster.name: my-application
node.name: 192.168.1.31
path.logs: /opt/logs/elasticsearch
network.host: 192.168.1.31

discovery.seed_hosts: ["192.168.1.31","192.168.1.32","192.168.1.33"]
cluster.initial_master_nodes: ["192.168.1.31","192.168.1.32","192.168.1.33"]
http.cors.enabled: true
http.cors.allow-origin: "*"

#开启master功能
node.master: true
#关闭ingest功能
node.ingest: false
#关闭data功能
node.data: false

▷ 192.168.1.32 /opt/elasticsearch/config/elasticsearch.yml

cluster.name: my-application
node.name: 192.168.1.32
path.logs: /opt/logs/elasticsearch
network.host: 192.168.1.32

discovery.seed_hosts: ["192.168.1.31","192.168.1.32","192.168.1.33"]
cluster.initial_master_nodes: ["192.168.1.31","192.168.1.32","192.168.1.33"]
http.cors.enabled: true
http.cors.allow-origin: "*"

#开启master功能
node.master: true
#关闭ingest功能
node.ingest: false
#关闭data功能
node.data: false

▷ 192.168.1.33 /opt/elasticsearch/config/elasticsearch.yml

cluster.name: my-application
node.name: 192.168.1.33
path.logs: /opt/logs/elasticsearch
network.host: 192.168.1.33

discovery.seed_hosts: ["192.168.1.31","192.168.1.32","192.168.1.33"]
cluster.initial_master_nodes: ["192.168.1.31","192.168.1.32","192.168.1.33"]
http.cors.enabled: true
http.cors.allow-origin: "*"

#开启master功能
node.master: true
#关闭ingest功能
node.ingest: false
#关闭data功能
node.data: false

elasticsearch集群状态

curl "http://192.168.1.31:9200/_cat/health?v"

elasticsearch-master状态

curl "http://192.168.1.31:9200/_cat/nodes?v"

至此，当node.role里所有服务器都不再出现“mid”，则表示一切顺利完成。