【ElasticSearch】03 部署

Windows集群部署：

把包解压成三个节点【kibana-7.16.3-windows-x86_64.zip】

- Node-1001

- Node-1002

- Node-1003

修改集群的配置信息

- Node-1001\config\elasticsearch.yml

- Node-1002\config\elasticsearch.yml

- Node-1003\config\elasticsearch.yml

配置信息：

============================== Node-1001 ==========================

# 集群名称一致

cluster.name: Cloud9-WIN-ES-CAT

#节点名称，集群内要唯一

node.name: node-1001

node.master: true

node.data: true

# ip 地址

network.host: localhost

# http 端口

http.port: 1001

# tcp 监听端口

transport.tcp.port: 9301

# 跨域配置

http.cors.enabled: true

http.cors.allow-origin: "*"

============================== Node-1002 ==========================

# 集群名称一致

cluster.name: Cloud9-WIN-ES-CAT

#节点名称，集群内要唯一

node.name: node-1002

node.master: true

node.data: true

# 候选主节点

discovery.seed_hosts: ["localhost:9301"]

discovery.zen.fd.ping_timeout: 1m

discovery.zen.fd.ping_retries: 5

# ip 地址

network.host: localhost

# http 端口

http.port: 1002

# tcp 监听端口

transport.tcp.port: 9302

# 跨域配置

http.cors.enabled: true

http.cors.allow-origin: "*"

============================== Node-1003 ==========================

# 集群名称一致

cluster.name: Cloud9-WIN-ES-CAT

#节点名称，集群内要唯一

node.name: node-1003

node.master: true

node.data: true

#候选主节点的地址，在开启服务后可以被选为主节点

discovery.seed_hosts: ["localhost:9301", "localhost:9302"]

discovery.zen.fd.ping_timeout: 1m

discovery.zen.fd.ping_retries: 5

# ip 地址

network.host: localhost

# http 端口

http.port: 1003

# tcp 监听端口

transport.tcp.port: 9303

# 跨域配置

http.cors.enabled: true

http.cors.allow-origin: "*"

(可选)编写辅助启动脚本【start-cluster.cmd】：

@echo off

start %CD%\Node-1001\bin\elasticsearch.bat

start %CD%\Node-1002\bin\elasticsearch.bat

start %CD%\Node-1003\bin\elasticsearch.bat

(可选)编写辅助清除脚本【clear-nodes.cmd】：

@ECHO OFF

RD /S /Q %CD%\Node-1001\data\nodes

RD /S /Q %CD%\Node-1002\data\nodes

RD /S /Q %CD%\Node-1003\data\nodes

启动之前，删除data目录下的nodes目录

然后依次启动

bin\elasticsearch.bat

请求节点地址，查看节点运行状态【GET请求】

http://localhost:1001/_cluster/health

http://localhost:1002/_cluster/health

http://localhost:1003/_cluster/health

响应这些信息表示正常

{

    "cluster_name": "Cloud9-WIN-ES-CAT",

    "status": "green",

    "timed_out": false,

    "number_of_nodes": 1,

    "number_of_data_nodes": 1,

    "active_primary_shards": 3,

    "active_shards": 3,

    "relocating_shards": 0,

    "initializing_shards": 0,

    "unassigned_shards": 0,

    "delayed_unassigned_shards": 0,

    "number_of_pending_tasks": 0,

    "number_of_in_flight_fetch": 0,

    "task_max_waiting_in_queue_millis": 0,

    "active_shards_percent_as_number": 100.0

}

或者是查看cat/nodes?v查看集群状态

http://localhost:1001/_cat/nodes?v

http://localhost:1002/_cat/nodes?v

http://localhost:1003/_cat/nodes?v

响应的信息：

ip        heap.percent ram.percent cpu load_1m load_5m load_15m node.role   master name

127.0.0.1           20          97  19                          cdfhilmrstw -      node-1003

127.0.0.1           21          97  19                          cdfhilmrstw -      node-1002

127.0.0.1           16          97  19                          cdfhilmrstw *      node-1001

# 测试集群是否是相互通信的？

# 在1001节点创建一个sample索引

PUT http://localhost:1001/sample

# 然后去1002节点和1003节点上面查看sample索引信息
# 如果其他节点能够找到1001节点的索引，说明集群是相互通信的

GET http://localhost:1003/sample

GET http://localhost:1002/sample

Linux集群部署：

首先提供集群服务环境（每台虚拟机最少提供2G内存，内存过低将导致服务无法集群通信）

192.168.177.128

192.168.177.129

192.168.177.130

每台机器都需要有ES的包

elasticsearch-7.16.0-linux-x86_64.tar.gz

解压包到/opt/module目录下面

mkdir -p /opt/module/

tar -zxvf elasticsearch-7.16.0-linux-x86_64.tar.gz -C /opt/module/

# （可选）变更目录名称

mv /opt/module/elasticsearch-7.16.0 /opt/module/es

# 创建elasticsearch用户, 设置密码123456

useradd user-es

passwd user-es

# 权限赋予

chown -R user-es:user-es /opt/module/elasticsearch-7.16.0/

# 编辑集群参数

mv /opt/module/elasticsearch-7.16.0/config/elasticsearch.yml /opt/module/elasticsearch-7.16.0/config/elasticsearch.yml.bak

vim /opt/module/elasticsearch-7.16.0/config/elasticsearch.yml

= = = = = 1号节点 192.168.177.128 = = = = =

# 集群名称

cluster.name: cluster-es

# 节点名称，每个节点的名称不能重复

node.name: node-1

# ip地址，每个节点的地址不能重复

network.host: 192.168.177.128

# 是否具有资格主节点

node.master: true

node.data: true

http.port: 9200

# head 插件需要这打开这两个配置

http.cors.allow-origin: "*"

http.cors.enabled: true

http.max_content_length: 200mb

# es7.x 之后新增的配置，初始化一个新的集群时需要此配置来选举 master

cluster.initial_master_nodes: ["node-1"]

# es7.x 之后新增的配置，节点发现

discovery.seed_hosts: ["192.168.177.128:9300","192.168.177.129:9300","192.168.177.130:9300"]

gateway.recover_after_nodes: 2

network.tcp.keep_alive: true

network.tcp.no_delay: true

transport.tcp.compress: true

# 集群内同时启动的数据任务个数，默认是 2 个

cluster.routing.allocation.cluster_concurrent_rebalance: 16

# 添加或删除节点及负载均衡时并发恢复的线程个数，默认 4 个

cluster.routing.allocation.node_concurrent_recoveries: 16

# 初始化数据恢复时，并发恢复线程的个数，默认 4 个

cluster.routing.allocation.node_initial_primaries_recoveries: 16

= = = = = 2号节点 192.168.177.129 = = = = =

# 集群名称

cluster.name: cluster-es

# 节点名称，每个节点的名称不能重复

node.name: node-2

# ip地址，每个节点的地址不能重复

network.host: 192.168.177.129

# 是否具有资格主节点

node.master: true

node.data: true

http.port: 9200

# head 插件需要这打开这两个配置

http.cors.allow-origin: "*"

http.cors.enabled: true

http.max_content_length: 200mb

# es7.x 之后新增的配置，初始化一个新的集群时需要此配置来选举 master

cluster.initial_master_nodes: ["node-1"]

# es7.x 之后新增的配置，节点发现

discovery.seed_hosts: ["192.168.177.128:9300","192.168.177.129:9300","192.168.177.130:9300"]

gateway.recover_after_nodes: 2

network.tcp.keep_alive: true

network.tcp.no_delay: true

transport.tcp.compress: true

# 集群内同时启动的数据任务个数，默认是 2 个

cluster.routing.allocation.cluster_concurrent_rebalance: 16

# 添加或删除节点及负载均衡时并发恢复的线程个数，默认 4 个

cluster.routing.allocation.node_concurrent_recoveries: 16

# 初始化数据恢复时，并发恢复线程的个数，默认 4 个

cluster.routing.allocation.node_initial_primaries_recoveries: 16

= = = = = 3号节点 192.168.177.130 = = = = =

# 集群名称

cluster.name: cluster-es

# 节点名称，每个节点的名称不能重复

node.name: node-3

# ip地址，每个节点的地址不能重复

network.host: 192.168.177.130

# 是否具有资格主节点

node.master: true

node.data: true

http.port: 9200

# head 插件需要这打开这两个配置

http.cors.allow-origin: "*"

http.cors.enabled: true

http.max_content_length: 200mb

# es7.x 之后新增的配置，初始化一个新的集群时需要此配置来选举 master

cluster.initial_master_nodes: ["node-1"]

# es7.x 之后新增的配置，节点发现

discovery.seed_hosts: ["192.168.177.128:9300","192.168.177.129:9300","192.168.177.130:9300"]

gateway.recover_after_nodes: 2

network.tcp.keep_alive: true

network.tcp.no_delay: true

transport.tcp.compress: true

# 集群内同时启动的数据任务个数，默认是 2 个

cluster.routing.allocation.cluster_concurrent_rebalance: 16

# 添加或删除节点及负载均衡时并发恢复的线程个数，默认 4 个

cluster.routing.allocation.node_concurrent_recoveries: 16

# 初始化数据恢复时，并发恢复线程的个数，默认 4 个

cluster.routing.allocation.node_initial_primaries_recoveries: 16

# 解除进程限制

vim /etc/security/limits.conf

= = = = = 配置参数 = = = =

# 每个进程可以打开的文件数的限制

user-es soft nofile 65536

user-es hard nofile 65536

vim /etc/security/limits.d/20-nproc.conf

= = = = = 配置参数 = = = =

# 每个进程可以打开的文件数的限制

user-es soft nofile 65536

user-es hard nofile 65536

# 操作系统级别对每个用户创建的进程数的限制

* hard nproc 4096

# 注：* 带表 Linux 所有用户名称

vim /etc/sysctl.conf

= = = = = 配置参数 = = = =

# 在文件中增加下面内容

# 一个进程可以拥有的 VMA(虚拟内存区域)的数量,默认值为 65536

vm.max_map_count=655360

# 重新加载

sysctl -p

# 关闭防火墙

systemctl stop firewalld

# 切换ES用户，启动elasticsearch集群

su user-es

# 清空遗留的节点数据
# 所有节点都执行一遍

rm -rf /opt/module/elasticsearch-7.16.0/data/nodes

# 启动elasticsearch

/opt/module/elasticsearch-7.16.0/bin/elasticsearch -d

# 访问

http://192.168.177.128:9200/_cat/nodes?v

http://192.168.177.129:9200/_cat/nodes?v

http://192.168.177.130:9200/_cat/nodes?v

# 如果每个地址访问都是下面一样的内容，则集群部署成功

ip              heap.percent ram.percent cpu load_1m load_5m load_15m node.role   master name

192.168.177.128           32          94   2    0.28    0.62     0.32 cdfhilmrstw *      node-1

192.168.177.129           25          95   7    0.19    0.65     0.37 cdfhilmrstw -      node-2

192.168.177.130           20          94   0    0.24    0.58     0.31 cdfhilmrstw -      node-3