前言

团队新采购一批大内存服务器作为Elasticsaerch节点。根据ES官网文档，单机单实例最多不超64G内存，这批服务器用于单机单实例有些"大材小用" .由于之前用docker做ES集群碰到过坑, 所以这次采用实体跑多实例的方案。

环境

ES单机多实例有两种情况,一种为单集群多实例,另外一种为多集群多实例。 本篇主要介绍单机多集群多实例的情况。

由于大多节点配置等信息都雷同，本文用两台服务器做为示例

需求架构图

部署过程

下载

node1,node2

curl -O https://artifacts.elastic.co/downloads/elasticsearch/elasticsearch-7.0.0-x86_64.rpmrpm -ivh elasticsearch-7.0.0-x86_64.rpm

创建目录

node1,node2

mkdir -p /data/elasticsearch/{t2,t1}/{data,repo,log}chown elasticsearch. elasticsearch/ -R

拷贝配置文件

node1,node2

cd /etc/cp -avP elasticsearch/ t1es/cp -avP elasticsearch/ t2es/

配置文件

node1

/etc/t2es/elasticsearch.yml

cluster:    name: t2es    routing:        allocation:            same_shard.host: true    initial_master_nodes:        - t2-node-70        - t2-node-71        - ...node:    name: t2-node-70    master: true    data: true    max_local_storage_nodes: 2path:    data: /data/elasticsearch/t2/data    logs: /data/elasticsearch/t2/log    repo: /data/elasticsearch/t2/reponetwork:    host: t2-node-70http:    port: 9200transport.tcp.port: 9300discovery.zen.ping.unicast.hosts: ["t2-node-70:9300","t2-node-71:9300" ...]discovery.zen.fd.ping_timeout: "10s"discovery.zen.fd.ping_retries: 10reindex.remote.whitelist: "192.168.1.*:*"

/etc/t1es/elasticsearch.yml

cluster:    name: t1es    routing:        allocation:            same_shard.host: true    initial_master_nodes:        - t1-node-70        - t1-node-71        - ...node:    name: t1-node-70    master: true    data: true    max_local_storage_nodes: 2path:    data: /data/elasticsearch/t1/data    logs: /data/elasticsearch/t1/log    repo: /data/elasticsearch/t1/reponetwork:    host: t1-node-70http:    port: 9201transport.tcp.port: 9301discovery.zen.ping.unicast.hosts: ["t1-node-70:9301","t1-node-71:9301" ...]discovery.zen.fd.ping_timeout: "10s"discovery.zen.fd.ping_retries: 10reindex.remote.whitelist: "192.168.1.*:*"

node2

/etc/t2es/elasticsearch.yml

...network:    host: t1-node-70http:    port: 9201transport.tcp.port: 9301...

/etc/t1es/elasticsearch.yml

...network:    host: t1-node-70http:    port: 9201transport.tcp.port: 9301...

启动服务配置

node1,node2

cd /usr/lib/systemd/systemcp -avp elasticsearch.service  t2es.servicecd /etc/sysconfigcp -avp cp -avP elasticsearch  t2escp -avp cp -avP elasticsearch  t1es

修改启动服务配置

node1,node2

/usr/lib/systemd/system/t2es

[Unit]Description=t2-ElasticsearchDocumentation=http://www.elastic.coWants=network-online.targetAfter=network-online.target[Service]RuntimeDirectory=elasticsearchPrivateTmp=trueEnvironment=ES_HOME=/usr/share/elasticsearchEnvironment=ES_PATH_CONF=/etc/t2esEnvironment=PID_DIR=/var/run/elasticsearchEnvironment=ENV_PATH=/etc/sysconfig/t2es # EnvironmentFile=-/etc/sysconfig/elasticsearch  WorkingDirectory=/usr/share/elasticsearchUser=elasticsearchGroup=elasticsearchExecStart=/usr/share/elasticsearch/bin/elasticsearch -p ${PID_DIR}/t2es.pid --quietStandardError=inheritLimitNOFILE=65535LimitNPROC=4096LimitAS=infinityLimitFSIZE=infinityTimeoutStopSec=0KillSignal=SIGTERMKillMode=processSendSIGKILL=noSuccessExitStatus=143[Install]WantedBy=multi-user.target

/usr/lib/systemd/system/t1es

...Environment=ES_PATH_CONF=/etc/t2esEnvironment=PID_DIR=/var/run/elasticsearchEnvironment=ENV_PATH=/etc/sysconfig/t2es # EnvironmentFile=-/etc/sysconfig/elasticsearch  WorkingDirectory=/usr/share/elasticsearchUser=elasticsearchGroup=elasticsearchExecStart=/usr/share/elasticsearch/bin/elasticsearch -p ${PID_DIR}/t2es.pid --quiet...

编辑/usr/share/elasticsearch/bin/elasticsearch-env

修改 source /etc/sysconfig/elasticsearch

if [ ! -z ${ENV_PATH} ]then    source ${ENV_PATH} else    source /etc/sysconfig/elasticsearchfi

/etc/sysconfig/t2es

所有参数配置可以在这里面加

ES_PATH_CONF=/etc/t2esES_STARTUP_SLEEP_TIME=5

启动

node1,node2

systemctl t1es startsystemctl t2es start

总结

• Elasticsearch 7.x系列需要配置initial_master_nodes.
• cluster.routing.allocation.same_shard.host防止一台主机存在多个相同切片.
• node.max_local_storage_nodes配置一台主机可以存在的实例数量.