mongodb分片

Mongodb 分片部署

配置mongodb集群，比如 在3个server上配置 3 shard的Mongodb集群：

架构：

1.每片数据需要3个mongod server，2个为主从数据节点；1个为仲裁节点(arbiter)，不存数据。

   一共三片，可以做成：  

   20.220-mongo1:   20001(sh1主)，20002(sh2从)，20003(sh3仲裁)

   20.221-mongo2:   20002(sh2主)，20003(sh3从)，20001(sh1仲裁)

   20.222-mongo3:   20003(sh3主)，20001(sh1从)，20002(sh2仲裁)

2. 需要3个mongod config server，登录3台机器执行如下

   20.220-mongo1:19999

         echo'export PATH=$PATH:/usr/local/mongodb/bin' >> /etc/profile

tar zxvfmongodb-linux-x86_64-2.6.0.tgz

mvmongodb-linux-x86_64-2.6.0 /usr/local/mongodb

mkdir -p/usr/local/mongodb/{etc,date,log}

cd /usr/local/mongodb/date

mkdir sh1

mkdir sh2

mkdir sh3

mkdir cf1

mkdir cf2

mkdir cf3

20.221-mongo2:19998

    echo 'exportPATH=$PATH:/usr/local/mongodb/bin' >> /etc/profile

tar zxvfmongodb-linux-x86_64-2.6.0.tgz

mvmongodb-linux-x86_64-2.6.0 /usr/local/mongodb

mkdir -p /usr/local/mongodb/{etc,date,log}

cd /usr/local/mongodb/date

mkdir sh1

mkdir sh2

mkdir sh3

mkdir cf1

mkdir cf2

mkdir cf3

   20.222-mongo3:19997

         echo'export PATH=$PATH:/usr/local/mongodb/bin' >> /etc/profile

tar zxvfmongodb-linux-x86_64-2.6.0.tgz

mvmongodb-linux-x86_64-2.6.0 /usr/local/mongodb

mkdir -p/usr/local/mongodb/{etc,date,log}

cd /usr/local/mongodb/date

mkdir sh1

mkdir sh2

mkdir sh3

mkdir cf1

mkdir cf2

mkdir cf3

3. mongos server 作为入口

   192.168.20.220:20000

实施步骤：

1.启动mongod datanode service

分别3台服务器上执行如下：

/usr/local/mongodb/bin/mongod --fork --rest--replSet sh1 --shardsvr --oplogSize 40000 --dbpath /usr/local/mongodb/date/sh1--logpath /usr/local/mongodb/log/sh1.log --port 20001

/usr/local/mongodb/bin/mongod --fork --rest--replSet sh2 --shardsvr --oplogSize 40000 --dbpath /usr/local/mongodb/date/sh2--logpath /usr/local/mongodb/log/sh2.log --port 20002

/usr/local/mongodb/bin/mongod --fork --rest--replSet sh3 --shardsvr --oplogSize 40000 --dbpath /usr/local/mongodb/date/sh3--logpath /usr/local/mongodb/log/sh3.log --port 20003

2.配置每片的 replica set，即每3台配成一个replica set互备

登录

mongo 192.168.20.220:20001/admin

config = {_id: 'sh1', members:[

{_id:0, host: '192.168.20.220:20001'},

{_id:1, host: '192.168.20.221:20001'},

{_id: 2, host: '192.168.20.222:20001',arbiterOnly: true}

]}

mongo 192.168.20.221:20002/admin

config = {_id: 'sh2', members:[

{_id:0, host: '192.168.20.220:20002'},

{_id:1, host: '192.168.20.221:20002'},

{_id: 2, host: '192.168.20.222:20002',arbiterOnly: true}

]}

mongo 192.168.20.222:20003/admin

config = {_id: 'sh3', members:[

{_id:0, host: '192.168.20.220:20003'},

{_id:1, host: '192.168.20.221:20003'},

{_id: 2, host: '192.168.20.222:20003',arbiterOnly: true}

]}

分别在3台机器上执行如下

rs.initiate(config) //初始化replset

rs.status()   //看结果是否成功建成repl set

3.启动 config server:

/usr/local/mongodb/bin/mongod --fork--configsvr --port 19999 --dbpath /usr/local/mongodb/date/cf1/ --logpath/usr/local/mongodb/log/cf1.log --rest

/usr/local/mongodb/bin/mongod --fork--configsvr --port 19998 --dbpath /usr/local/mongodb/date/cf2/ --logpath/usr/local/mongodb/log/cf2.log --rest

/usr/local/mongodb/bin/mongod --fork--configsvr --port 19997 --dbpath /usr/local/mongodb/date/cf3/ --logpath/usr/local/mongodb/log/cf3.log --rest

4. 启动  mongos server

/usr/local/mongodb19999/bin/mongos--configdb 192.168.20.220:19999,192.168.20.221:19998,192.168.20.222:19997--logpath /usr/local/mongodb19999/log/mongos.log --logappend --fork --port20000

5. 配置mongo 分片集群

登录mongos:  

mongo 192.168.20.220:20000/admin

运行：

>db.runCommand({addshard:"sh1/l-mo1.prod.cn2:20001,l-mo2.prod.cn2:20001",name:"sh1"});

>db.runCommand({addshard:"sh2/l-mo2.prod.cn2:20002,l-mo3.prod.cn2:20002",name:"sh2"});

>db.runCommand({addshard:"sh3/l-mo3.prod.cn2:20003,l-mo1.prod.cn2:20003",name:"sh3"});

> db.runCommand({listshards:1})

6.配置 库表分片

登录mongos:

>use admin;

>db.runCommand( { enablesharding :"dbname" } );

>db.runCommand( { shardcollection :"dbname.collectionname", key :{ "keyfield" :1 }});

注意分片使用的keyfield需要是表索引。

相比而言，压力不高，数据量不大的情况下，还是部署副本集比较方便