mongodb集群方式-分片+副本集方式

 

分片就是水平的扩展，将数据分拆到不同的机器上，以达到存储更多的数据，处理更大的负载。可以选定将按照指定的文档键值进行分片。

配置

整体部署方案：

启动三个分片服务，两个做数据存储，另一个作为config，

配置分片的副本集，创建管理员用户，关闭mongod，打开keyfile，启动mongos，配置分片服务。

角色

分片一般有三个组成部分：

分片服务(Shard Server)，mongod 实例，2个以上，负责存储实际的数据分片，生产环境中一个Shard Server可由几台服务器组成一个Replica Set代替，避免主机单点故障；
路由服务(Routing Process)，mongos实例，1个以上，它负责管理分片，客户端由此前端路由接入，且让整个集群看起来像单一数据库，客户端应用可以透明使用，Routing Process不存储数据，数据来自Config Server；

配置服务(Config Server)，mongod 实例，1个以上，负责存储整个集群的配置信息：即数据和片的对应关系。

 

因为测试资源有限， 采取三台虚拟机的方式，

 

分片的架构图如下：

192.168.100.101所有配置如下：

############config-1############
configsvr = true
replSet = config
port = 30001
dbpath = /opt/mongo/data/config-1
logpath = /opt/mongo/logs/config-1.log
logappend = true
fork = true
profile = 1
slowms = 500
keyFile = /opt/mongo/config/keyfile
maxConns=20000
storageEngine = wiredTiger
############route############
configdb = config/192.168.100.101:30001,192.168.100.102:30002,192.168.100.103:30003
port = 20000
logpath = /opt/mongo/logs/route.log
logappend = true
fork = true
#chunkSize = 256
keyFile = /opt/mongo/config/keyfile
maxConns=20000
############rs1-1############
port = 10001
fork = true
dbpath = /opt/mongo/data/rs1-1
logpath = /opt/mongo/logs/rs1-1.log
replSet = test1
logappend = true
profile = 1
slowms = 500
directoryperdb = true
keyFile = /opt/mongo/config/keyfile
maxConns=20000
storageEngine = wiredTiger
############rs2-a############
port = 20003
fork = true
dbpath = /opt/mongo/data/rs2-a
logpath = /opt/mongo/logs/rs2-a.log
replSet = test2
logappend = true
profile = 1
slowms = 500
directoryperdb = true
keyFile = /opt/mongo/config/keyfile
maxConns=20000
storageEngine = wiredTiger

192.168.100.102所有配置如下：

############config-2############
configsvr = true
replSet = config
port = 30002
dbpath = /opt/mongo/data/config-2
logpath = /opt/mongo/logs/config-2.log
logappend = true
fork = true
profile = 1
slowms = 500
keyFile = /opt/mongo/config/keyfile
maxConns=20000
storageEngine = wiredTiger
############route############
configdb = config/192.168.100.101:30001,192.168.100.102:30002,192.168.100.103:30003
port = 20000
logpath = /opt/mongo/logs/route.log
logappend = true
fork = true
#chunkSize = 256
keyFile = /opt/mongo/config/keyfile
maxConns=20000
############rs1-2############
port = 10002
fork = true
dbpath = /opt/mongo/data/rs1-2
logpath = /opt/mongo/logs/rs1-2.log
replSet = test1
logappend = true
profile = 1
slowms = 500
directoryperdb = true
keyFile = /opt/mongo/config/keyfile
maxConns=20000
storageEngine = wiredTiger
############rs2-2############
port = 20002
fork = true
dbpath = /opt/mongo/data/rs2-2
logpath = /opt/mongo/logs/rs2-2.log
replSet = test2
logappend = true
profile = 1
slowms = 500
directoryperdb = true
keyFile = /opt/mongo/config/keyfile
maxConns=20000
storageEngine = wiredTiger

192.168.100.103所有配置如下：

############config-3############
configsvr = true
replSet = config
port = 30003
dbpath = /opt/mongo/data/config-3
logpath = /opt/mongo/logs/config-3.log
logappend = true
fork = true
profile = 1
slowms = 500
keyFile = /opt/mongo/config/keyfile
maxConns=20000
storageEngine = wiredTiger
############route############
configdb = config/192.168.100.101:30001,192.168.100.102:30002,192.168.100.103:30003
port = 20000
logpath = /opt/mongo/logs/route.log
logappend = true
fork = true
#chunkSize = 256
keyFile = /opt/mongo/config/keyfile
maxConns=20000
############rs1-a############
port = 10003
fork = true
dbpath = /opt/mongo/data/rs1-a
logpath = /opt/mongo/logs/rs1-a.log
replSet = test1
logappend = true
profile = 1
slowms = 500
directoryperdb = true
keyFile = /opt/mongo/config/keyfile
maxConns=20000
storageEngine = wiredTiger
############rs2-1############
port = 20001
fork = true
dbpath = /opt/mongo/data/rs2-1
logpath = /opt/mongo/logs/rs2-1.log
replSet = test2
logappend = true
profile = 1
slowms = 500
directoryperdb = true
keyFile = /opt/mongo/config/keyfile
maxConns=20000
storageEngine = wiredTiger

 

openfile 可以使用命令生成：

openssl rand -base64 500 > keyfile
chmod 400 keyfile

确保每台机器的keyfile一致

 

依次按照上面配置文件的内容创建配置文件和存储目录日志目录

创建完成依次启动所有的mongod节点

注意此时要把keyfile选项注释掉，否则启动之后未创建角色就要认证，无法进入操作

ls |grep -v keyfile| xargs sed -i "s/^keyFile/#keyFile/g"

启动之后依次连接每个分片的主执行以下操作：

>config = {
    _id: "test1",
    members: [{
        _id: 0,
        host: "192.168.100.101:10001"
    }, {
        _id: 1,
        host: "192.168.100.102:10002"
    }, {
        _id: 2,
        host: "192.168.100.103:10003",
        arbiterOnly: true
    }]
}
>rs.initiate(config)
>rs.status()
>db.isMaster( )
>use admin;
>db.createRole({role:"superman", privileges:[{resource:{anyResource: true}, actions:["anyAction"]}], roles:["root"]})
>db.createUser({user:"test",pwd:"test",roles:[{role:"superman", db:"admin"}]})

依次执行成功之后，

停止所有的mongod节点:

for i in `seq 10`;do killall mongod ;done

开启keyfile

ls |grep -v keyfile| xargs sed -i "s/^#keyFile/keyFile/g"

之后依次启动mongod节点

 

然后启动mongos节点

连接mongos节点

mongos>use admin;
mongos>db.auth("test","test")
mongos>sh.addShard("test1/192.168.100.101:10001")
mongos>sh.addShard("test2/192.168.100.103:20001")
mongos>sh.status()

至此分片配置完成。