实战mongodb3.06 Relica Sets+sharding集群
实战mongodb3.06 Relica Sets+sharding集群
前 言
MongoDB 的Sharding机制解决了海量存储和动态扩容的问题，但离实际生产环境所需要的高可靠、高可用还有些距离，例如Shard Server的单点故障就无法解决，所以提出”ReplicatSets +Sharding”的解决方案。本方案是某某公司真实实例介绍采用MongoDB复制集和Sharding高可能用方案。本方案采用MongoDB 3.06版本。
MongoDB3.0以上版本提升7到10倍的写入效率以及增加80%的数据压缩率,还能减少95%的运维成本 。
MonoDB3.0新特性主括：可插入式的存储引擎API、支持WiredTiger存储引擎、MMAPv1提升、复制集全面提升、集群方面的改进、提升了安全性。
MongoDB3.0以上版本拥有大幅度的改进，本作者所以采用最新的3.06版本讲解。本实例采用最新的配置文件，在官方网可以看到。
1．Relica Sets+sharding架构
Replica Sets+sharding解决方案内容如下：
Shard服务器：使用Replica Sets确保每个数据节点都具有备份、自动容错转移、自动恢复能力
配置服务器：使用3个配置服务器确保元数据完整性。
路由进程：使用3个路由进程实现负载均衡，提高客户端接入性能。
配置完成的Replica Sets+sharding环境如下图所示。
2．搭建一个高可用架构
采用Replica Sets+sharding 架构，可以避免单机Sharding架构中的ShardServer单点故障，这方案组合解决的sharding架构中的高可用问题。
各服务器开放的监听端口如图所示。
|
主机 |
IP |
服务及端口 |
|
Mongodb01 |
172.16.202.201 |
Mongod shard1_1 11731 |
|
Mongod shard2_1 11732 | ||
|
Mongod shard3_1 11733 | ||
|
Mongod config 30000 | ||
|
Mongos 1 60000 | ||
|
Mongodb02 |
172.16.202.202 |
Mongod shard1_2 11731 |
|
Mongod shard2_2 11732 | ||
|
Mongod shard3_2 11733 | ||
|
Mongod config 30000 | ||
|
Mongos 2 60000 | ||
|
Mongodb03 |
172.16.202.203 |
Mongod shard1_3 11731 |
|
Mongod shard2_3 11732 | ||
|
Mongod shard3_3 11733 | ||
|
Mongod config 30000 | ||
|
Mongos 3 60000 |
2.1.创建mongo用户
在三台服务器中创建mongo用户，如下面的代码所示
[[email protected] ~]# useradd mongo
[[email protected] ~]#passwd mongo
[[email protected] ~]# su - mongo
[[email protected] ~]$
2.2.创建数据目录
首先要在mongo用户下创建shard server和Config Server的数据目录，用于存储数据，创建logs的日志目录、创建config存放配置文件目录。
在mongodb01上创建shard server和Config Server的数据目录、logs的日志目录、config存放配置文件目录。
[[email protected] ~]$ mkdir -p /home/mongo/data/shard1_1
[[email protected] ~]$ mkdir -p /home/mongo/data/shard2_1
[[email protected] ~]$ mkdir -p /home/mongo/data/shard3_1
[[email protected] ~]$ mkdir -p /home/mongo/data/config
[[email protected] ~]$ mkdir -p /home/mongo/data/logs
[[email protected] ~]$ mkdir -p /home/mongo/config
如以上代码所示，目录/home/mongo/data/shard1_1供 shard1主节点使用，目录/home/mongo/data/shard2_1供shard2仲裁使用，目录/home/mongo/data/shard3_1 供shard3副本使用，目录/home/mongo/data/config 供整个ReplicaSets+sharding架构中的1个configServer使用，目录/home/mongo/data/logs供日志使用，
目录/home/mongo/config供配置文件使用。
在mongodb02上创建shard server和Config Server的数据目录、logs的日志目录、config存放配置文件目录。
[[email protected] ~]$ mkdir -p /home/mongo/data/shard1_2
[[email protected] ~]$ mkdir -p /home/mongo/data/shard2_2
[[email protected] ~]$ mkdir -p /home/mongo/data/shard3_2
[[email protected] ~]$ mkdir -p /home/mongo/data/config
[[email protected] ~]$ mkdir -p /home/mongo/data/logs
[[email protected] ~]$ mkdir -p /home/mongo/config
如以上代码所示，目录/home/mongo/data/shard1_2供 shard1副本使用，目录/home/mongo/data/shard2_2供shard2主节点使用，目录/home/mongo/data/shard3_1 供shard3仲裁使用，目录/home/mongo/data/config 供整个ReplicaSets+sharding架构中的1个configServer使用，目录/home/mongo/data/logs供日志使用，
目录/home/mongo/config供配置文件使用。
在mongodb03上创建shard server和Config Server的数据目录、logs的日志目录、config存放配置文件目录。
[[email protected] ~]$ mkdir -p /home/mongo/data/shard1_3
[[email protected] ~]$ mkdir -p /home/mongo/data/shard2_3
[[email protected] ~]$ mkdir -p /home/mongo/data/shard3_3
[[email protected] ~]$ mkdir -p /home/mongo/data/config
[[email protected] ~]$ mkdir -p /home/mongo/data/logs
[[email protected] ~]$ mkdir -p /home/mongo/config
如以上代码所示，目录/home/mongo/data/shard1_3供 shard1仲裁使用，目录/home/mongo/data/shard2_3供shard2副本使用，目录/home/mongo/data/shard3_3 供shard3主节点使用，目录/home/mongo/data/config 供整个ReplicaSets+sharding架构中的1个configServer使用，目录/home/mongo/data/logs供日志使用，
目录/home/mongo/config供配置文件使用。
2.3.配置Replica Sets
在三台服务器上解压mongodb-linux-x86_64-3.0.6.tgz
[[email protected] ~]$ tar zxvfmongodb-linux-x86_64-3.0.6.tgz
[[email protected] ~]$ mvmongodb-linux-x86_64-3.0.6 mongodb
2.3.1.配置shard1所用到的Relica Set 1
#注意配置文件缩进
在mongodb01上操作，如下的代码所示：
[[email protected] ~]$ cd config/
[[email protected] config]$ cat shard1_1.conf
systemLog:
destination: file
##Log
path:/home/mongo/data/logs/shard1_1.log
logAppend: true
storage:
journal:
enabled: true
dbPath: /home/mongo/data/shard1_1
directoryPerDB: true
engine: wiredTiger
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes:true
collectionConfig:
blockCompressor: snappy
processManagement:
fork: true
net:
bindIp: 172.16.202.201
port: 11731
replication:
oplogSizeMB: 500
replSetName: shard1
sharding:
clusterRole: shardsvr
#sercurity:
#authorization: enabled
#keyFile:/home/mongo/key/security
[[email protected] config]$ /home/mongo/mongodb/bin/mongod -f/home/mongo/config/shard1_1.conf
如以上代码所示，在mongodb01启动Replica Set1中的1个成员节点，复制集名字是shard1,监听端口是11731。
在mongodb02上操作，如下的代码所示：
[[email protected] ~]$ cd config/
[[email protected] config]$ cat shard1_2.conf
systemLog:
destination: file
##Log
path: /home/mongo/data/logs/shard1_2.log
logAppend: true
storage:
journal:
enabled: true
dbPath: /home/mongo/data/shard1_2
directoryPerDB: true
engine: wiredTiger
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes:true
collectionConfig:
blockCompressor: snappy
processManagement:
fork: true
net:
bindIp: 172.16.202.202
port: 11731
replication:
oplogSizeMB: 500
replSetName: shard1
sharding:
clusterRole: shardsvr
#sercurity:
#authorization: enabled
#keyFile:/home/mongo/key/security
[[email protected] config]$ /home/mongo/mongodb/bin/mongod -f/home/mongo/config/shard1_2.conf
如以上代码所示，在mongodb2启动Replica Set1中的1个成员节点，复制集名字是shard1,监听端口是11731。
在mongodb03上操作，如下的代码所示：
[[email protected]~]$ cd config/
[[email protected] config]$ cat shard1_3.conf
systemLog:
destination: file
##Log
path: /home/mongo/data/logs/shard1_3.log
logAppend: true
storage:
journal:
enabled: true
dbPath: /home/mongo/data/shard1_3
directoryPerDB: true
engine: wiredTiger
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes:true
collectionConfig:
blockCompressor: snappy
processManagement:
fork: true
net:
bindIp: 172.16.202.203
port: 11731
replication:
oplogSizeMB: 500
replSetName: shard1
sharding:
clusterRole: shardsvr
#sercurity:
#authorization: enabled
#keyFile:/home/mongo/key/security
[[email protected] config]$ /home/mongo/mongodb/bin/mongod -f/home/mongo/config/shard1_3.conf
如以上代码所示，在mongodb3启动Replica Set1中的1个成员节点，复制集名字是shard1,监听端口是11731。
连接mongodb01的11731端口的mongod，初始化Replicat Set1,如下代码所示：
[[email protected] ~]$ /home/mongo/mongodb/bin/mongo 172.16.202.201:11731
MongoDB shell version: 3.0.6
connecting to: 172.16.202.201:11731/test
> config={_id:'shard1',members:[{_id:0,host:'172.16.202.201:11731',priority:2},{_id:1,host:'172.16.202.202:11731'},{_id:2,host:'172.16.202.203:11731',arbiterOnly:true}]}
{
"_id": "shard1",
"members": [
{
"_id": 0,
"host": "172.16.202.201:11731",
"priority": 2
},
{
"_id": 1,
"host": "172.16.202.202:11731"
},
{
"_id": 2,
"host": "172.16.202.203:11731",
"arbiterOnly": true
}
]
}
>rs.initiate(config)
{ "ok" : 1 }
以上代码通过执行rs.initiate(config)命令来初始化shard1的复制集Replica Set 1。
2.3.2.配置shard2所用到的Relica Set 2
在mongodb01上操作，如下的代码所示：
[[email protected] ~]$ cd config/
[[email protected] config]$ cat shard2_1.conf
systemLog:
destination: file
##Log
path: /home/mongo/data/logs/shard2_1.log
logAppend: true
storage:
journal:
enabled: true
dbPath: /home/mongo/data/shard2_1
directoryPerDB: true
engine: wiredTiger
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes:true
collectionConfig:
blockCompressor: snappy
processManagement:
fork: true
net:
bindIp: 172.16.202.201
port: 11732
replication:
oplogSizeMB: 500
replSetName: shard2
sharding:
clusterRole: shardsvr
#sercurity:
#authorization: enabled
#keyFile:/home/mongo/key/security
[[email protected] config]$ /home/mongo/mongodb/bin/mongod -f/home/mongo/config/shard2_1.conf
如以上代码所示，在mongodb01启动Replica Set1中的1个成员节点，复制集名字是shard2,监听端口是11732。
在mongodb02上操作，如下的代码所示：
[[email protected] ~]$ cd config/
[[email protected] config]$ cat shard2_2.conf
systemLog:
destination: file
##Log
path: /home/mongo/data/logs/shard2_2.log
logAppend: true
storage:
journal:
enabled: true
dbPath: /home/mongo/data/shard2_2
directoryPerDB: true
engine: wiredTiger
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes:true
collectionConfig:
blockCompressor: snappy
processManagement:
fork: true
net:
bindIp: 172.16.202.202
port: 11732
replication:
oplogSizeMB: 500
replSetName: shard2
sharding:
clusterRole: shardsvr
#sercurity:
#authorization: enabled
#keyFile:/home/mongo/key/security
[[email protected] config]$ /home/mongo/mongodb/bin/mongod -f/home/mongo/config/shard2_2.conf
如以上代码所示，在mongodb2启动Replica Set1中的1个成员节点，复制集名字是shard2,监听端口是11732。
在mongodb03上操作，如下的代码所示：
[[email protected]~]$ cd config/
[[email protected] config]$ cat shard2_3.conf
systemLog:
destination: file
##Log
path: /home/mongo/data/logs/shard2_3.log
logAppend: true
storage:
journal:
enabled: true
dbPath: /home/mongo/data/shard2_3
directoryPerDB: true
engine: wiredTiger
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes:true
collectionConfig:
blockCompressor: snappy
processManagement:
fork: true
net:
bindIp: 172.16.202.203
port: 11732
replication:
oplogSizeMB: 500
replSetName: shard2
sharding:
clusterRole: shardsvr
#sercurity:
#authorization: enabled
#keyFile:/home/mongo/key/security
[[email protected] config]$ /home/mongo/mongodb/bin/mongod -f/home/mongo/config/shard2_3.conf
如以上代码所示，在mongodb3启动Replica Set1中的1个成员节点，复制集名字是shard2,监听端口是11732。
连接mongodb02的11732端口的mongod，初始化Replicat Set1,如下代码所示：
[[email protected] config]$/home/mongo/mongodb/bin/mongo 172.16.202.202:11732
MongoDB shell version: 3.0.6
connecting to: 172.16.202.202:11732/test
> config={_id:'shard2',members:[{_id:0,host:'172.16.202.201:11732',arbiterOnly:true},{_id:1,host:'172.16.202.202:11732',priority:2},{_id:2,host:'172.16.202.203:11732'}]}
{
"_id": "shard2",
"members": [
{
"_id": 0,
"host": "172.16.202.201:11732",
"arbiterOnly": true
},
{
"_id": 1,
"host": "172.16.202.202:11732",
"priority": 2
},
{
"_id": 2,
"host": "172.16.202.203:11732"
}
]
}
> rs.initiate(config)
{ "ok" : 1 }
以上代码通过执行rs.initiate(config)命令来初始化shard2的复制集Replica Set 1。
2.3.3.配置shard3所用到的Relica Set 3
在mongodb01上操作，如下的代码所示：
[[email protected] ~]$ cd config/
[[email protected] config]$ cat shard3_1.conf
systemLog:
destination: file
##Log
path: /home/mongo/data/logs/shard3_1.log
logAppend: true
storage:
journal:
enabled: true
dbPath: /home/mongo/data/shard3_1
directoryPerDB: true
engine: wiredTiger
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes:true
collectionConfig:
blockCompressor: snappy
processManagement:
fork: true
net:
bindIp: 172.16.202.201
port: 11733
replication:
oplogSizeMB: 500
replSetName: shard3
sharding:
clusterRole: shardsvr
#sercurity:
#authorization: enabled
#keyFile:/home/mongo/key/security
[[email protected] config]$/home/mongo/mongodb/bin/mongod -f /home/mongo/config/shard3_1.conf
如以上代码所示，在mongodb01启动Replica Set1中的1个成员节点，复制集名字是shard3,监听端口是11733。
在mongodb02上操作，如下的代码所示：
[[email protected] ~]$ cd config/
[[email protected] config]$ cat shard3_2.conf
systemLog:
destination: file
##Log
path: /home/mongo/data/logs/shard3_2.log
logAppend: true
storage:
journal:
enabled: true
dbPath: /home/mongo/data/shard3_2
directoryPerDB: true
engine: wiredTiger
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes:true
collectionConfig:
blockCompressor: snappy
processManagement:
fork: true
net:
bindIp: 172.16.202.202
port: 11733
replication:
oplogSizeMB: 500
replSetName: shard3
sharding:
clusterRole: shardsvr
#sercurity:
#authorization: enabled
#keyFile:/home/mongo/key/security
[[email protected] config]$ /home/mongo/mongodb/bin/mongod -f/home/mongo/config/shard3_2.conf
如以上代码所示，在mongodb2启动Replica Set1中的1个成员节点，复制集名字是shard3,监听端口是11733。
在mongodb03上操作，如下的代码所示：
[[email protected]~]$ cd config/
[[email protected] config]$ cat shard3_3.conf
systemLog:
destination: file
##Log
path: /home/mongo/data/logs/shard3_3.log
logAppend: true
storage:
journal:
enabled: true
dbPath: /home/mongo/data/shard3_3
directoryPerDB: true
engine: wiredTiger
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes:true
collectionConfig:
blockCompressor: snappy
processManagement:
fork: true
net:
bindIp: 172.16.202.203
port: 11733
replication:
oplogSizeMB: 500
replSetName: shard3
sharding:
clusterRole: shardsvr
#sercurity:
#authorization: enabled
#keyFile:/home/mongo/key/security
[[email protected] config]$ /home/mongo/mongodb/bin/mongod -f/home/mongo/config/shard3_3.conf
如以上代码所示，在mongodb3启动Replica Set1中的1个成员节点，复制集名字是shard3,监听端口是11733。
连接mongodb03的11733端口的mongod，初始化Replicat Set1,如下代码所示：
[[email protected] config]$/home/mongo/mongodb/bin/mongo 172.16.202.203:11733
MongoDB shell version: 3.0.6
connecting to: 172.16.202.203:11733/test
> config={_id:'shard3',members:[{_id:0,host:'172.16.202.201:11733'},{_id:1,host:'172.16.202.202:11733',arbiterOnly:true},{_id:2,host:'172.16.202.203:11733',priority:2}]}
{
"_id": "shard3",
"members": [
{
"_id": 0,
"host": "172.16.202.201:11733"
},
{
"_id": 1,
"host": "172.16.202.202:11733",
"arbiterOnly": true
},
{
"_id": 2,
"host": "172.16.202.203:11733",
"priority": 2
}
]
}
> rs.initiate(config)
{ "ok" : 1 }
以上代码通过执行rs.initiate(config)命令来初始化shard2的复制集Replica Set 1。
2.3.4.查看复制集状态
shard1:PRIMARY> rs.status()
{
"set": "shard1",
"date": ISODate("2015-11-25T10:53:06.091Z"),
"myState": 1,
"members": [
{
"_id": 0,
"name": "172.16.202.201:11731",
"health": 1,
"state": 1,
"stateStr": "PRIMARY", #主库
"uptime": 3009,
"optime": Timestamp(1448448493, 1),
"optimeDate": ISODate("2015-11-25T10:48:13Z"),
"electionTime": Timestamp(1448448497, 1),
"electionDate": ISODate("2015-11-25T10:48:17Z"),
"configVersion": 1,
"self": true
},
{
"_id": 1,
"name": "172.16.202.202:11731",
"health": 1,
"state": 2,
"stateStr": "SECONDARY", #复本
"uptime": 292,
"optime": Timestamp(1448448493, 1),
"optimeDate": ISODate("2015-11-25T10:48:13Z"),
"lastHeartbeat": ISODate("2015-11-25T10:53:05.389Z"),
"lastHeartbeatRecv": ISODate("2015-11-25T10:53:05.391Z"),
"pingMs": 0,
"lastHeartbeatMessage": "could not find member to sync from",
"configVersion": 1
},
{
"_id": 2,
"name": "172.16.202.203:11731",
"health": 1,
"state": 7,
"stateStr": "ARBITER", #仲裁
"uptime" : 292,
"lastHeartbeat": ISODate("2015-11-25T10:53:05.391Z"),
"lastHeartbeatRecv": ISODate("2015-11-25T10:53:05.390Z"),
"pingMs": 0,
"configVersion": 1
}
],
"ok": 1
}
2.4.配置3台Config Server
三台上执行操作如下代码所示：
[[email protected] config]$ catconfig.conf
systemLog:
destination: file
##Log
path: /home/mongo/data/logs/config.log
logAppend: true
storage:
journal:
enabled: true
dbPath:/home/mongo/data/config
directoryPerDB: true
processManagement:
fork: true
net:
#bindIp:172.16.202.201 #这里根据自己来是否绑定IP
port: 30000
sharding:
clusterRole: configsvr
[[email protected] config]$ /home/mongo/mongodb/bin/mongos-f /home/mongo/config/config.conf
如以上代码所示，在三台服务器上分别执行启动ConfigServer进程，并指定此进程监听是30000
2.5.配置3台Route Process
三台上执行操作如下代码所示：
[[email protected] config]$ catmongos.conf
systemLog:
destination: file
##Log
path: /home/mongo/data/logs/mongo.log
logAppend: true
processManagement:
fork: true
net:
#bindIp:172.16.202.201
port: 6000
sharding:
configDB: 172.16.202.201:30000,172.16.202.202:30000,172.16.202.203:30000
[[email protected]]$ /home/mongo/mongodb/bin/mongos -f /home/mongo/config/mongos.conf
如以上代码所示，在三台服务器上分别启动路由控制器，并指定此进程监听端口是60000，同时指定三台服务器上的Config Server的IP和端口。
2.6.配置Shard Cluster
连接到其中一台机器的端口60000的mongos进程，并切换到admin数据到开始配置Sharding环境，如下面的代码所示：
[[email protected] logs]$/home/mongo/mongodb/bin/mongo 172.16.202.201:60000
MongoDB shell version: 3.0.6
connecting to: 172.16.202.201:60000/test
mongos> use admin
switched to db admin
mongos>db.runCommand({addshard:"shard1/172.16.202.201:11731,172.16.202.202:11731,172.16.202.203:11731",name:"shard1"});
{ "shardAdded" :"shard1", "ok" : 1 }
mongos>db.runCommand({addshard:"shard2/172.16.202.201:11732,172.16.202.202:11732,172.16.202.203:11732",name:"shard2"});
{ "shardAdded" :"shard2", "ok" : 1 }
mongos>db.runCommand({addshard:"shard3/172.16.202.201:11733,172.16.202.202:11733,172.16.202.203:11733",name:"shard3"});
{ "shardAdded" :"shard3", "ok" : 1 }
以上代码通过执行以下命令：
db.runCommand({addshard:"shard1/172.16.202.201:11731,172.16.202.202:11731,172.16.202.203:11731",name:"shard1"});
将Replica Set 1在三台服务器上的3个成员节点作为Shard Server 1添加进sharding环境中。
以上代码通过执行以下命令：
db.runCommand({addshard:"shard2/172.16.202.201:11732,172.16.202.202:11732,172.16.202.203:11732",name:"shard2"});
将Replica Set 2在三台服务器上的3个成员节点作为Shard Server 2添加进sharding环境中。
以上代码通过执行以下命令：
db.runCommand({addshard:"shard3/172.16.202.201:11733,172.16.202.202:11733,172.16.202.203:11733",name:"shard3"});
将Replica Set 3在三台服务器上的3个成员节点作为Shard Server 3添加进sharding环境中。
接下来**分片，如下面的代码所示： 采用hash分片
[[email protected] logs]$/home/mongo/mongodb/bin/mongo 172.16.202.201:60000
MongoDB shell version: 3.0.6
connecting to: 172.16.202.201:60000/test
mongos> use admin
switched to db admin
mongos> db.runCommand({enablesharding:"logs"})
{"ok" : 1 }
mongos>db.runCommand({shardcollection:"logs.users",key:{id:"hashed"}})
{ "collectionsharded" :"logs.users", "ok" : 1 }
如以上代码所示，首先执行db.runCommand({enablesharding:"logs"})命令**logs库上的分片功能;然后执行
db.runCommand({shardcollection:"logs.methodlog",key:{_id:1}})
命令**users表的分片功能。
本文转自 jxzhfei 51CTO博客，原文链接:http://blog.51cto.com/jxzhfei/1722243