[TOC]
# MongDB分片副本集集群
副本集解决主节点发生故障导致数据丢失或不可用的问题,但遇到需要存储海量数据的情况时,副本集机制就束手无策了。副本集中的一台机器可能不足以存储数据,或者说集群不足以提供可接受的读写吞吐量。这就需要用到 MongoDB 的分片(Sharding)技术,这也是 MongoDB 的另外一种集群部署模式。
分片是指将数据拆分并分散存放在不同机器上的过程。有时也用分区来表示这个概念。将数据分散到不同的机器上,不需要功能强大的大型计算机就可以存储更多的数据,处理更大的负载。
MongoDB 支持自动分片,可以使数据库架构对应用程序不可见,简化系统管理。对应用程序而言,就如同始终在使用一个单机的 MongoDB 服务器一样。
MongoDB 的分片机制允许创建一个包含许多台机器的集群,将数据子集分散在集群中,每个分片维护着一个数据集合的子集。与副本集相比,使用集群架构可以使应用程序具有更强大的数据处理能力。
下图展示了在MongoDB中使用分片集群结构分布:
![image](http://yanxuan.nosdn.127.net/c210a44615a3928580f85e7660c1e1fd.png)
上图中主要有如下所述三个主要组件:
- Shard:
用于存储实际的数据块,实际生产环境中一个shard server角色可由几台机器组个一个relica set承担,防止主机单点故障
- Config Server:
mongod实例,存储了整个 ClusterMetadata,其中包括 chunk信息。4.0版本以后要求最低3个副本集
- Query Routers:
前端路由,客户端由此接入,且让整个集群看上去像单一数据库,前端应用可以透明使用。
# 开始搭建
准备三台机器,这里用的三台虚拟机
系统: CentOS-7.2
软件: MongDB-4.2.7
role/port | 192.168.0.41 | 192.168.0.42 | 192.168.0.43
---|---|---|---
config | 30000 | 30000 | 30000
mongos | 30010 | 30010 | 30010
shard1 | 27001 | 27001 | 27001
shard2 | 27002 | 27002 | 27002
shard3 | 27003 | 27003 | 27003
官网新版本配置文件说明:
[https://docs.mongodb.com/manual/reference/configuration-options/](https://note.youdao.com/)
## config集群
1. 创建文件及文件夹
```
[root@localhost /]# mkdir -p /usr/local/mongodb/data/config
[root@localhost /]# mkdir -p /usr/local/mongodb/log/config
[root@localhost /]# touch /usr/local/mongodb/log/config.log
```
2. 创建config.cfg
```
[root@localhost /]# cat /usr/local/mongodb/conf/config.cfg
systemLog:
destination: file
logAppend: true
path: /usr/local/mongodb/log/config.log
storage:
dbPath: /usr/local/mongodb/data/config
journal:
enabled: true
# engine:
# wiredTiger:
processManagement:
fork: true # fork and run in background
#pidFilePath: /var/run/mongodb/config1.pid # location of pidfile
timeZoneInfo: /usr/share/zoneinfo
# network interfaces
net:
port: 30000
bindIp: 0.0.0.0
#security:
# authorization: enabled
# keyFile: /usr/local/mongodb/conf/mongodb.keyfile
#operationProfiling:
replication:
replSetName: configs
sharding:
clusterRole: configsvr
## Enterprise-Only Options
#auditLog:
#snmp:
```
3. 启动config
```
[root@localhost mongodb]# mongod -f /usr/local/mongodb/conf/config.cfg
about to fork child process, waiting until server is ready for connections.
forked process: 223
child process started successfully, parent exiting
```
ps: 30000端口成功监听
另外2台的操作同上
4. 创建config集群
任意一台机器上操作
```
[root@localhost mongodb]# mongo --port 30000
> rs.initiate(
{
_id: "configs",
configsvr: true,
members: [
{ _id : 0, host : "192.168.0.41:30000" },
{ _id : 1, host : "192.168.0.42:30000" },
{ _id : 2, host : "192.168.0.43:30000" },
]
}
)
"ok" : 1,
"$gleStats" : {
"lastOpTime" : Timestamp(1591233702, 1),
"electionId" : ObjectId("000000000000000000000000")
},
"lastCommittedOpTime" : Timestamp(0, 0),
"$clusterTime" : {
"clusterTime" : Timestamp(1591233702, 1),
"signature" : {
"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
"keyId" : NumberLong(0)
}
},
"operationTime" : Timestamp(1591233702, 1)
}
```
ps: 初始化id必须要和config.cfg内定义的replSetName相同才能成功
rs.status() 查看集群状态
## shard分片集群
1. 创建文件及文件夹
```
[root@localhost /]# mkdir -p /usr/local/mongodb/data/shard{1..3}
[root@localhost /]# mkdir -p /usr/local/mongodb/log/shard{1..3}
[root@localhost /]# touch /usr/local/mongodb/log/shard{1..3}/shard.log
```
2. 创建shard1.cfg,shard2.cfg,shard3.cfg
```
[root@localhost /]# cat /usr/local/mongodb/conf/shard1.cfg
systemLog:
destination: file
logAppend: true
path: /usr/local/mongodb/log/shard1/shard.log
storage:
dbPath: /usr/local/mongodb/data/shard1
journal:
enabled: true
# engine:
# wiredTiger:
processManagement:
fork: true # fork and run in background
#pidFilePath: /var/run/mongodb/shard1.pid # location of pidfile
timeZoneInfo: /usr/share/zoneinfo
net:
port: 27001
bindIp: 0.0.0.0
#security:
# authorization: enabled
# keyFile: /usr/local/mongodb/conf/mongodb.keyfile
#operationProfiling:
replication:
replSetName: shard1
sharding:
clusterRole: shardsvr
## Enterprise-Only Options
#auditLog:
#snmp:
[root@localhost /]# cat /usr/local/mongodb/conf/shard2.cfg
systemLog:
destination: file
logAppend: true
path: /usr/local/mongodb/log/shard2/shard.log
storage:
dbPath: /usr/local/mongodb/data/shard2
journal:
enabled: true
# engine:
# wiredTiger:
processManagement:
fork: true # fork and run in background
#pidFilePath: /var/run/mongodb/shard1.pid # location of pidfile
timeZoneInfo: /usr/share/zoneinfo
net:
port: 27002
bindIp: 0.0.0.0
#security:
#operationProfiling:
replication:
replSetName: shard2
sharding:
clusterRole: shardsvr
## Enterprise-Only Options
#auditLog:
#snmp:
```
shard3.cfg参考shard2.cfg
3. 启动shard
```
[root@localhost mongodb]# mongod -f /usr/local/mongodb/conf/shard1.cfg
[root@localhost mongodb]# mongod -f /usr/local/mongodb/conf/shard2.cfg
[root@localhost mongodb]# mongod -f /usr/local/mongodb/conf/shard3.cfg
```
另外2台的操作同上
4. 创建shard集群
任意一台机器上执行
```
[root@localhost mongodb]# mongo --port 27001
> rs.initiate(
{
_id: "shard1",
members: [
{ _id : 0, host : "192.168.0.41:27001" },
{ _id : 1, host : "192.168.0.42:27001" },
{ _id : 2, host : "192.168.0.43:27001" },
]
}
)
[root@localhost mongodb]# mongo --port 27002
> rs.initiate(
{
_id: "shard2",
members: [
{ _id : 0, host : "192.168.0.41:27002" },
{ _id : 1, host : "192.168.0.42:27002" },
{ _id : 2, host : "192.168.0.43:27002" },
]
}
)
[root@localhost mongodb]# mongo --port 27003
> rs.initiate(
{
_id: "shard3",
members: [
{ _id : 0, host : "192.168.0.41:27003" },
{ _id : 1, host : "192.168.0.42:27003" },
{ _id : 2, host : "192.168.0.43:27003" },
]
}
)
```
目前已经创建了shard1,shard2,shard3三个具有三副本的复制集,后续还会创建分片集群
## mongos服务
1. 准备文件
monogs作为集群访问的入口,是不存储数据的,只需要创建日志文件。每个服务器上启动一个mongos,提高可用性。
```
[root@localhost mongodb]# touch /usr/local/mongodb/log/mongos.log
```
2. 创建mongos.cfg
```
[root@localhost /]# cat /usr/local/mongodb/conf/mongos.cfg
systemLog:
destination: file
logAppend: true
path: /usr/local/mongodb/log/mongos.log
processManagement:
fork: true # fork and run in background
#pidFilePath: /var/run/mongodb/mongos1.pid # location of pidfile
timeZoneInfo: /usr/share/zoneinfo
net:
port: 30010
bindIp: 0.0.0.0
#security:
#operationProfiling:
#eplication:
sharding:
# 配置哪些ip 需要添加
#configDB: configs/192.168.0.153:30000,192.168.0.216:30000,192.168.0.218:30000
configDB: configs/192.168.0.41:30000,192.168.0.42:30000,192.168.0.43:30000
# clusterRole: configsvr
## Enterprise-Only Options
#auditLog:
#snmp:
```
ps: configDB指向三个config服务的地址和端口
3. 启动mongos
```
[root@localhost mongodb]# mongos -f /usr/local/mongodb/conf/mongos.cfg
```
ps:mongos启动不是使用mongod
每台机器都要启动一个mongos服务
## 为各副本集添加秘钥认证
1. 生成秘钥文件
任意一台机器操作
```
[root@localhost mongodb]# openssl rand 700 -base64 > /usr/local/mongodb/conf/mongodb.keyfile
[root@localhost mongodb]# chmod 600 /usr/local/mongodb/conf/mongodb.keyfile
```
2. 将秘钥分发到其他服务器
```
[root@localhost mongodb]# scp /usr/local/mongodb/conf/mongodb.keyfile root@192.168.0.42:/usr/local/mongodb/conf
[root@localhost mongodb]# scp /usr/local/mongodb/conf/mongodb.keyfile root@192.168.0.43:/usr/local/mongodb/conf
```
3. 修改配置文件
config.cfg,shard.cfg,mongos.cfg三类文件新增以下内容
```
security:
keyFile: /usr/local/mongodb/conf/mongodb.keyfile
```
## 创建用于连接的用户
1. 命令行创建用户
```
mongos> use admin
switched to db admin
mongos> db.createUser({user:"mongouser",pwd:"mongopwd",roles:["root"]})
mongos> db.auth("mongouser","mongopwd")
mongos> use config
switched to db config
mongos> db.createUser({user:"mongouser",pwd:"mongopwd",roles:["root"]})
mongos> db.auth("mongouser","mongopwd")
```
2. 修改配置文件
config.cfg,shard.cfg两类文件新增以下内容
```
security:
authorization: enabled
```
3. 重启服务mongo服务
4. 客户端连接
5. 用户管理常用命令
- show users // 查看当前库下的用户
- db.dropUser('testadmin') // 删除用户
- db.updateUser('admin', {pwd: '654321'}) // 修改用户密码
- db.auth('admin', '654321') // 密码认证
## 可能遇到的问题
1. 启用认证后,遇到"command replSetGetStatus requires authentication"
```
configs:SECONDARY> rs.status()
{
"operationTime" : Timestamp(1591253807, 2),
"ok" : 0,
"errmsg" : "command replSetGetStatus requires authentication",
"code" : 13,
"codeName" : "Unauthorized",
"lastCommittedOpTime" : Timestamp(1591253807, 2),
"$clusterTime" : {
"clusterTime" : Timestamp(1591253807, 2),
"signature" : {
"hash" : BinData(0,"QCmeAVDoq7jZzwHfPYNn/mU4eS8="),
"keyId" : NumberLong("6834296761922617360")
}
}
}
configs:SECONDARY> use admin
switched to db admin
configs:SECONDARY> db.auth('mongouser','mongopwd')
1
```
2. 启动mongos实例一直无响应
- 可能是启动顺序,需要启动config实例后操作
- 认证秘钥问题
# 使用实例
1. 连接mongos
任意一台上连接
```
[root@localhost mongodb]# mongo --port 30010
```
2. 添加分片
```
mongos> use admin
switched to db admin
mongos> sh.addShard("shard1/192.168.0.41:27001,192.168.0.42:27001,192.168.0.43:27001")
{
"shardAdded" : "shard1",
"ok" : 1,
"operationTime" : Timestamp(1591241739, 4),
"$clusterTime" : {
"clusterTime" : Timestamp(1591241739, 4),
"signature" : {
"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
"keyId" : NumberLong(0)
}
}
}
mongos> sh.addShard("shard2/192.168.0.41:27002,192.168.0.42:27002,192.168.0.43:27002")
{
"shardAdded" : "shard2",
"ok" : 1,
"operationTime" : Timestamp(1591241810, 2),
"$clusterTime" : {
"clusterTime" : Timestamp(1591241810, 2),
"signature" : {
"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
"keyId" : NumberLong(0)
}
}
}
mongos> sh.addShard("shard3/192.168.0.41:27003,192.168.0.42:27003,192.168.0.43:27003")
{
"shardAdded" : "shard3",
"ok" : 1,
"operationTime" : Timestamp(1591241833, 6),
"$clusterTime" : {
"clusterTime" : Timestamp(1591241833, 6),
"signature" : {
"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
"keyId" : NumberLong(0)
}
}
}
```
- 查看分片状态
sh.status()
- 移除分片
db.runCommand({removeShard : "shard1"})
3. 开启数据库分片(创建数据库)
```
mongos> use admin
mongos> sh.enableSharding("notice_set")
```
4. 指定数据库表中的字段key为分片键,并对其使用哈希分片策略
```
mongos> sh.shardCollection("notice_set.notice",{"msgid": "hashed"})
```
5. 添加测试数据
```
mongos> use notice_set
mongos> for (i = 1; i <= 10; i++) db.notice.insert({"msgid":i})
```
可以查看每个分片的数据量等于10
# 连接集群的方式
1. 单机连接
选任一mongos直接连接,如果该mongos不可用,服务也无法继续使用
```
mongos --host --port
```
2. 集群连接
配置上所有mongos,某一个mongos不可用,服务将继续
```
mongodb://[username:password@]host1[:port1][,host2[:port2],...[,hostN[:portN]]][/[database][?options]]
```
- mongodb:// 前缀,代表这是一个Connection String
- username:password@ 如果启用了鉴权,需要指定用户密码
- hostX:portX 多个 mongos 的地址列表
- /database 鉴权时,用户帐号所属的数据库
- ?options 指定额外的连接选项