一、Mongodb的分片(MSC)
1.分片的概念
mongodb的副本集跟redis的高可用相同,只能读,分担不了主库的压力,只能在主库出现故障的时候接替主库的工作
mongodb能够使用的内存,只是主库的内存和磁盘,当副本集中机器配置不一致时也会有问题
2.分片的介绍
#优点:
1.提高机器资源的利用率
2.减轻主库的压力
#缺点:
1.机器需要的更多
2.配置和管理更加的复杂和困难
3.分片配置好之后想修改很困难
3.分片的原理
1)路由服务 mongos server
类似于代理,跟数据库的atlas类似,可以将客户端的数据分配到后端的mongo服务器上
2)分片配置服务器信息 config server
mongos server是不知道后端服务器mongo有几台,地址是什么,他只能连接到这个config server,而config server就是记录后端服务器地址和数据的一个服务
作用:
1.记录后端mongo节点的信息
2.记录数据写入存到了哪个节点
3.提供给mongos后端服务器的信息
3)片键
config server只存储信息,而不会主动将数据写入节点,所以还有一个片键的概念,片键就是索引
作用:
1.将数据根据规则分配到不同的节点
2.相当于建立索引,加快访问速度
分类:
1.区间片键(很有可能出现数据分配不均匀的情况)
可以以时间区间分片,根据时间建立索引
可以以地区区间分片,根据地区建立索引
2.hash片键(足够平均,足够随机)
根据id或者数据数量进行分配
4)分片
存储数据的节点,这种方式就是分布式集群
二、分片集群的搭建
做分片只是针对单节点,mongo服务相当于还是只有一个,所以我们还有对分片进行副本集的操作
跟ES一样,我们不能一台机器上部署多节点,自己做自己的副本,那当机器挂了时,还是有问题
所以我们要错开进行副本集的建立,而且一台机器上不能有相同的数据节点,否则选举又会出现问题
1.服务器规划15台
主机 |
ip |
部署 |
端口 |
mongodb01 |
10.0.0.81 |
Shard1_Master Shard2_Slave Shard3_Arbiter Config Server Mongos Server |
20010 28020 28030 40000 60000 |
mongodb02 |
10.0.0.82 |
Shard2_Master Shard3_Slave Shard1_Arbiter Config Server Mongos Server |
20010 28020 28030 40000 60000 |
mongodb03 |
10.0.0.83 |
Shard3_Master Shard1_Slave Shard2_Arbiter Config Server Mongos Server |
20010 28020 28030 40000 60000 |
2.目录规划
#服务目录
mkdir /server/mongodb/master/{conf,log,pid,data} -p
mkdir /server/mongodb/slave/{conf,log,pid,data} -p
mkdir /server/mongodb/arbiter/{conf,log,pid,data} -p
mkdir /server/mongodb/config/{conf,log,pid,data} -p
mkdir /server/mongodb/mongos/{conf,log,pid} -p
3.安装mongo
#安装依赖
yum install -y libcurl openssl
#上传或下载包
rz mongodb-linux-x86_64-3.6.13.tgz
#解压
tar xf mongodb-linux-x86_64-3.6.13.tgz -C /usr/local/
#做软连接
ln -s /usr/local/mongodb-linux-x86_64-3.6.13 /usr/local/mongodb
4.配置mongodb01
1)配置shard上的master
vim /server/mongodb/master/conf/mongo.conf
systemLog:
destination: file
logAppend: true
path: /server/mongodb/master/log/master.log
storage:
journal:
enabled: true
dbPath: /server/mongodb/master/data/
directoryPerDB: true
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes: true
collectionConfig:
blockCompressor: zlib
indexConfig:
prefixCompression: true
processManagement:
fork: true
pidFilePath: /server/mongodb/master/pid/master.pid
timeZoneInfo: /usr/share/zoneinfo
net:
port: 28010
bindIp: 127.0.0.1,10.0.0.81
replication:
oplogSizeMB: 1024
replSetName: shard1
sharding:
clusterRole: shardsvr #分片规则:分片集群服务
2)配置shard上的slave
vim /server/mongodb/slave/conf/mongo.conf
systemLog:
destination: file
logAppend: true
path: /server/mongodb/slave/log/slave.log
storage:
journal:
enabled: true
dbPath: /server/mongodb/slave/data/
directoryPerDB: true
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes: true
collectionConfig:
blockCompressor: zlib
indexConfig:
prefixCompression: true
processManagement:
fork: true
pidFilePath: /server/mongodb/slave/pid/slave.pid
timeZoneInfo: /usr/share/zoneinfo
net:
port: 28020
bindIp: 127.0.0.1,10.0.0.81
replication:
oplogSizeMB: 1024
replSetName: shard2
sharding:
clusterRole: shardsvr
3)配置shard上的arbiter
vim /server/mongodb/arbiter/conf/mongo.conf
systemLog:
destination: file
logAppend: true
path: /server/mongodb/arbiter/log/arbiter.log
storage:
journal:
enabled: true
dbPath: /server/mongodb/arbiter/data/
directoryPerDB: true
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes: true
collectionConfig:
blockCompressor: zlib
indexConfig:
prefixCompression: true
processManagement:
fork: true
pidFilePath: /server/mongodb/arbiter/pid/arbiter.pid
timeZoneInfo: /usr/share/zoneinfo
net:
port: 28030
bindIp: 127.0.0.1,10.0.0.81
replication:
oplogSizeMB: 1024
replSetName: shard3
sharding:
clusterRole: shardsvr
5.配置mongodb02
1)配置shard上的master
vim /server/mongodb/master/conf/mongo.conf
systemLog:
destination: file
logAppend: true
path: /server/mongodb/master/log/master.log
storage:
journal:
enabled: true
dbPath: /server/mongodb/master/data/
directoryPerDB: true
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes: true
collectionConfig:
blockCompressor: zlib
indexConfig:
prefixCompression: true
processManagement:
fork: true
pidFilePath: /server/mongodb/master/pid/master.pid
timeZoneInfo: /usr/share/zoneinfo
net:
port: 28010
bindIp: 127.0.0.1,10.0.0.82
replication:
oplogSizeMB: 1024
replSetName: shard2
sharding:
clusterRole: shardsvr
2)配置shard上的slave
vim /server/mongodb/slave/conf/mongo.conf
systemLog:
destination: file
logAppend: true
path: /server/mongodb/slave/log/slave.log
storage:
journal:
enabled: true
dbPath: /server/mongodb/slave/data/
directoryPerDB: true
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes: true
collectionConfig:
blockCompressor: zlib
indexConfig:
prefixCompression: true
processManagement:
fork: true
pidFilePath: /server/mongodb/slave/pid/slave.pid
timeZoneInfo: /usr/share/zoneinfo
net:
port: 28020
bindIp: 127.0.0.1,10.0.0.82
replication:
oplogSizeMB: 1024
replSetName: shard3
sharding:
clusterRole: shardsvr
3)配置shard上的arbiter
vim /server/mongodb/arbiter/conf/mongo.conf
systemLog:
destination: file
logAppend: true
path: /server/mongodb/arbiter/log/arbiter.log
storage:
journal:
enabled: true
dbPath: /server/mongodb/arbiter/data/
directoryPerDB: true
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes: true
collectionConfig:
blockCompressor: zlib
indexConfig:
prefixCompression: true
processManagement:
fork: true
pidFilePath: /server/mongodb/arbiter/pid/arbiter.pid
timeZoneInfo: /usr/share/zoneinfo
net:
port: 28030
bindIp: 127.0.0.1,10.0.0.82
replication:
oplogSizeMB: 1024
replSetName: shard1
sharding:
clusterRole: shardsvr
6.配置mongodb03
1)配置shard上的master
vim /server/mongodb/master/conf/mongo.conf
systemLog:
destination: file
logAppend: true
path: /server/mongodb/master/log/master.log
storage:
journal:
enabled: true
dbPath: /server/mongodb/master/data/
directoryPerDB: true
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes: true
collectionConfig:
blockCompressor: zlib
indexConfig:
prefixCompression: true
processManagement:
fork: true
pidFilePath: /server/mongodb/master/pid/master.pid
timeZoneInfo: /usr/share/zoneinfo
net:
port: 28010
bindIp: 127.0.0.1,10.0.0.83
replication:
oplogSizeMB: 1024
replSetName: shard3
sharding:
clusterRole: shardsvr
2)配置shard上的slave
vim /server/mongodb/slave/conf/mongo.conf
systemLog:
destination: file
logAppend: true
path: /server/mongodb/slave/log/slave.log
storage:
journal:
enabled: true
dbPath: /server/mongodb/slave/data/
directoryPerDB: true
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes: true
collectionConfig:
blockCompressor: zlib
indexConfig:
prefixCompression: true
processManagement:
fork: true
pidFilePath: /server/mongodb/slave/pid/slave.pid
timeZoneInfo: /usr/share/zoneinfo
net:
port: 28020
bindIp: 127.0.0.1,10.0.0.83
replication:
oplogSizeMB: 1024
replSetName: shard1
sharding:
clusterRole: shardsvr
3)配置shard上的arbiter
vim /server/mongodb/arbiter/conf/mongo.conf
systemLog:
destination: file
logAppend: true
path: /server/mongodb/arbiter/log/arbiter.log
storage:
journal:
enabled: true
dbPath: /server/mongodb/arbiter/data/
directoryPerDB: true
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes: true
collectionConfig:
blockCompressor: zlib
indexConfig:
prefixCompression: true
processManagement:
fork: true
pidFilePath: /server/mongodb/arbiter/pid/arbiter.pid
timeZoneInfo: /usr/share/zoneinfo
net:
port: 28030
bindIp: 127.0.0.1,10.0.0.83
replication:
oplogSizeMB: 1024
replSetName: shard2
sharding:
clusterRole: shardsvr
7.配置环境变量
[root@redis01 ~]# vim /etc/profile.d/mongo.sh
export PATH="/usr/local/mongodb/bin:$PATH"
[root@redis01 ~]# source /etc/profile
8.优化警告
useradd mongo -s /sbin/nologin -M
echo "never" > /sys/kernel/mm/transparent_hugepage/enabled
echo "never" > /sys/kernel/mm/transparent_hugepage/defrag
9.配置system管理
1)配置master管理
vim /usr/lib/systemd/system/mongod-master.service
[Unit]
Description=MongoDB Database Server
Documentation=https://docs.mongodb.org/manual
After=network.target
[Service]
User=mongo
Group=mongo
ExecStart=/usr/local/mongodb/bin/mongod -f /server/mongodb/master/conf/mongo.conf
ExecStartPre=/usr/bin/chown -R mongo:mongo /server/mongodb/master/
ExecStop=/usr/local/mongodb/bin/mongod -f /server/mongodb/master/conf/mongo.conf --shutdown
PermissionsStartOnly=true
PIDFile=/server/mongodb/master/pid/master.pid
Type=forking
[Install]
WantedBy=multi-user.target
2)配置管理salve
vim /usr/lib/systemd/system/mongod-slave.service
[Unit]
Description=MongoDB Database Server
Documentation=https://docs.mongodb.org/manual
After=network.target
[Service]
User=mongo
Group=mongo
ExecStart=/usr/local/mongodb/bin/mongod -f /server/mongodb/slave/conf/mongo.conf
ExecStartPre=/usr/bin/chown -R mongo:mongo /server/mongodb/slave/
ExecStop=/usr/local/mongodb/bin/mongod -f /server/mongodb/slave/conf/mongo.conf --shutdown
PermissionsStartOnly=true
PIDFile=/server/mongodb/slave/pid/slave.pid
Type=forking
[Install]
WantedBy=multi-user.target
3)配置管理arbiter
vim /usr/lib/systemd/system/mongod-arbiter.service
[Unit]
Description=MongoDB Database Server
Documentation=https://docs.mongodb.org/manual
After=network.target
[Service]
User=mongo
Group=mongo
ExecStart=/usr/local/mongodb/bin/mongod -f /server/mongodb/arbiter/conf/mongo.conf
ExecStartPre=/usr/bin/chown -R mongo:mongo /server/mongodb/arbiter/
ExecStop=/usr/local/mongodb/bin/mongod -f /server/mongodb/arbiter/conf/mongo.conf --shutdown
PermissionsStartOnly=true
PIDFile=/server/mongodb/arbiter/pid/arbiter.pid
Type=forking
[Install]
WantedBy=multi-user.target
4)刷新启动程序
systemctl daemon-reload
10.启动mongodb所有节点
systemctl start mongod-master.service
systemctl start mongod-slave.service
systemctl start mongod-arbiter.service
11.配置副本集(交错方式)
1)mongodb01初始化副本集shard
#连接主库
mongo --port 28010
rs.add("10.0.0.83:28020")
rs.addArb("10.0.0.82:28030")
或者使用:
config = {_id: 'sh1', members: [
{_id: 0, host: '10.0.0.81:28010'},
{_id: 1, host: '10.0.0.83:28020'},
{_id: 2, host: '10.0.0.82:28030',"arbiterOnly":true}]
}
2)mongodb02初始化副本集shard
#连接主库
mongo --port 28010
rs.add("10.0.0.81:28020")
rs.addArb("10.0.0.83:28030")
3)mongodb03初始化副本集shard
#连接主库
mongo --port 28010
rs.add("10.0.0.82:28020")
rs.addArb("10.0.0.81:28030")
4)检查所有节点副本集状态
#三台主节点
mongo --port 28010
rs.status()
rs.isMaster()
12.配置config server
1)创建目录
2)配置config server
vim /server/mongodb/config/conf/mongo.conf
systemLog:
destination: file
logAppend: true
path: /server/mongodb/config/log/mongodb.log
storage:
journal:
enabled: true
dbPath: /server/mongodb/config/data/
directoryPerDB: true
wiredTiger:
engineConfig:
cacheSizeGB: 1
directoryForIndexes: true
collectionConfig:
blockCompressor: zlib
indexConfig:
prefixCompression: true
processManagement:
fork: true
pidFilePath: /server/mongodb/config/pid/mongod.pid
timeZoneInfo: /usr/share/zoneinfo
net:
port: 40000
bindIp: 127.0.0.1,10.0.0.81
replication:
replSetName: configset
sharding:
clusterRole: configsvr
3)启动
/usr/local/mongodb/bin/mongod -f /server/mongodb/config/conf/mongo.conf
4)mongodb01上初始化副本集config server
mongo --port 40000
rs.initiate({
_id:"configset",
1configsvr: true,
members:[
{_id:0,host:"10.0.0.51:40000"},
{_id:1,host:"10.0.0.52:40000"},
{_id:2,host:"10.0.0.53:40000"},
]})
5)检查
rs.status()
rs.isMaster()
13.配置mongos
1)创建目录
2)配置mongos
vim /server/mongodb/mongos/conf/mongo.conf
=======================
systemLog:
destination: file
logAppend: true
path: /server/mongodb/mongos/log/mongos.log
processManagement:
fork: true
pidFilePath: /server/mongodb/mongos/pid/mongos.pid
timeZoneInfo: /usr/share/zoneinfo
net:
port: 60000
bindIp: 127.0.0.1,10.0.0.81
sharding:
configDB:
configReplset/10.0.0.81:40000,10.0.0.82:40000,10.0.0.83:40000
=======================
#有的低版本中configReplset替换使用configSet
3)启动
/usr/local/mongodb/bin/mongod -f /server/mongodb/mongos/conf/mongo.conf
4)添加分片成员
#登录mongos
mongo --port 60000
#添加成员
use admin
db.runCommand({addShard:'shard1/10.0.0.81:28100,10.0.0.83:28200,10.0.0.82:28300'})
db.runCommand({addShard:'shard2/10.0.0.82:28100,10.0.0.81:28200,10.0.0.83:28300'})
db.runCommand({addShard:'shard3/10.0.0.83:28100,10.0.0.82:28200,10.0.0.81:28300'})
5)查看分片信息
db.runCommand( { listshards : 1 } )
或者
sh.status()
"在db02和db03上也重复同样步骤即可."
- 以上即完成了集群的搭建,已经可以使用,但分片是使用自动分片的功能,会多做一步chunk迁移,影响io性能,所以使用以下分片的进一步优化步骤.
三、手动配置区间分片策略
- range方式分片
- hash方式分片(推荐,更快更均衡)
1. range方式分片
1)开启区间分片功能
需要在mongos中开启
#数据库开启分片
mongo --port 60000
use admin
#指定库开启分片
db.runCommand( { enablesharding : "test" } )
2)创建集合索引
mongo --port 60000
use test
db.range.ensureIndex( { id: 1 } )
#在range表上创建id列索引顺序排序.
3)对集合开启分片,片键是id
use admin
db.runCommand( { shardcollection : "test.range",key : {id: 1} } )
4)插入测试数据
use test
for(i=1;i<10000;i++){ db.range.insert({"id":i,"name":"shanghai","age":28,"date":new Date()}); }
db.range.stats()
db.range.count()
2. hash方式分片
#数据库开启分片
mongo --port 60000
use admin
db.runCommand( { enablesharding : "testhash" } )
1)集合创建索引
use testhash
db.hash.ensureIndex( { id: "hashed" } )
#在testhash库的hash表下建立索引列id,类型为hashed(哈希)
2)集合开启哈希分片
use admin
sh.shardCollection( "testhash.hash", { id: "hashed" } )
3)生成测试数据
use testhash
for(i=1;i<10000;i++){ db.hash.insert({"id":i,"name":"shanghai","age":70}); }
4)验证数据
分片验证
#mongodb01
mongo --port 28010
use testhash
db.hash.count()
33755
#mongodb01
mongo --port 28010
use testhash
db.hash.count()
33142
#mongodb01
mongo --port 28010
use testhash
db.hash.count()
33102
四、分片集群常用管理命令
## 可以查看以下所有信息
sh.status()
# 1.列出分片所有详细信息
db.printShardingStatus()
sh.status()
# 2.列出所有分片成员信息
use admin
db.runCommand({ listshards : 1})
# 3.列出开启分片的数据库
use config
db.databases.find({"partitioned": true })
# 4.查看分片的片键
use config
db.collections.find().pretty()
# 5. 其他命令
删除分片节点(谨慎)
(1)确认blance是否在工作
sh.getBalancerState()
(2)删除shard2节点(谨慎)
mongos> db.runCommand( { removeShard: "shard2" } )
#注意:删除操作一定会立即触发blancer然后迁移数据。最好在停机状态下操作,速度很慢,会影响很大。
五、balancer操作管理
1. balancer介绍
# mongos的一个重要功能,自动巡查所有shard节点上的chunk的情况,自动做chunk迁移。
什么时候工作?
1、自动运行,会检测系统不繁忙的时候做迁移
2、在做节点删除的时候,立即开始迁移工作
3、balancer只能在预设定的时间窗口内运行
# 有需要时可以关闭和开启blancer(备份的时候)
mongos> sh.stopBalancer()
mongos> sh.startBalancer()
2. 自定义balancer进行的时间段
#官方说明文档:
https://docs.mongodb.com/manual/tutorial/manage-sharded-cluster-balancer/#schedule-the-balancing-window
========================
#!生产中必须要调整的配置.(重要)
use config
sh.setBalancerState( true )
db.settings.update({ _id : "balancer" }, { $set : { activeWindow : { start : "3:00", stop : "5:00" } } }, true )
#设定balancer开始和结束的时间段
sh.getBalancerWindow()
sh.status()
========================
#关于集合的balancer(了解):
1.关闭某个集合的balance
sh.disableBalancing("students.grades")
2.打开某个集合的balancer
sh.enableBalancing("students.grades")
3.确定某个集合的balance是开启或者关闭
db.getSiblingDB("config").collections.findOne({_id : "students.grades"}).noBalance;
六、mongo配置密码做副本集
openssl rand -base64 123 > /server/mongodb/mongo.key
chown -R mongod.mongod /server/mongodb/mongo.key
chmod -R 600 /server/mongodb/mongo.key
scp -r /server/mongodb/mongo.key 192.168.1.82:/server/mongodb/
scp -r /server/mongodb/mongo.key 192.168.1.83:/server/mongodb/