副本集（Replica Set）是一组MongoDB实例组成的集群，由一个主（Primary）服务器和多个备份（Secondary）服务器构成。通过Replication，将数据的更新由Primary推送到其他实例上，在一定的延迟之后，每个MongoDB实例维护相同的数据集副本。通过维护冗余的数据库副本，能够实现数据的异地备份，读写分离和自动故障转移。
也就是说如果主服务器崩溃了，备份服务器会自动将其中一个成员升级为新的主服务器。使用复制功能时，如果有一台服务器宕机了，仍然可以从副本集的其他服务器*问数据。如果服务器上的数据损坏或者不可访问，可以从副本集的某个成员中创建一份新的数据副本。
早期的MongoDB版本使用master-slave，一主一从和MySQL类似，但slave在此架构中为只读，当主库宕机后，从库不能自动切换为主。目前已经淘汰master-slave模式，改为副本集，这种模式下有一个主(primary)，和多个从(secondary)，只读。支持给它们设置权重，当主宕掉后，权重最高的从切换为主。在此架构中还可以建立一个仲裁(arbiter)的角色，它只负责裁决，而不存储数据。此架构中读写数据都是在主上，要想实现负载均衡的目的需要手动指定读库的目标server。
简而言之MongoDB 副本集是有自动故障恢复功能的主从集群，有一个Primary节点和一个或多个Secondary节点组成。类似于MySQL的MMM架构。更多关于副本集的介绍请见官方文档：
官方文档地址：
副本集架构图：
21.34 mongodb副本集搭建
我这里使用了三台机器搭建副本集：
192.168.77.128 (primary)
192.168.77.130 (secondary)
192.168.77.134 (secondary)
这三台机器上都已经安装好了MongoDB。
开始搭建：
1.编辑三台机器的配置文件，更改或增加以下内容：
[root@localhost ~]# vim /etc/mongod.conf
replication: # 取消这行的注释
oplogSizeMB: 20 # 增加这一行配置定义oplog的大小，注意前面需要有两个空格
replSetName: zero # 定义复制集的名称，同样的前面需要有两个空格注：需要确保每台机器的配置文件中的bindIp都有配置监听自身的内网IP
2.编辑完成后，分别重启三台机器的MongoDB服务：
[root@localhost ~]# systemctl restart mongod.service
[root@localhost ~]# ps aux |grep mongod
mongod 2578 0.7 8.9 1034696 43592 ? Sl 18:21 0:00 /usr/bin/mongod -f /etc/mongod.conf
root 2605 0.0 0.1 112660 964 pts/0 S+ 18:21 0:00 grep --color=auto mongod
[root@localhost ~]# netstat -lntp |grep mongod
tcp 0 0 192.168.77.134:27017 0.0.0.0:* LISTEN 2578/mongod
tcp 0 0 127.0.0.1:27017 0.0.0.0:* LISTEN 2578/mongod
[root@localhost ~]#3.关闭三台机器的防火墙，或者清空iptables规则
4.连接主机器的MongoDB，在主机器上运行命令mongo，然后配置副本集：
[root@localhost ~]# mongo
> use admin
switched to db admin
> config={_id:"zero",members:[{_id:0,host:"192.168.77.128:27017"},{_id:1,host:"192.168.77.130:27017"},{_id:2,host:"192.168.77.134:27017"}]} # 分别配置三台机器的ip
{
"_id" : "zero", # 副本集的名称
"members" : [
{
"_id" : 0,
"host" : "192.168.77.128:27017"
},
{
"_id" : 1,
"host" : "192.168.77.130:27017"
},
{
"_id" : 2,
"host" : "192.168.77.134:27017"
}
]
}
> rs.initiate(config) # 初始化
{
"ok" : 1,
"operationTime" : Timestamp(1515465317, 1),
"$clusterTime" : {
"clusterTime" : Timestamp(1515465317, 1),
"signature" : {
"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
"keyId" : NumberLong(0)
}
}
}
zero:PRIMARY> rs.status() # 查看状态
{
"set" : "zero",
"date" : ISODate("2018-01-09T02:37:13.713Z"),
"myState" : 1,
"term" : NumberLong(1),
"heartbeatIntervalMillis" : NumberLong(2000),
"optimes" : {
"lastCommittedOpTime" : {
"ts" : Timestamp(1515465429, 1),
"t" : NumberLong(1)
},
"readConcernMajorityOpTime" : {
"ts" : Timestamp(1515465429, 1),
"t" : NumberLong(1)
},
"appliedOpTime" : {
"ts" : Timestamp(1515465429, 1),
"t" : NumberLong(1)
},
"durableOpTime" : {
"ts" : Timestamp(1515465429, 1),
"t" : NumberLong(1)
}
},
"members" : [
{
"_id" : 0,
"name" : "192.168.77.128:27017",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"uptime" : 527,
"optime" : {
"ts" : Timestamp(1515465429, 1),
"t" : NumberLong(1)
},
"optimeDate" : ISODate("2018-01-09T02:37:09Z"),
"infoMessage" : "could not find member to sync from",
"electionTime" : Timestamp(1515465327, 1),
"electionDate" : ISODate("2018-01-09T02:35:27Z"),
"configVersion" : 1,
"self" : true
},
{
"_id" : 1,
"name" : "192.168.77.130:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 116,
"optime" : {
"ts" : Timestamp(1515465429, 1),
"t" : NumberLong(1)
},
"optimeDurable" : {
"ts" : Timestamp(1515465429, 1),
"t" : NumberLong(1)
},
"optimeDate" : ISODate("2018-01-09T02:37:09Z"),
"optimeDurableDate" : ISODate("2018-01-09T02:37:09Z"),
"lastHeartbeat" : ISODate("2018-01-09T02:37:13.695Z"),
"lastHeartbeatRecv" : ISODate("2018-01-09T02:37:13.661Z"),
"pingMs" : NumberLong(0),
"syncingTo" : "192.168.77.128:27017",
"configVersion" : 1
},
{
"_id" : 2,
"name" : "192.168.77.134:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 116,
"optime" : {
"ts" : Timestamp(1515465429, 1),
"t" : NumberLong(1)
},
"optimeDurable" : {
"ts" : Timestamp(1515465429, 1),
"t" : NumberLong(1)
},
"optimeDate" : ISODate("2018-01-09T02:37:09Z"),
"optimeDurableDate" : ISODate("2018-01-09T02:37:09Z"),
"lastHeartbeat" : ISODate("2018-01-09T02:37:13.561Z"),
"lastHeartbeatRecv" : ISODate("2018-01-09T02:37:13.660Z"),
"pingMs" : NumberLong(0),
"syncingTo" : "192.168.77.128:27017",
"configVersion" : 1
}
],
"ok" : 1,
"operationTime" : Timestamp(1515465429, 1),
"$clusterTime" : {
"clusterTime" : Timestamp(1515465429, 1),
"signature" : {
"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
"keyId" : NumberLong(0)
}
}
}
zero:PRIMARY>以上我们需要关注三台机器的stateStr状态，主机器的stateStr状态需要为PRIMARY，两台从机器的stateStr状态需要为SECONDARY才是正常。
如果出现两个从上的stateStr状态为"stateStr" : "STARTUP", 则需要进行如下操作：
> config={_id:"zero",members:[{_id:0,host:"192.168.77.128:27017"},{_id:1,host:"192.168.77.130:27017"},{_id:2,host:"192.168.77.134:27017"}]}
> rs.reconfig(config)然后再次查看状态：rs.status()，确保从的状态变为SECONDARY。
21.35 mongodb副本集测试
1.在主机器上创建库以及创建集合：
zero:PRIMARY> use testdb # 创建库
switched to db testdb
zero:PRIMARY> db.test.insert({AccountID:1,UserName:"zero",password:"123456"}) # 创建集合，并且插入一条数据
WriteResult({ "nInserted" : 1 })
zero:PRIMARY> show dbs # 查看所有的库
admin 0.000GB
config 0.000GB
local 0.000GB
testdb 0.000GB
zero:PRIMARY> show tables # 查看当前库的集合
test
zero:PRIMARY>2.然后到从机器上查看是否有同步主机器上的数据：
[root@localhost ~]# mongo
zero:SECONDARY> show dbs
2018-01-09T18:46:09.959+0800 E QUERY [thread1] Error: listDatabases failed:{
"operationTime" : Timestamp(1515466399, 1),
"ok" : 0,
"errmsg" : "not master and slaveOk=false",
"code" : 13435,
"codeName" : "NotMasterNoSlaveOk",
"$clusterTime" : {
"clusterTime" : Timestamp(1515466399, 1),
"signature" : {
"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
"keyId" : NumberLong(0)
}
}
} :
_getErrorWithCode@src/mongo/shell/utils.js:25:13
Mongo.prototype.getDBs@src/mongo/shell/mongo.js:65:1
shellHelper.show@src/mongo/shell/utils.js:813:19
shellHelper@src/mongo/shell/utils.js:703:15
@(shellhelp2):1:1
zero:SECONDARY> rs.slaveOk() # 如果出现以上错误，需要执行这条命令
zero:SECONDARY> show dbs # 然后就不会报错了
admin 0.000GB
config 0.000GB
local 0.000GB
testdb 0.000GB
zero:SECONDARY> use testdb
switched to db testdb
zero:SECONDARY> show tables
test
zero:SECONDARY>如上可以看到数据已经成功同步到从机器上了。
副本集更改权重模拟主宕机
使用rs.config()命令可以查看每台机器的权重：
zero:PRIMARY> rs.config()
{
"_id" : "zero",
"version" : 1,
"protocolVersion" : NumberLong(1),
"members" : [
{
"_id" : 0,
"host" : "192.168.77.128:27017",
"arbiterOnly" : false,
"buildIndexes" : true,
"hidden" : false,
"priority" : 1,
"tags" : {
},
"slaveDelay" : NumberLong(0),
"votes" : 1
},
{
"_id" : 1,
"host" : "192.168.77.130:27017",
"arbiterOnly" : false,
"buildIndexes" : true,
"hidden" : false,
"priority" : 1,
"tags" : {
},
"slaveDelay" : NumberLong(0),
"votes" : 1
},
{
"_id" : 2,
"host" : "192.168.77.134:27017",
"arbiterOnly" : false,
"buildIndexes" : true,
"hidden" : false,
"priority" : 1,
"tags" : {
},
"slaveDelay" : NumberLong(0),
"votes" : 1
}
],
"settings" : {
"chainingAllowed" : true,
"heartbeatIntervalMillis" : 2000,
"heartbeatTimeoutSecs" : 10,
"electionTimeoutMillis" : 10000,
"catchUpTimeoutMillis" : -1,
"catchUpTakeoverDelayMillis" : 30000,
"getLastErrorModes" : {
},
"getLastErrorDefaults" : {
"w" : 1,
"wtimeout" : 0
},
"replicaSetId" : ObjectId("5a542a65e491a43160eb92f0")
}
}
zero:PRIMARY>priority的值表示该机器的权重，默认都为1。
增加一条防火墙规则来阻断通信模拟主机器宕机：
# 注意这是在主机器上执行
[root@localhost ~]# iptables -I INPUT -p tcp --dport 27017 -j DROP然后到从机器上查看状态：
zero:SECONDARY> rs.status()
{
"set" : "zero",
"date" : ISODate("2018-01-09T14:06:24.127Z"),
"myState" : 1,
"term" : NumberLong(4),
"heartbeatIntervalMillis" : NumberLong(2000),
"optimes" : {
"lastCommittedOpTime" : {
"ts" : Timestamp(1515506782, 1),
"t" : NumberLong(4)
},
"readConcernMajorityOpTime" : {
"ts" : Timestamp(1515506782, 1),
"t" : NumberLong(4)
},
"appliedOpTime" : {
"ts" : Timestamp(1515506782, 1),
"t" : NumberLong(4)
},
"durableOpTime" : {
"ts" : Timestamp(1515506782, 1),
"t" : NumberLong(4)
}
},
"members" : [
{
"_id" : 0,
"name" : "192.168.77.128:27017",
"health" : 0,
"state" : 8,
"stateStr" : "(not reachable/healthy)",
"uptime" : 0,
"optime" : {
"ts" : Timestamp(0, 0),
"t" : NumberLong(-1)
},
"optimeDurable" : {
"ts" : Timestamp(0, 0),
"t" : NumberLong(-1)
},
"optimeDate" : ISODate("1970-01-01T00:00:00Z"),
"optimeDurableDate" : ISODate("1970-01-01T00:00:00Z"),
"lastHeartbeat" : ISODate("2018-01-09T14:06:20.243Z"),
"lastHeartbeatRecv" : ISODate("2018-01-09T14:06:23.491Z"),
"pingMs" : NumberLong(0),
"lastHeartbeatMessage" : "Couldn't get a connection within the time limit",
"configVersion" : -1
},
{
"_id" : 1,
"name" : "192.168.77.130:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 1010,
"optime" : {
"ts" : Timestamp(1515506782, 1),
"t" : NumberLong(4)
},
"optimeDurable" : {
"ts" : Timestamp(1515506782, 1),
"t" : NumberLong(4)
},
"optimeDate" : ISODate("2018-01-09T14:06:22Z"),
"optimeDurableDate" : ISODate("2018-01-09T14:06:22Z"),
"lastHeartbeat" : ISODate("2018-01-09T14:06:23.481Z"),
"lastHeartbeatRecv" : ISODate("2018-01-09T14:06:23.178Z"),
"pingMs" : NumberLong(0),
"syncingTo" : "192.168.77.134:27017",
"configVersion" : 1
},
{
"_id" : 2,
"name" : "192.168.77.134:27017",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"uptime" : 1250,
"optime" : {
"ts" : Timestamp(1515506782, 1),
"t" : NumberLong(4)
},
"optimeDate" : ISODate("2018-01-09T14:06:22Z"),
"electionTime" : Timestamp(1515506731, 1),
"electionDate" : ISODate("2018-01-09T14:05:31Z"),
"configVersion" : 1,
"self" : true
}
],
"ok" : 1,
"operationTime" : Timestamp(1515506782, 1),
"$clusterTime" : {
"clusterTime" : Timestamp(1515506782, 1),
"signature" : {
"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
"keyId" : NumberLong(0)
}
}
}
zero:PRIMARY>如上，可以看到 192.168.77.128 的 stateStr 的值变成 not reachable/healthy 了，而 192.168.77.134 自动切换成主了，也可以看到192.168.77.134 的 stateStr 的值变成 了PRIMARY。因为权重是相同的，所以切换是有一定的随机性的。
接下来我们指定每台机器权重，让权重高的机器自动切换为主。
1.先删除192.168.77.128上的防火墙规则：
[root@localhost ~]# iptables -D INPUT -p tcp --dport 27017 -j DROP
2.回到192.168.77.134机器上，指定各个机器的权重：
zero:PRIMARY> cfg = rs.conf()
zero:PRIMARY> cfg.members[0].priority = 3
3
zero:PRIMARY> cfg.members[1].priority = 2
2
zero:PRIMARY> cfg.members[2].priority = 1
zero:PRIMARY> rs.reconfig(cfg)
{
"ok" : 1,
"operationTime" : Timestamp(1515507322, 1),
"$clusterTime" : {
"clusterTime" : Timestamp(1515507322, 1),
"signature" : {
"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
"keyId" : NumberLong(0)
}
}
}
zero:PRIMARY>3.这时候192.168.77.128应该就切换成主了，到192.168.77.128上执行rs.config()进行查看：
zero:PRIMARY> rs.config()
{
"_id" : "zero",
"version" : 2,
"protocolVersion" : NumberLong(1),
"members" : [
{
"_id" : 0,
"host" : "192.168.77.128:27017",
"arbiterOnly" : false,
"buildIndexes" : true,
"hidden" : false,
"priority" : 3,
"tags" : {
},
"slaveDelay" : NumberLong(0),
"votes" : 1
},
{
"_id" : 1,
"host" : "192.168.77.130:27017",
"arbiterOnly" : false,
"buildIndexes" : true,
"hidden" : false,
"priority" : 2,
"tags" : {
},
"slaveDelay" : NumberLong(0),
"votes" : 1
},
{
"_id" : 2,
"host" : "192.168.77.134:27017",
"arbiterOnly" : false,
"buildIndexes" : true,
"hidden" : false,
"priority" : 1,
"tags" : {
},
"slaveDelay" : NumberLong(0),
"votes" : 1
}
],
"settings" : {
"chainingAllowed" : true,
"heartbeatIntervalMillis" : 2000,
"heartbeatTimeoutSecs" : 10,
"electionTimeoutMillis" : 10000,
"catchUpTimeoutMillis" : -1,
"catchUpTakeoverDelayMillis" : 30000,
"getLastErrorModes" : {
},
"getLastErrorDefaults" : {
"w" : 1,
"wtimeout" : 0
},
"replicaSetId" : ObjectId("5a542a65e491a43160eb92f0")
}
}
zero:PRIMARY>如上，可以看到每个机器权重的变化，192.168.77.128也自动切换回主角色了。如果192.168.77.128再宕掉的话，那么192.168.77.130就会是候选主节点，因为除了192.168.77.128之外它的权重最高。
本文转自 ZeroOne01 51CTO博客，原文链接：http://blog.51cto.com/zero01/2059033，如需转载请自行联系原作者