1. 副本集的特性
1)一台主服务器(Primary),多台从服务器(Secondary)
2)Primary挂了之后，从服务器自动完成从它们之中选举一台服务器作为主服务器，继续工作，这就解决了单点故障，因此，在这种情况下，MongoDB集群能够继续工作
3)挂了的主服务器恢复到集群中只能以Secondary服务器的角色加入进来
2. 副本集环境搭建
2.1副本集闭环配置(以三台机器配置为例)
port=27017 bind_ip=127.0.0.1 dbpath=/home/tom/development/mongoDBdata/replicaA replSet=child/127.0.0.1:27018
port=27018 bind_ip=127.0.0.1 dbpath=/home/tom/development/mongoDBdata/replicaB replSet=child/127.0.0.1:27019
port=27019 bind_ip=127.0.0.1 dbpath=/home/tom/development/mongoDBdata/replicaC replSet=child/127.0.0.1:27017
2.2 查看副本集状态
rs.status()
2.3 启动三台服务器
在~/development/mongoDBConfig目录建立三个配置文件A.conf,B.conf和C.conf,内容分别是上面闭环设置的3台机器的配置，如A.conf的内容是
port=27017 bind_ip=127.0.0.1 dbpath=/home/tom/development/mongoDBdata/replicaA replSet=child/127.0.0.1:27018
在~/devlopment/mongDBConfig目录下，运行命令启动三台服务器
mongd --config A.conf mongd --config B.conf mongd --config C.conf
此时三台服务器都在输出如下类似的日志，不断尝试跟指定的conf文件中指定的child host进行通信(contact)，例如27017的
Wed Aug 6 07:26:19.088 [rsStart] trying to contact 127.0.0.1:27018 Wed Aug 6 07:26:19.089 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG) Wed Aug 6 07:26:29.089 [rsStart] trying to contact 127.0.0.1:27018 Wed Aug 6 07:26:29.090 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG) Wed Aug 6 07:26:39.090 [rsStart] trying to contact 127.0.0.1:27018 Wed Aug 6 07:26:39.091 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG) Wed Aug 6 07:26:49.091 [rsStart] trying to contact 127.0.0.1:27018 Wed Aug 6 07:26:49.092 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG) Wed Aug 6 07:26:59.092 [rsStart] trying to contact 127.0.0.1:27018 Wed Aug 6 07:26:59.094 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG)
Wed Aug 6 07:27:38.614 [rsStart] trying to contact 127.0.0.1:27019 Wed Aug 6 07:27:38.615 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG) Wed Aug 6 07:27:48.615 [rsStart] trying to contact 127.0.0.1:27019 Wed Aug 6 07:27:48.616 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG) Wed Aug 6 07:27:58.616 [rsStart] trying to contact 127.0.0.1:27019 Wed Aug 6 07:27:58.617 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG) Wed Aug 6 07:28:08.617 [rsStart] trying to contact 127.0.0.1:27019 Wed Aug 6 07:28:08.618 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG) Wed Aug 6 07:28:18.618 [rsStart] trying to contact 127.0.0.1:27019 Wed Aug 6 07:28:18.619 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG)
Wed Aug 6 07:28:35.770 [rsStart] trying to contact 127.0.0.1:27017 Wed Aug 6 07:28:35.770 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG) Wed Aug 6 07:28:45.771 [rsStart] trying to contact 127.0.0.1:27017 Wed Aug 6 07:28:45.772 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG) Wed Aug 6 07:28:55.772 [rsStart] trying to contact 127.0.0.1:27017 Wed Aug 6 07:28:55.773 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG) Wed Aug 6 07:29:05.773 [rsStart] trying to contact 127.0.0.1:27017 Wed Aug 6 07:29:05.774 [rsStart] replSet can't get local.system.replset config from self or any seed (EMPTYCONFIG)
2.4 副本集环境初始化
上面的日志表明副本集还没有进行初始化，使用如下命令在任意一台服务器的Javascript Shell上运行，完成副本集初始化
use admin
db.runCommand({"replSetInitiate": {
"_id": "child",
"members": [
{
"_id": 1,
"host": "127.0.0.1:27017"
},
{
"_id": 2,
"host": "127.0.0.1:27018"
},
{
"_id": 3,
"host": "127.0.0.1:27019"
}
]
}
});
此时查看副本集的服务器日志输出，没台机器有如下的日志输出，表明闭环建立成功
Wed Aug 6 07:45:45.954 [initandlisten] connection accepted from 127.0.0.1:46181 #8 (3 connections now open)
rs.status()察看副本集状态，得到如下的输出，可见27017是Primary服务器，其它的是Secodary服务器
child:PRIMARY> rs.status();
{
"set" : "child",
"date" : ISODate("2014-08-05T23:56:21Z"),
"myState" : 1,
"members" : [
{
"_id" : 1,
"name" : "127.0.0.1:27017",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"uptime" : 2338,
"optime" : Timestamp(1407282254, 1),
"optimeDate" : ISODate("2014-08-05T23:44:14Z"),
"self" : true
},
{
"_id" : 2,
"name" : "127.0.0.1:27018",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 722,
"optime" : Timestamp(1407282254, 1),
"optimeDate" : ISODate("2014-08-05T23:44:14Z"),
"lastHeartbeat" : ISODate("2014-08-05T23:56:19Z"),
"lastHeartbeatRecv" : ISODate("2014-08-05T23:56:20Z"),
"pingMs" : 0,
"syncingTo" : "127.0.0.1:27017"
},
{
"_id" : 3,
"name" : "127.0.0.1:27019",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 720,
"optime" : Timestamp(1407282254, 1),
"optimeDate" : ISODate("2014-08-05T23:44:14Z"),
"lastHeartbeat" : ISODate("2014-08-05T23:56:19Z"),
"lastHeartbeatRecv" : ISODate("2014-08-05T23:56:20Z"),
"pingMs" : 0,
"syncingTo" : "127.0.0.1:27017"
}
],
"ok" : 1
}
2.5 验证
2.5.1.在主服务器27017上执行如下命令：
use foolbar
db.persons.insert({"name":"tom"});
db.persons.find();
>>>{ "_id" : ObjectId("53e1709f07c1625ed566a67f"), "name" : "tom" }
上面创建了一个数据库foolbar，一个集合persons，插入一条文档,然后执行查询
2.5.2.在从服务器27018上执行查询
use foolbar
db.persons.find()
>>>error: { "$err" : "not master and slaveOk=false", "code" : 13435 }
查询错误，提示27018是Secondary服务器，但是SlaveOk是false，不能做查询操作，所以需要设置从服务器，使之能够支持数据读取操作
2.5.3.关掉主服务器27017
那么MongoDB从27018和27019上自动选举一台作为主服务器，此时如果重启27017，那么27017将作为Secondary服务器加入到集群中。假如27019被选举为Primary，那么对27019做针对foolbar数据库persons集合的查询，会查询来1中创建的数据
2.5.4.设置读写分离
无论是主从复制还是副本集复制，一方面提供数据的备份，另外一点通过读写分离提高读写性能，在副本集复制方式下，默认情况下，从服务器只进行数据的同步，但是不提供数据读取功能，要想使从服务器提供数据读取功能，需要设置从服务器的slaveOkay为true，但是遗憾的是，如此重要的设置，MongoDB的Javascript Shell竟然没有提供修改的API，难道副本集的本意只做数据同步备份，不做读写分离？It doesn't make any sense!!!要修改slaveOkay，只能通过MongoDB的语言客户端如Java进行修改，这在以后再讲。