|
一、部署主从复制
1、主服务器和从服务器必须开启安全认证:--auth
2、主服务器和从服务器的admin数据库中必须有全局用户。
3、主服务器的local数据库和从服务器的local数据均有名为repl且密码相同的用户名。
主服务器设置：
dongguo@mongodb:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
> use admin
switched to db admin
> db.addUser('rootm','rootm')
{
"user" : "rootm",
"readOnly" : false,
"pwd" : "aa6526e3b7cbcecc18b2bd822f7c3547",
"_id" : ObjectId("50659e14d2fe6be605337c18")
}
> db.addUser('repl','repl')
{
"user" : "repl",
"readOnly" : false,
"pwd" : "c9f242649c23670ff94c4ca00ea06fe7",
"_id" : ObjectId("5065a85eccf77b17681365b7")
}
> use cloud
switched to db cloud
> db.addUser('repl','repl')
{
"_id" : ObjectId("5065a7cbb70f43c4d157e8ec"),
"user" : "repl",
"readOnly" : false,
"pwd" : "c9f242649c23670ff94c4ca00ea06fe7"
}
> use local
switched to db local
> db.addUser('repl','repl')
{
"user" : "repl",
"readOnly" : false,
"pwd" : "c9f242649c23670ff94c4ca00ea06fe7",
"_id" : ObjectId("50659e2cd2fe6be605337c19")
}
> exit
bye
dongguo@mongodb:~$ sudo /etc/init.d/mongod stop
dongguo@mongodb:~$ sudo vim /opt/mongodb/etc/mongod.conf
修改如下设置：
master = true
auth = true
dongguo@mongodb:~$ sudo /etc/init.d/mongod start
dongguo@mongodb:~$ vim /etc/init.d/mongod
增加用户名与密码属性
$CLIEXEC admin -u rootm -p rootm --eval "db.shutdownServer()"
在日志中可以看到auth与master参数都已经被启用。
dongguo@mongodb:~$ tailf /opt/mongodb/log/mongodb.log
Fri Sep 28 20:58:33 [initandlisten] MongoDB starting : pid=15992 port=27017 dbpath=/opt/mongodb/data master=1 64-bit
host=mongodb
Fri Sep 28 20:58:33 [initandlisten] db version v2.2.0, pdfile version 4.5
Fri Sep 28 20:58:33 [initandlisten] git version: f5e83eae9cfbec7fb7a071321928f00d1b0c5207
Fri Sep 28 20:58:33 [initandlisten] build info: Linux ip-10-2-29-40 2.6.21.7-2.ec2.v1.2.fc8xen #1 SMP Fri Nov 20 17:48:28 EST
2009 x86_64 BOOST_LIB_VERSION=1_49
Fri Sep 28 20:58:33 [initandlisten] options: { auth: "true", config: "/opt/mongodb/etc/mongod.conf", dbpath:
"/opt/mongodb/data", fork: "true", logappend: "true", logpath: "/opt/mongodb/log/mongodb.log", master: "true", pidfilepath:
"/opt/mongodb/run/mongod.pid", port: 27017 }
Fri Sep 28 20:58:33 [initandlisten] journal dir=/opt/mongodb/data/journal
从服务器设置：
dongguo@mongodb-node1:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
> use admin
switched to db admin
> db.addUser('roots','roots')
{
"user" : "roots",
"readOnly" : false,
"pwd" : "430b42be2092c65349e29043412a9155",
"_id" : ObjectId("50659fc1eb520e528f23454d")
}
> db.addUser('repl','repl')
{
"user" : "repl",
"readOnly" : false,
"pwd" : "c9f242649c23670ff94c4ca00ea06fe7",
"_id" : ObjectId("5065a9772485dc71fdfe1af5")
}
> db.addUser('repl','repl')
{
"user" : "repl",
"readOnly" : false,
"pwd" : "c9f242649c23670ff94c4ca00ea06fe7",
"_id" : ObjectId("5065a9a22485dc71fdfe1af6")
}
> use local
switched to db local
> db.addUser('repl','repl')
{
"user" : "repl",
"readOnly" : false,
"pwd" : "c9f242649c23670ff94c4ca00ea06fe7",
"_id" : ObjectId("50659fd1eb520e528f23454e")
}
> exit
bye
dongguo@mongodb-node1:~$ sudo /etc/init.d/mongod stop
dongguo@mongodb-node1:~$ sudo vim /opt/mongodb/etc/mongod.conf
修改如下设置：
slave = true
auth = true
source = 10.6.1.145:27017
dongguo@mongodb-node1:~$ sudo /etc/init.d/mongod start
dongguo@mongodb-node1:~$ vim /etc/init.d/mongod
增加用户名与密码属性
$CLIEXEC admin -u roots -p roots --eval "db.shutdownServer()"
在日志中可以看到auth与master参数都已经被启用，并且数据也开始进行了复制。
dongguo@mongodb-node1:~$ tailf /opt/mongodb/log/mongodb.log
Fri Sep 28 21:05:26 [initandlisten] options: { auth: "true", config: "/opt/mongodb/etc/mongod.conf", dbpath:
"/opt/mongodb/data", fork: "true", logappend: "true", logpath: "/opt/mongodb/log/mongodb.log", pidfilepath:
"/opt/mongodb/run/mongod.pid", port: 27017, slave: "true", source: "10.6.1.145:27017" }
Fri Sep 28 21:05:26 [initandlisten] journal dir=/opt/mongodb/data/journal
Fri Sep 28 21:05:26 [initandlisten] recover : no journal files present, no recovery needed
Fri Sep 28 21:05:26 [initandlisten] waiting for connections on port 27017
Fri Sep 28 21:05:26 [websvr] admin web console waiting for connections on port 28017
Fri Sep 28 21:05:27 [replslave] build index local.sources { _id: 1 }
Fri Sep 28 21:05:27 [replslave] build index done. scanned 0 total records. 0 secs
Fri Sep 28 21:05:27 [replslave] repl: syncing from host:10.6.1.145:27017
Fri Sep 28 21:05:27 [replslave] build index local.me { _id: 1 }
Fri Sep 28 21:05:27 [replslave] build index done. scanned 0 total records. 0 secs
Fri Sep 28 21:05:36 [replslave] repl: applied 1 operations
Fri Sep 28 21:05:36 [replslave] repl: end sync_pullOpLog syncedTo: Sep 28 21:05:38 5065a0a2:1
Fri Sep 28 21:05:36 [replslave] repl: syncing from host:10.6.1.145:27017
Fri Sep 28 21:05:36 [replslave] resync: dropping database cloud
Fri Sep 28 21:05:36 [replslave] removeJournalFiles
Fri Sep 28 21:05:36 [replslave] resync: cloning database cloud to get an initial copy
Fri Sep 28 21:05:36 [FileAllocator] allocating new datafile /opt/mongodb/data/cloud.ns, filling with zeroes...
Fri Sep 28 21:05:36 [FileAllocator] done allocating datafile /opt/mongodb/data/cloud.ns, size: 16MB, took 0 secs
Fri Sep 28 21:05:36 [FileAllocator] allocating new datafile /opt/mongodb/data/cloud.0, filling with zeroes...
Fri Sep 28 21:05:36 [FileAllocator] done allocating datafile /opt/mongodb/data/cloud.0, size: 64MB, took 0.001 secs
Fri Sep 28 21:05:36 [FileAllocator] allocating new datafile /opt/mongodb/data/cloud.1, filling with zeroes...
Fri Sep 28 21:05:36 [FileAllocator] done allocating datafile /opt/mongodb/data/cloud.1, size: 128MB, took 0.004 secs
Fri Sep 28 21:05:36 [replslave] build index cloud.vm_instance { _id: 1 }
Fri Sep 28 21:05:36 [replslave] fastBuildIndex dupsToDrop:0
Fri Sep 28 21:05:36 [replslave] build index done. scanned 79 total records. 0.014 secs
Fri Sep 28 21:05:36 [replslave] build index cloud.system.users { _id: 1 }
Fri Sep 28 21:05:36 [replslave] fastBuildIndex dupsToDrop:0
Fri Sep 28 21:05:36 [replslave] build index done. scanned 1 total records. 0 secs
Fri Sep 28 21:05:36 [replslave] resync: done with initial clone for db: cloud
Fri Sep 28 21:05:42 [replslave] repl: syncing from host:10.6.1.145:2701
接下来登陆数据库进行手工验证：
dongguo@mongodb-node1:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
> use cloud;
switched to db cloud
> db.auth('repl','repl')
1
> db.vm_instance
db.vm_instance
> db.vm_instance.find()
"uuid" : "1e5de227-547e-4ed0-957a-8884559d01cd", "created" : "2012-06-29 00:40:53" }
{ "_id" : ObjectId("50658bd04ff74b0e8f2f6488"), "id" : 5, "instance_name" : "i-2-5-VM", "private_ip_address" : "10.6.223.41",
"uuid" : "123f8676-2fc1-413a-bdec-89e54835ad6a", "created" : "2012-06-29 01:20:37" }
{ "_id" : ObjectId("50658bd04ff74b0e8f2f6489"), "id" : 6, "instance_name" : "i-2-6-VM", "private_ip_address" : "10.6.223.139",
"uuid" : "7d39898e-cc0d-46b2-a1d4-170201acf832", "created" : "2012-06-29 03:55:32" }
{ "_id" : ObjectId("50658bd04ff74b0e8f2f648a"), "id" : 7, "instance_name" : "i-2-7-VM", "private_ip_address" : "10.6.223.140",
"uuid" : "ad0690bb-6738-42e3-a3ee-5a93b47fd3a4", "created" : "2012-06-29 03:55:52" }
{ "_id" : ObjectId("50658bd04ff74b0e8f2f648b"), "id" : 8, "instance_name" : "i-2-8-VM", "private_ip_address" : "10.6.223.142",
"uuid" : "5beb13b5-8319-43a0-a65f-e7c7dde4334f", "created" : "2012-06-29 03:56:07" }
{ "_id" : ObjectId("50658bd04ff74b0e8f2f648c"), "id" : 9, "instance_name" : "i-2-9-VM", "private_ip_address" : "10.6.223.143",
Type "it" for more
>
bye
数据库cloud中的数据已经成功的同步过来了，主从复制搭建成功。
二、部署副本集群
接下来我们开始搭建副本集群，这是一个更高级方式，和主从复制不同的是：
1. 该集群没有特定的主数据库；
2. 如果某个主数据库宕机了，集群中就会推选出一个从数据库作为主数据库顶上，实现自动故障恢复功能，非常不错。
首先，我们准备三台服务器 mongodb,mongodb-node1,mongodb-node2。
确保只有mongodb上有数据，而mongodb-node1与mongodb-node2上都没有数据，如果有则通过 use dbname和db.dropDatabase()将数据删除。
dongguo@mongodb-node1:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
> show dbs
local (empty)
>
bye
dongguo@mongodb-node2:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
> show dbs
local (empty)
>
bye
dongguo@mongodb:~$ mongo admin
MongoDB shell version: 2.2.0
connecting to: admin
> show dbs
admin 0.203125GB
cloud 0.203125GB
local 1.203125GB
并取消之前在mongodb与mongodb-node1的相关主从设置
dongguo@mongodb:~$ sudo vim /opt/mongodb/etc/mongod.conf
修改如下设置：
#master = true
#auth = true
dongguo@mongodb-node1:~$ sudo vim /opt/mongodb/etc/mongod.conf
修改如下设置：
#slave = true
#auth = true
#source = 10.6.1.145:27017
并修改服务管理脚本
$ vim /etc/init.d/mongod
$CLIEXEC admin --eval "db.shutdownServer()"
然后在三台服务器上加入启动参数 --replSet
需要给集群一个名称，比如MoboTap，--replSet 表示让服务器知道MoboTap下还有其他数据库。
$ sudo vim /opt/mongodb/etc/mongod.conf
replSet = MoboTap
然后重启三台服务器的MongoDB
$ sudo /etc/init.d/mongod stop
$ sudo /etc/init.d/mongod start
在mongodb服务器上初始化整个副本集群
dongguo@mongodb:~$ mongo admin
MongoDB shell version: 2.2.0
connecting to: admin
> show dbs
admin 0.203125GB
cloud 0.203125GB
local 1.203125GB
> db.runCommand({"replSetInitiate":{ ... "_id":"MoboTap", ... "members":[ ... { ... "_id":1, ... "host":"10.6.1.145:27017" ... }, ... { ... "_id":2, ... "host":"10.6.1.146:27017" ... }, ... { ... "_id":3, ... "host":"10.6.1.147:27017" ... } ... ]}}) { "info" : "Config now saved locally. Should come online in about a minute.", "ok" : 1 }
然后我们检查目前整个副本集群的主从状态：
dongguo@mongodb:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
MoboTap:PRIMARY>
dongguo@mongodb-node1:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
MoboTap:SECONDARY>
MoboTap:SECONDARY> show dbs
admin 0.203125GB
cloud 0.203125GB
local 1.203125GB
dongguo@mongodb-node2:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
MoboTap:SECONDARY>
MoboTap:SECONDARY> show dbs
admin 0.203125GB
cloud 0.203125GB
local 1.203125GB
我们可以看到，目前通过内部的自动“选举”，mongodb担任了PRIMARY的角色，而另外的node1与node2则担任了SECONDARY，并且成功的从mongodb中同步了数据。
我们可以通过rs.status()很方便的查看集群中的服务器状态。
dongguo@mongodb:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
MoboTap:PRIMARY> rs.status()
{
"set" : "MoboTap",
"date" : ISODate("2012-09-28T21:38:29Z"),
"myState" : 1,
"members" : [
{
"_id" : 1,
"name" : "10.6.1.145:27017",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"uptime" : 8665,
"optime" : Timestamp(1348866843000, 1),
"optimeDate" : ISODate("2012-09-28T21:14:03Z"),
"self" : true
},
{
"_id" : 2,
"name" : "10.6.1.146:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 1464,
"optime" : Timestamp(1348866843000, 1),
"optimeDate" : ISODate("2012-09-28T21:14:03Z"),
"lastHeartbeat" : ISODate("2012-09-28T21:38:28Z"),
"pingMs" : 0
},
{
"_id" : 3,
"name" : "10.6.1.147:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 1464,
"optime" : Timestamp(1348866843000, 1),
"optimeDate" : ISODate("2012-09-28T21:14:03Z"),
"lastHeartbeat" : ISODate("2012-09-28T21:38:28Z"),
"pingMs" : 0
}
],
"ok" : 1
}
在MongoDB的副本集群中，还有一个角色叫做仲裁服务器，它的特点是只参与投票选举，而不会同步数据。
我们可以通过下面的步骤来新增一台仲裁服务器mongodb-node3。
MongoDB的配置与上面的mongodb-node2完全相同，同样也要保证没有数据且启用了--replSet参数。
启动mongodb-node3之后，我们在mongodb上执行如下操作：
dongguo@mongodb:~$ mongo admin
MongoDB shell version: 2.2.0
connecting to: admin
MoboTap:PRIMARY> rs.addArb("10.6.1.148:27017")
{ "ok" : 1 }
MoboTap:PRIMARY> rs.status()
{
"set" : "MoboTap",
"date" : ISODate("2012-09-28T21:56:32Z"),
"myState" : 1,
"members" : [
{
"_id" : 1,
"name" : "10.6.1.145:27017",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"uptime" : 9748,
"optime" : Timestamp(1348869269000, 1),
"optimeDate" : ISODate("2012-09-28T21:54:29Z"),
"self" : true
},
{
"_id" : 2,
"name" : "10.6.1.146:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 2547,
"optime" : Timestamp(1348869269000, 1),
"optimeDate" : ISODate("2012-09-28T21:54:29Z"),
"lastHeartbeat" : ISODate("2012-09-28T21:56:31Z"),
"pingMs" : 0
},
{
"_id" : 3,
"name" : "10.6.1.147:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 464,
"optime" : Timestamp(1348869269000, 1),
"optimeDate" : ISODate("2012-09-28T21:54:29Z"),
"lastHeartbeat" : ISODate("2012-09-28T21:56:30Z"),
"pingMs" : 0
},
{
"_id" : 4,
"name" : "10.6.1.148:27017",
"health" : 1,
"state" : 7,
"stateStr" : "ARBITER",
"uptime" : 17,
"lastHeartbeat" : ISODate("2012-09-28T21:56:31Z"),
"pingMs" : 0
}
],
"ok" : 1
}
我们可以看到仲裁服务器已经被添加进来了，角色为ARBITER，就是仲裁的意思。
好的，目前整个副本集群就搭建好了，那么这个集群所具备的一个非常好的特性 “自动故障恢复”，我们就可以来好好见识一下了。
下面，我们将目前担任PRIMARY角色的服务器的进程直接Kill掉：
dongguo@mongodb:~$ ps aux | grep mongod
root 17029 0.3 12.9 5480648 64836 ? Sl 03:14 0:31 /opt/mongodb/bin/mongod --config /opt/mongodb/etc/mongod.conf
dongguo 18010 0.0 0.1 6224 584 pts/0 S+ 06:01 0:00 grep --color=auto mongod
dongguo@mongodb:~$ sudo killall -9 mongod
dongguo@mongodb:~$ ps aux | grep mongod
dongguo 18013 0.0 0.1 6224 584 pts/0 S+ 06:01 0:00 grep --color=auto mongod
进程已经被杀掉了，那么目前的主从关系变成了什么样子呢？我们通过mongodb-node1上查看一下：
dongguo@mongodb-node1:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
MoboTap:PRIMARY> rs.status()
{
"set" : "MoboTap",
"date" : ISODate("2012-09-28T22:02:23Z"),
"myState" : 1,
"members" : [
{
"_id" : 1,
"name" : "10.6.1.145:27017",
"health" : 0,
"state" : 8,
"stateStr" : "(not reachable/healthy)",
"uptime" : 0,
"optime" : Timestamp(1348869269000, 1),
"optimeDate" : ISODate("2012-09-28T21:54:29Z"),
"lastHeartbeat" : ISODate("2012-09-28T22:01:12Z"),
"pingMs" : 0,
"errmsg" : "socket exception [CONNECT_ERROR] for 10.6.1.145:27017"
},
{
"_id" : 2,
"name" : "10.6.1.146:27017",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"uptime" : 3111,
"optime" : Timestamp(1348869269000, 1),
"optimeDate" : ISODate("2012-09-28T21:54:29Z"),
"self" : true
},
{
"_id" : 3,
"name" : "10.6.1.147:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 816,
"optime" : Timestamp(1348869269000, 1),
"optimeDate" : ISODate("2012-09-28T21:54:29Z"),
"lastHeartbeat" : ISODate("2012-09-28T22:02:23Z"),
"pingMs" : 0,
"errmsg" : "syncing to: 10.6.1.146:27017"
},
{
"_id" : 4,
"name" : "10.6.1.148:27017",
"health" : 1,
"state" : 7,
"stateStr" : "ARBITER",
"uptime" : 369,
"lastHeartbeat" : ISODate("2012-09-28T22:02:22Z"),
"pingMs" : 0
}
],
"ok" : 1
}
很明显，我们的mongodb-node1自动接管了PRIMARY的角色，并且可以看到mongodb的状态目前变成了(not reachable/healthy)。
接着，我们重新恢复mongodb的服务，再看看主从关系有没有什么变化：
dongguo@mongodb:~$ sudo /etc/init.d/mongod start
dongguo@mongodb:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
MoboTap:SECONDARY> rs.status()
{
"set" : "MoboTap",
"date" : ISODate("2012-09-28T22:04:38Z"),
"myState" : 2,
"syncingTo" : "10.6.1.146:27017",
"members" : [
{
"_id" : 1,
"name" : "10.6.1.145:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 19,
"optime" : Timestamp(1348869269000, 1),
"optimeDate" : ISODate("2012-09-28T21:54:29Z"),
"errmsg" : "syncing to: 10.6.1.146:27017",
"self" : true
},
{
"_id" : 2,
"name" : "10.6.1.146:27017",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"uptime" : 19,
"optime" : Timestamp(1348869269000, 1),
"optimeDate" : ISODate("2012-09-28T21:54:29Z"),
"lastHeartbeat" : ISODate("2012-09-28T22:04:37Z"),
"pingMs" : 0
},
{
"_id" : 3,
"name" : "10.6.1.147:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 19,
"optime" : Timestamp(1348869269000, 1),
"optimeDate" : ISODate("2012-09-28T21:54:29Z"),
"lastHeartbeat" : ISODate("2012-09-28T22:04:37Z"),
"pingMs" : 0
},
{
"_id" : 4,
"name" : "10.6.1.148:27017",
"health" : 1,
"state" : 7,
"stateStr" : "ARBITER",
"uptime" : 19,
"lastHeartbeat" : ISODate("2012-09-28T22:04:37Z"),
"pingMs" : 0
}
],
"ok" : 1
}
我们可以看到mongodb现在已经自动成为SECONDARY，而PRIMARY则仍然由mongodb-node1担任。
下面，我们来实践一下节点的动态增加与删除。
我们新增一台mongodb-node4，相关配置与mongodb-node3方法一致。
启动好mongodb-node4以后，我们在PRIMARY上添加该节点：
dongguo@mongodb-node1:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
MoboTap:PRIMARY> rs.add("10.6.1.149:27017");
{ "ok" : 1 }
MoboTap:PRIMARY> rs.status();
{
"set" : "MoboTap",
"date" : ISODate("2012-09-28T22:21:29Z"),
"myState" : 1,
"members" : [
{
"_id" : 1,
"name" : "10.6.1.145:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 1031,
"optime" : Timestamp(1348870876000, 1),
"optimeDate" : ISODate("2012-09-28T22:21:16Z"),
"lastHeartbeat" : ISODate("2012-09-28T22:21:29Z"),
"pingMs" : 0
},
{
"_id" : 2,
"name" : "10.6.1.146:27017",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"uptime" : 4257,
"optime" : Timestamp(1348870876000, 1),
"optimeDate" : ISODate("2012-09-28T22:21:16Z"),
"self" : true
},
{
"_id" : 3,
"name" : "10.6.1.147:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 1962,
"optime" : Timestamp(1348870876000, 1),
"optimeDate" : ISODate("2012-09-28T22:21:16Z"),
"lastHeartbeat" : ISODate("2012-09-28T22:21:28Z"),
"pingMs" : 1
},
{
"_id" : 4,
"name" : "10.6.1.148:27017",
"health" : 1,
"state" : 7,
"stateStr" : "ARBITER",
"uptime" : 222,
"lastHeartbeat" : ISODate("2012-09-28T22:21:29Z"),
"pingMs" : 0
},
{
"_id" : 5,
"name" : "10.6.1.149:27017",
"health" : 1,
"state" : 5,
"stateStr" : "STARTUP2",
"uptime" : 13,
"optime" : Timestamp(0, 0),
"optimeDate" : ISODate("1970-01-01T00:00:00Z"),
"lastHeartbeat" : ISODate("2012-09-28T22:21:28Z"),
"pingMs" : 0,
"errmsg" : "initial sync need a member to be primary or secondary to do our initial sync"
}
],
"ok" : 1
}
目前数据还未同步完成，等待一段时间过后我们再看一下：
dongguo@mongodb-node4:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
MoboTap:SECONDARY> rs.status()
{
"set" : "MoboTap",
"date" : ISODate("2012-09-29T10:10:06Z"),
"myState" : 2,
"syncingTo" : "10.6.1.146:27017",
"members" : [
{
"_id" : 1,
"name" : "10.6.1.145:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 96,
"optime" : Timestamp(1348870876000, 1),
"optimeDate" : ISODate("2012-09-28T22:21:16Z"),
"lastHeartbeat" : ISODate("2012-09-29T10:10:04Z"),
"pingMs" : 0
},
{
"_id" : 2,
"name" : "10.6.1.146:27017",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"uptime" : 96,
"optime" : Timestamp(1348870876000, 1),
"optimeDate" : ISODate("2012-09-28T22:21:16Z"),
"lastHeartbeat" : ISODate("2012-09-29T10:10:04Z"),
"pingMs" : 1
},
{
"_id" : 3,
"name" : "10.6.1.147:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 96,
"optime" : Timestamp(1348870876000, 1),
"optimeDate" : ISODate("2012-09-28T22:21:16Z"),
"lastHeartbeat" : ISODate("2012-09-29T10:10:04Z"),
"pingMs" : 0
},
{
"_id" : 4,
"name" : "10.6.1.148:27017",
"health" : 1,
"state" : 7,
"stateStr" : "ARBITER",
"uptime" : 96,
"lastHeartbeat" : ISODate("2012-09-29T10:10:04Z"),
"pingMs" : 0
},
{
"_id" : 5,
"name" : "10.6.1.149:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 228,
"optime" : Timestamp(1348870876000, 1),
"optimeDate" : ISODate("2012-09-28T22:21:16Z"),
"self" : true
}
],
"ok" : 1
}
新的节点已经成功加入到了服务器当中。
接着，我们来测试动态的删除节点，同样拿刚刚新增的mongodb-node4为例：
dongguo@mongodb-node1:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
MoboTap:PRIMARY> rs.remove("10.6.1.149:27017");
{ "ok" : 1 }
MoboTap:PRIMARY> rs.status()
{
"set" : "MoboTap",
"date" : ISODate("2012-09-28T22:26:00Z"),
"myState" : 1,
"members" : [
{
"_id" : 1,
"name" : "10.6.1.145:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 62,
"optime" : Timestamp(1348871096000, 1),
"optimeDate" : ISODate("2012-09-28T22:24:56Z"),
"lastHeartbeat" : ISODate("2012-09-28T22:26:00Z"),
"pingMs" : 0,
"errmsg" : "syncing to: 10.6.1.146:27017"
},
{
"_id" : 2,
"name" : "10.6.1.146:27017",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"uptime" : 4528,
"optime" : Timestamp(1348871096000, 1),
"optimeDate" : ISODate("2012-09-28T22:24:56Z"),
"self" : true
},
{
"_id" : 3,
"name" : "10.6.1.147:27017",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 62,
"optime" : Timestamp(1348871096000, 1),
"optimeDate" : ISODate("2012-09-28T22:24:56Z"),
"lastHeartbeat" : ISODate("2012-09-28T22:26:00Z"),
"pingMs" : 0,
"errmsg" : "syncing to: 10.6.1.146:27017"
},
{
"_id" : 4,
"name" : "10.6.1.148:27017",
"health" : 1,
"state" : 7,
"stateStr" : "ARBITER",
"uptime" : 62,
"lastHeartbeat" : ISODate("2012-09-28T22:26:00Z"),
"pingMs" : 0
}
],
"ok" : 1
}
我们可以看到，节点mongodb-node4已经被动态的从副本集群中删除了。
我们登录到mongodb-node4上确认一下：
dongguo@mongodb-node4:~$ mongo
MongoDB shell version: 2.2.0
connecting to: test
MoboTap:REMOVED> rs.status()
{
"set" : "MoboTap",
"date" : ISODate("2012-09-29T10:14:03Z"),
"myState" : 10,
"members" : [
{
"_id" : 5,
"name" : "10.6.1.149:27017",
"health" : 1,
"state" : 10,
"stateStr" : "REMOVED",
"uptime" : 465,
"optime" : Timestamp(1348870876000, 1),
"optimeDate" : ISODate("2012-09-28T22:21:16Z"),
"self" : true
}
],
"ok" : 1
}
很清楚的可以看到自己的状态变成了REMOVED。
至此，整个mongodb的副本集群的自动故障恢复与节点的动态增加与删除我们都成功的实现了。
这种方式明显要比上面的主从复制方式更可用，并且支持多节点，能够进行动态增加与删除节点，实现水平的扩容，非常不错。
mongoDB的两个特性：主从复制、副本集
一、主从复制
1.主从复制是一个简单的数据库同步备份的集群技术.
1.1在数据库集群中要明确的知道谁是主服务器,主服务器只有一台.
1.2从服务器要知道自己的数据源也就是对于的主服务是谁.
1.3--master用来确定主服务器,--slave 和 ?source 来控制从服务器
1.主从复制集群案例
2.主从复制的其他设置项
--only 从节点->指定复制某个数据库,默认是复制全部数据库
--slavedelay 从节点->设置主数据库同步数据的延迟(单位是秒)
--fastsync 从节点->以主数据库的节点快照为节点启动从数据库
--autoresync 从节点->如果不同步则从新同步数据库
--oplogSize 主节点->设置oplog的大小(主节点操作记录存储到local的oplog中)
3.利用shell 动态添加和删除从节点
不难看出从节点中关于主节点的信息全部存到local的sources的集合中
我们只要对集合进行操作就可以动态操作主从关系
挂接主节点:操作之前只留下从数据库服务
db.sources.insert({“host”:”127.0.0.1:8888”})
删除已经挂接的主节点:操作之前只留下从数据库服务
db.sources.remove({“host”:”127.0.0.1:8888”})
二、 副本集概念
1.1第一张图表明A是活跃的B和C是用于备份的
1.2第二张图当A出现了故障,这时候集群根据权重算法推选出B为活跃的数据库
1.3第三张图当A恢复后他自动又会变为备份数据库
 2.初始化副本集
use admin
db.runCommand({"replSetInitiate":
{
"_id":'child',
"members":[{
"_id":1,
"host":"127.0.0.1:1111"
},{
"_id":2,
"host":"127.0.0.1:2222"
},{
"_id":3,
"host":"127.0.0.1:3333"
}]
}
})
2.查看副本集状态
rs.status()
3.节点和初始化高级参数
standard 常规节点:参与投票有可能成为活跃节点
passive 副本节点:参与投票,但是不能成为活跃节点
arbiter 仲裁节点:只是参与投票不复制节点也不能成为活跃节点
4.高级参数
Priority 0到1000之间 ,0代表是副本节点 ,1到1000是常规节点
arbiterOnly : true 仲裁节点
用法
members":[{
"_id":1,
"host":"127.0.0.1:1111“,
arbiterOnly : true
}]”
5.优先级相同时候仲裁组建的规则
对于优先级相同的时候会根据时间，比如：A是活跃节点，B、C为副本集的时候，
当A宕机的时候，mongo会选择B、C中更新时间最近的为活跃节点。
6.读写分离操作à扩展读
6.1一般情况下作为副本的节点是不能进行数据库读操作的
但是在读取密集型的系统中读写分离是十分必要的
6.2设置读写分离
slaveOkay : true
很遗憾他在shell中无法掩饰,这个特性是被写到mongoDB的
驱动程序中的,在java和node等其他语言中可以完成
7.Oplog
他是被存储在本地数据库local中的,他的每一个文档保证这一个节点操作
如果想故障恢复可以更彻底oplog可已经尽量设置大一些用来保存更多的操作
信息
改变oplog大小
主库 --master --oplogSize size
|
