在上一篇文章 提到了几个问题还没有解决。
- 主节点挂了能否自动切换连接？目前需要手工切换。
- 主节点的读写压力过大如何解决？
- 从节点每个上面的数据都是对数据库全量拷贝，从节点压力会不会过大？
- 数据压力大到机器支撑不了的时候能否做到自动扩展？
这篇文章看完这些问题就可以搞定了。NoSQL的产生就是为了解决大数据量、高扩展性、高性能、灵活数据模型、高可用性。但是光通过主从模式的架构远远达不到上面几点，由此MongoDB设计了副本集和分片的功能。这篇文章主要介绍副本集：
mongoDB官方已经不建议使用主从模式了，替代方案是采用副本集的模式， ，如图：
那什么是副本集呢？打魔兽世界总说打副本，其实这两个概念差不多一个意思。游戏里的副本是指玩家集中在高峰时间去一个场景打怪，会出现玩家暴多怪物少的情况，游戏开发商为了保证玩家的体验度，就为每一批玩家单独开放一个同样的空间同样的数量的怪物，这一个复制的场景就是一个副本，不管有多少个玩家各自在各自的副本里玩不会互相影响。 mongoDB的副本也是这个，主从模式其实就是一个单副本的应用，没有很好的扩展性和容错性。而副本集具有多个副本保证了容错性，就算一个副本挂掉了还有很多副本存在，并且解决了上面第一个问题“主节点挂掉了，整个集群内会自动切换”。难怪mongoDB官方推荐使用这种模式。我们来看看mongoDB副本集的架构图：
由图可以看到客户端连接到整个副本集，不关心具体哪一台机器是否挂掉。主服务器负责整个副本集的读写，副本集定期同步数据备份，一但主节点挂掉，副本节点就会选举一个新的主服务器，这一切对于应用服务器不需要关心。我们看一下主服务器挂掉后的架构：
副本集中的副本节点在主节点挂掉后通过心跳机制检测到后，就会在集群内发起主节点的选举机制，自动选举一位新的主服务器。看起来很牛X的样子，我们赶紧操作部署一下！
官方推荐的副本集机器数量为至少3个，那我们也按照这个数量配置测试。
1、准备两台机器 192.168.1.136、192.168.1.137、192.168.1.138。 192.168.1.136 当作副本集主节点，192.168.1.137、192.168.1.138作为副本集副本节点。
2、分别在每台机器上建立mongodb副本集测试文件夹
|
1
2
3
4
5
6
7
8
|
#存放整个mongodb文件mkdir -p /data/mongodbtest/replset#存放mongodb数据文件mkdir -p /data/mongodbtest/replset/data#进入mongodb文件夹cd /data/mongodbtest |
3、下载mongodb的安装程序包
|
1
|
wget http://fastdl.mongodb.org/linux/mongodb-linux-x86_64-2.4.8.tgz |
注意linux生产环境不能安装32位的mongodb，因为32位受限于操作系统最大2G的文件限制。
|
1
2
|
#解压下载的压缩包 tar xvzf mongodb-linux-x86_64-2.4.8.tgz |
4、分别在每台机器上启动mongodb
|
1
|
/data/mongodbtest/mongodb-linux-x86_64-2.4.8/bin/mongod --dbpath /data/mongodbtest/replset/data --replSet repset |
可以看到控制台上显示副本集还没有配置初始化信息。
|
1
2
|
Sun Dec 29 20:12:02.953 [rsStart] replSet can‘t get local.system.replset config from self or any seed (EMPTYCONFIG)Sun Dec 29 20:12:02.953 [rsStart] replSet info you may need to run replSetInitiate -- rs.initiate() in the shell -- if that is not already done |
5、初始化副本集
在三台机器上任意一台机器登陆mongodb
|
1
2
3
4
|
/data/mongodbtest/mongodb-linux-x86_64-2.4.8/bin/mongo#使用admin数据库use admin |
#定义副本集配置变量，这里的 _id:”repset” 和上面命令参数“ –replSet repset” 要保持一样。
|
1
2
3
4
5
|
config = { _id:"repset", members:[... {_id:0,host:"192.168.1.136:27017"},... {_id:1,host:"192.168.1.137:27017"},... {_id:2,host:"192.168.1.138:27017"}]... } |
#输出
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
|
{ "_id" : "repset", "members" : [ { "_id" : 0, "host" : "192.168.1.136:27017" }, { "_id" : 1, "host" : "192.168.1.137:27017" }, { "_id" : 2, "host" : "192.168.1.138:27017" } ]} |
|
1
2
|
#初始化副本集配置rs.initiate(config); |
#输出成功
|
1
2
3
4
|
{ "info" : "Config now saved locally. Should come online in about a minute.", "ok" : 1} |
#查看日志，副本集启动成功后，138为主节点PRIMARY，136、137为副本节点SECONDARY。
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
|
Sun Dec 29 20:26:13.842 [conn3] replSet replSetInitiate admin command received from clientSun Dec 29 20:26:13.842 [conn3] replSet replSetInitiate config object parses ok, 3 members specifiedSun Dec 29 20:26:13.847 [conn3] replSet replSetInitiate all members seem upSun Dec 29 20:26:13.848 [conn3] ******Sun Dec 29 20:26:13.848 [conn3] creating replication oplog of size: 990MB...Sun Dec 29 20:26:13.849 [FileAllocator] allocating new datafile /data/mongodbtest/replset/data/local.1, filling with zeroes...Sun Dec 29 20:26:13.862 [FileAllocator] done allocating datafile /data/mongodbtest/replset/data/local.1, size: 1024MB, took 0.012 secsSun Dec 29 20:26:13.863 [conn3] ******Sun Dec 29 20:26:13.863 [conn3] replSet info saving a newer config version to local.system.replsetSun Dec 29 20:26:13.864 [conn3] replSet saveConfigLocally doneSun Dec 29 20:26:13.864 [conn3] replSet replSetInitiate config now saved locally. Should come online in about a minute.Sun Dec 29 20:26:23.047 [rsStart] replSet I am 192.168.1.138:27017Sun Dec 29 20:26:23.048 [rsStart] replSet STARTUP2Sun Dec 29 20:26:23.049 [rsHealthPoll] replSet member 192.168.1.137:27017 is upSun Dec 29 20:26:23.049 [rsHealthPoll] replSet member 192.168.1.136:27017 is upSun Dec 29 20:26:24.051 [rsSync] replSet SECONDARYSun Dec 29 20:26:25.053 [rsHealthPoll] replset info 192.168.1.136:27017 thinks that we are downSun Dec 29 20:26:25.053 [rsHealthPoll] replSet member 192.168.1.136:27017 is now in state STARTUP2Sun Dec 29 20:26:25.056 [rsMgr] not electing self, 192.168.1.136:27017 would veto with ‘I don‘t think 192.168.1.138:27017 is electable‘Sun Dec 29 20:26:31.059 [rsHealthPoll] replset info 192.168.1.137:27017 thinks that we are downSun Dec 29 20:26:31.059 [rsHealthPoll] replSet member 192.168.1.137:27017 is now in state STARTUP2Sun Dec 29 20:26:31.062 [rsMgr] not electing self, 192.168.1.137:27017 would veto with ‘I don‘t think 192.168.1.138:27017 is electable‘Sun Dec 29 20:26:37.074 [rsMgr] replSet info electSelf 2Sun Dec 29 20:26:38.062 [rsMgr] replSet PRIMARYSun Dec 29 20:26:39.071 [rsHealthPoll] replSet member 192.168.1.137:27017 is now in state RECOVERINGSun Dec 29 20:26:39.075 [rsHealthPoll] replSet member 192.168.1.136:27017 is now in state RECOVERINGSun Dec 29 20:26:42.201 [slaveTracking] build index local.slaves { _id: 1 }Sun Dec 29 20:26:42.207 [slaveTracking] build index done. scanned 0 total records. 0.005 secsSun Dec 29 20:26:43.079 [rsHealthPoll] replSet member 192.168.1.136:27017 is now in state SECONDARYSun Dec 29 20:26:49.080 [rsHealthPoll] replSet member 192.168.1.137:27017 is now in state SECONDARY |
|
1
2
|
#查看集群节点的状态 rs.status(); |
#输出
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
|
{ "set" : "repset", "date" : ISODate("2013-12-29T12:54:25Z"), "myState" : 1, "members" : [ { "_id" : 0, "name" : "192.168.1.136:27017", "health" : 1, "state" : 2, "stateStr" : "SECONDARY", "uptime" : 1682, "optime" : Timestamp(1388319973, 1), "optimeDate" : ISODate("2013-12-29T12:26:13Z"), "lastHeartbeat" : ISODate("2013-12-29T12:54:25Z"), "lastHeartbeatRecv" : ISODate("2013-12-29T12:54:24Z"), "pingMs" : 1, "syncingTo" : "192.168.1.138:27017" }, { "_id" : 1, "name" : "192.168.1.137:27017", "health" : 1, "state" : 2, "stateStr" : "SECONDARY", "uptime" : 1682, "optime" : Timestamp(1388319973, 1), "optimeDate" : ISODate("2013-12-29T12:26:13Z"), "lastHeartbeat" : ISODate("2013-12-29T12:54:25Z"), "lastHeartbeatRecv" : ISODate("2013-12-29T12:54:24Z"), "pingMs" : 1, "syncingTo" : "192.168.1.138:27017" }, { "_id" : 2, "name" : "192.168.1.138:27017", "health" : 1, "state" : 1, "stateStr" : "PRIMARY", "uptime" : 2543, "optime" : Timestamp(1388319973, 1), "optimeDate" : ISODate("2013-12-29T12:26:13Z"), "self" : true } ], "ok" : 1} |
整个副本集已经搭建成功了。
6、测试副本集数据复制功能
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
|
#在主节点192.168.1.138 上连接到终端：mongo 127.0.0.1#建立test 数据库。use test;往testdb表插入数据。> db.testdb.insert({"test1":"testval1"})#在副本节点 192.168.1.136、192.168.1.137 上连接到mongodb查看数据是否复制过来。/data/mongodbtest/mongodb-linux-x86_64-2.4.8/bin/mongo 192.168.1.136:27017#使用test 数据库。repset:SECONDARY> use test;repset:SECONDARY> show tables; |
#输出
|
1
|
Sun Dec 29 21:50:48.590 error: { "$err" : "not master and slaveOk=false", "code" : 13435 } at src/mongo/shell/query.js:128 |
|
1
2
3
4
5
|
#mongodb默认是从主节点读写数据的，副本节点上不允许读，需要设置副本节点可以读。repset:SECONDARY> db.getMongo().setSlaveOk();#可以看到数据已经复制到了副本集。repset:SECONDARY> db.testdb.find(); |
|
1
2
|
#输出{ "_id" : ObjectId("52c028460c7505626a93944f"), "test1" : "testval1" } |
7、测试副本集故障转移功能
先停掉主节点mongodb 138，查看136、137的日志可以看到经过一系列的投票选择操作，137 当选主节点，136从137同步数据过来。
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
|
Sun Dec 29 22:03:05.351 [rsBackgroundSync] replSet sync source problem: 10278 dbclient error communicating with server: 192.168.1.138:27017Sun Dec 29 22:03:05.354 [rsBackgroundSync] replSet syncing to: 192.168.1.138:27017Sun Dec 29 22:03:05.356 [rsBackgroundSync] repl: couldn‘t connect to server 192.168.1.138:27017Sun Dec 29 22:03:05.356 [rsBackgroundSync] replSet not trying to sync from 192.168.1.138:27017, it is vetoed for 10 more secondsSun Dec 29 22:03:05.499 [rsHealthPoll] DBClientCursor::init call() failedSun Dec 29 22:03:05.499 [rsHealthPoll] replset info 192.168.1.138:27017 heartbeat failed, retryingSun Dec 29 22:03:05.501 [rsHealthPoll] replSet info 192.168.1.138:27017 is down (or slow to respond):Sun Dec 29 22:03:05.501 [rsHealthPoll] replSet member 192.168.1.138:27017 is now in state DOWNSun Dec 29 22:03:05.511 [rsMgr] not electing self, 192.168.1.137:27017 would veto with ‘192.168.1.136:27017 is trying to elect itself but 192.168.1.138:27017 is already primary and more up-to-date‘Sun Dec 29 22:03:07.330 [conn393] replSet info voting yea for 192.168.1.137:27017 (1)Sun Dec 29 22:03:07.503 [rsHealthPoll] replset info 192.168.1.138:27017 heartbeat failed, retryingSun Dec 29 22:03:08.462 [rsHealthPoll] replSet member 192.168.1.137:27017 is now in state PRIMARYSun Dec 29 22:03:09.359 [rsBackgroundSync] replSet syncing to: 192.168.1.137:27017Sun Dec 29 22:03:09.507 [rsHealthPoll] replset info 192.168.1.138:27017 heartbeat failed, retrying |
查看整个集群的状态，可以看到138为状态不可达。
|
1
2
3
|
/data/mongodbtest/mongodb-linux-x86_64-2.4.8/bin/mongo 192.168.1.136:27017repset:SECONDARY> rs.status(); |
#输出
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
|
{ "set" : "repset", "date" : ISODate("2013-12-29T14:28:35Z"), "myState" : 2, "syncingTo" : "192.168.1.137:27017", "members" : [ { "_id" : 0, "name" : "192.168.1.136:27017", "health" : 1, "state" : 2, "stateStr" : "SECONDARY", "uptime" : 9072, "optime" : Timestamp(1388324934, 1), "optimeDate" : ISODate("2013-12-29T13:48:54Z"), "self" : true }, { "_id" : 1, "name" : "192.168.1.137:27017", "health" : 1, "state" : 1, "stateStr" : "PRIMARY", "uptime" : 7329, "optime" : Timestamp(1388324934, 1), "optimeDate" : ISODate("2013-12-29T13:48:54Z"), "lastHeartbeat" : ISODate("2013-12-29T14:28:34Z"), "lastHeartbeatRecv" : ISODate("2013-12-29T14:28:34Z"), "pingMs" : 1, "syncingTo" : "192.168.1.138:27017" }, { "_id" : 2, "name" : "192.168.1.138:27017", "health" : 0, "state" : 8, "stateStr" : "(not reachable/healthy)", "uptime" : 0, "optime" : Timestamp(1388324934, 1), "optimeDate" : ISODate("2013-12-29T13:48:54Z"), "lastHeartbeat" : ISODate("2013-12-29T14:28:35Z"), "lastHeartbeatRecv" : ISODate("2013-12-29T14:28:23Z"), "pingMs" : 0, "syncingTo" : "192.168.1.137:27017" } ], "ok" : 1} |
再启动原来的主节点 138，发现138 变为 SECONDARY，还是137 为主节点 PRIMARY。
|
1
2
3
4
5
6
7
8
|
Sun Dec 29 22:21:06.619 [rsStart] replSet I am 192.168.1.138:27017Sun Dec 29 22:21:06.619 [rsStart] replSet STARTUP2Sun Dec 29 22:21:06.627 [rsHealthPoll] replset info 192.168.1.136:27017 thinks that we are downSun Dec 29 22:21:06.627 [rsHealthPoll] replSet member 192.168.1.136:27017 is upSun Dec 29 22:21:06.627 [rsHealthPoll] replSet member 192.168.1.136:27017 is now in state SECONDARYSun Dec 29 22:21:07.628 [rsSync] replSet SECONDARYSun Dec 29 22:21:08.623 [rsHealthPoll] replSet member 192.168.1.137:27017 is upSun Dec 29 22:21:08.624 [rsHealthPoll] replSet member 192.168.1.137:27017 is now in state PRIMARY |
8、java程序连接副本集测试。三个节点有一个节点挂掉也不会影响应用程序客户端对整个副本集的读写！
