在之前我们使用mysql数据库或者redis时大家广泛用到，采用双机备份后主节点挂掉了后从节点可以接替主机继续服务。所以这种模式比单节点的高可用性要好很多。
1、环境准备
实际应用中，需要两台机器一主一从。我这里因资源问题，使用一台机器解决。
192.168.221.161:27021当做master
192.168.221.161:27022当做slave
2、分别建立两个文件夹/data/db_master，/data/db_slave
3、分别配置两个配置文件mongodb_master.conf，mongodb_salve.conf
[root@MidApp mongodb]# cat mongodb_master.conf dbpath=/data/db_master logpath=/usr/local/mongodb/logs/mongodb_master.log logappend=true port=27021 fork=true auth=false nohttpinterface=false bind_ip=192.168.221.161 journal=true quiet=true master=true
[root@MidApp mongodb]# cat mongodb_salve.conf dbpath=/data/db_slave logpath=/usr/local/mongodb/logs/mongodb_slave.log logappend=true port=27022 fork=true auth=false nohttpinterface=false bind_ip=192.168.221.161 journal=true quiet=true slave=true source=192.168.221.161:27021
4. 启动master节点
[root@MidApp mongodb]# mongod -f mongodb_master.conf about to fork child process, waiting until server is ready for connections. forked process: 55489 child process started successfully, parent exiting
Master上输出日志内容如下：
2017-11-14T09:02:39.999-0800 I CONTROL [initandlisten] options: { config: "mongodb_master.conf", master: true, net: { bindIp: "192.168.221.161", http: { enabled: true }, port: 27021 }, processManagement: { fork: true }, security: { authorization: "disabled" }, storage: { dbPath: "/data/db_master", journal: { enabled: true } }, systemLog: { destination: "file", logAppend: true, path: "/usr/local/mongodb/logs/mongodb_master.log", quiet: true } }
2017-11-14T09:02:40.014-0800 I INDEX [initandlisten] allocating new ns file /data/db_master/local.ns, filling with zeroes...
2017-11-14T09:02:40.018-0800 I NETWORK [websvr] admin web console waiting for connections on port 28021
2017-11-14T09:02:40.254-0800 I STORAGE [FileAllocator] allocating new datafile /data/db_master/local.0, filling with zeroes...
2017-11-14T09:02:40.254-0800 I STORAGE [FileAllocator] creating directory /data/db_master/_tmp
2017-11-14T09:02:40.263-0800 I STORAGE [FileAllocator] done allocating datafile /data/db_master/local.0, size: 64MB, took 0.001 secs
2017-11-14T09:02:40.318-0800 I REPL [initandlisten] ******
2017-11-14T09:02:40.318-0800 I REPL [initandlisten] creating replication oplog of size: 990MB...
2017-11-14T09:02:40.319-0800 I STORAGE [FileAllocator] allocating new datafile /data/db_master/local.1, filling with zeroes...
2017-11-14T09:02:40.349-0800 I STORAGE [FileAllocator] done allocating datafile /data/db_master/local.1, size: 1024MB, took 0.029 secs
2017-11-14T09:02:40.354-0800 I REPL [initandlisten] ******
2017-11-14T09:02:40.358-0800 I NETWORK [initandlisten] waiting for connections on port 270215. 启动从节点
[root@MidApp mongodb]# mongod -f mongodb_salve.conf about to fork child process, waiting until server is ready for connections. forked process: 55577 child process started successfully, parent exiting
Slave上输出日志如下：
2017-11-14T09:05:10.757-0800 I CONTROL [initandlisten] allocator: tcmalloc
2017-11-14T09:05:10.757-0800 I CONTROL [initandlisten] options: { config: "mongodb_salve.conf", net: { bindIp: "192.168.221.161", http: { enabled: true }, port: 27022 }, processManagement: { fork: true }, security: { authorization: "disabled" }, slave: true, source: "192.168.221.161:27021", storage: { dbPath: "/data/db_slave", journal: { enabled: true } }, systemLog: { destination: "file", logAppend: true, path: "/usr/local/mongodb/logs/mongodb_slave.log", quiet: true } }
2017-11-14T09:05:10.759-0800 I INDEX [initandlisten] allocating new ns file /data/db_slave/local.ns, filling with zeroes...
2017-11-14T09:05:10.763-0800 I NETWORK [websvr] admin web console waiting for connections on port 28022
2017-11-14T09:05:11.039-0800 I STORAGE [FileAllocator] allocating new datafile /data/db_slave/local.0, filling with zeroes...
2017-11-14T09:05:11.039-0800 I STORAGE [FileAllocator] creating directory /data/db_slave/_tmp
2017-11-14T09:05:11.045-0800 I STORAGE [FileAllocator] done allocating datafile /data/db_slave/local.0, size: 64MB, took 0.001 secs
2017-11-14T09:05:11.066-0800 I NETWORK [initandlisten] waiting for connections on port 270226. 验证一下主从复制
登录master插入数据：
[root@MidApp mongodb]# mongo 192.168.221.161:27021
MongoDB shell version: 3.0.6
connecting to: 192.168.221.161:27021/test
> show dbs
local 1.078GB
> use test
switched to db test
> db.testdb.insert({"test1":"item1"})
WriteResult({ "nInserted" : 1 })
> db.testdb.find()
{ "_id" : ObjectId("5a0b23702c969986fca777ed"), "test1" : "item1" }登录slave节点，验证数据：
[root@MidApp mongodb]# mongo 192.168.221.161:27022
MongoDB shell version: 3.0.6
connecting to: 192.168.221.161:27022/test
> show dbs
2017-11-14T09:07:45.745-0800 E QUERY Error: listDatabases failed:{ "note" : "from execCommand", "ok" : 0, "errmsg" : "not master" }
at Error (<anonymous>)
at Mongo.getDBs (src/mongo/shell/mongo.js:47:15)
at shellHelper.show (src/mongo/shell/utils.js:630:33)
at shellHelper (src/mongo/shell/utils.js:524:36)
at (shellhelp2):1:1 at src/mongo/shell/mongo.js:47
> db.getMongo().setSlaveOk()#由于slave节点默认是不可读的，需要先执行这句命令才能查询数据
> show dbs
local 0.078GB
> show dbs
local 0.078GB
test 0.078GB
> db.testdb.find()
{ "_id" : ObjectId("5a0b23702c969986fca777ed"), "test1" : "item1" }从节点上可以通过命令db.printReplicationInfo()查看服务状态：
> db.printReplicationInfo() this is a slave, printing slave replication info. source: 192.168.221.161:27021 syncedTo: Tue Nov 14 2017 09:12:04 GMT-0800 (PST) 7 secs (0 hrs) behind the freshest member (no primary available at the moment)
至此，mongodb的主从架构已经完成了。为防止机器负载过大，也可以配置一主多从服务，master节点只负责写操作，slave节点只提供读操作。
不过，存在几个疑问：
a. 从服务器可以当做主服务器吗，也就是从服务器可写吗？
试一下：
> db.testdb.insert({"test3":"item3"})
WriteResult({ "writeError" : { "code" : undefined, "errmsg" : "not master" } })可以看到，slave节点是不可写的
b.如果master挂掉，从服务器会自动接管主服务，变为可写吗？
把master进程杀掉：
kill -9 `ps -ef|grep mongod|grep -v grep|awk ‘{print $2}‘`测试slave节点：
> db.testdb.insert({"test3":"item3"})
WriteResult({ "writeError" : { "code" : undefined, "errmsg" : "not master" } })发现，还是不可写！
这种情况只能手动干预了。。。
继续思考一个问题：
我们怎么实现主节点挂了之后能够自动切换？下一篇接着学习mongodb的集群搭建
本文出自 “清淼淡写” 博客，请务必保留此出处http://qingmiao.blog.51cto.com/7286083/1981606
原文地址：http://qingmiao.blog.51cto.com/7286083/1981606
