MongoDB主从集群

摘要: MongoDB主从集群

一、环境

$ cat /etc/redhat-release
CentOS Linux release 7.0.1406 (Core)
$ uname -a
Linux zhaopin-2-201 3.10.0-123.el7.x86_64 #1 SMP Mon Jun 30 12:09:22 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux
$ mongo --version
MongoDB shell version: 3.0.6

二、准备

1.创建目录

$ sudo mkdir -p /data/mongodb/{data/db0,backup/db0,log/db0,conf/db0}

2.编写配置文件
1）master：

$ sudo vim /data/mongodb/conf/db0/mongodb.conf
# base
port = 27017
maxConns = 800
filePermissions = 0700
fork = true
noauth = true
directoryperdb = true
dbpath = /data/mongodb/data/db0
pidfilepath = /data/mongodb/data/db0/mongodb.pid
journal = true

# security
nohttpinterface = true
rest = false

# log
logpath = /data/mongodb/log/db0/monodb.log
logRotate = rename
logappend = true
slowms = 50
master = true

2）slave:

$ sudo vim /data/mongodb/conf/db0/mongodb.conf
# base
port = 27017
maxConns = 800
filePermissions = 0700
fork = true
noauth = true
directoryperdb = true
dbpath = /data/mongodb/data/db0
pidfilepath = /data/mongodb/data/db0/mongodb.pid
journal = true

# security
nohttpinterface = true
rest = false

# log
logpath = /data/mongodb/log/db0/mongodb.log
logRotate = rename
logappend = true
slowms = 50

slave = true
source = 172.30.2.201:27017

三、启动

1.先启动master

$ sudo /opt/mongodb/bin/mongod --config /data/mongodb/conf/db0/mongodb.conf
about to fork child process, waiting until server is ready for connections.
forked process: 48583
child process started successfully, parent exiting
$ mongo
MongoDB shell version: 3.0.6
connecting to: test
> db.isMaster();
{
"ismaster" : 0,
"info" : "dead: data too stale halted replication",
"maxBsonObjectSize" : 16777216,
"maxMessageSizeBytes" : 48000000,
"maxWriteBatchSize" : 1000,
"localTime" : ISODate("2015-09-23T07:47:52.957Z"),
"maxWireVersion" : 3,
"minWireVersion" : 0,
"ok" : 1
}
>
bye

2.启动slave

$ sudo /opt/mongodb/bin/mongod --config /data/mongodb/conf/db0/mongodb.conf
about to fork child process, waiting until server is ready for connections.
forked process: 36715
child process started successfully, parent exiting
$ mongo
MongoDB shell version: 3.0.6
connecting to: test
> db.isMaster();
{
"ismaster" : false,
"maxBsonObjectSize" : 16777216,
"maxMessageSizeBytes" : 48000000,
"maxWriteBatchSize" : 1000,
"localTime" : ISODate("2015-09-23T07:49:32.934Z"),
"maxWireVersion" : 3,
"minWireVersion" : 0,
"ok" : 1
}
>
bye

3. 设置开机启动

$ sudo vim /etc/rc.local
/opt/mongodb/bin/mongod --config /data/mongodb/conf/db0/mongodb.conf

四、验证

1.数据同步
1) master：

$ mongo
MongoDB shell version: 3.0.6
connecting to: test
> use aaa;
switched to db aaa
> db.createCollection("test");
{ "ok" : 1 }
> show collections;
system.indexes
system.profile
test
>

2) slave：

$ mongo
MongoDB shell version: 3.0.6
connecting to: test
> rs.slaveOk();
> show dbs;
aaa    0.078GB
local  0.078GB
> use aaa;
switched to db aaa
> show collections;
system.indexes
test
>

2.读写验证
由上可知，master是可读写的

在从库上写测试：

$ mongo
MongoDB shell version: 3.0.6
connecting to: test
> use aaa;
switched to db aaa
> db.createCollection("test2");
{ "note" : "from execCommand", "ok" : 0, "errmsg" : "not master" }
> show collections;
2015-09-23T15:57:49.240+0800 E QUERY    Error: listCollections failed: { "note" : "from execCommand", "ok" : 0, "errmsg" : "not master" }
at Error (<anonymous>)
at DB._getCollectionInfosCommand (src/mongo/shell/db.js:646:15)
at DB.getCollectionInfos (src/mongo/shell/db.js:658:20)
at DB.getCollectionNames (src/mongo/shell/db.js:669:17)
at shellHelper.show (src/mongo/shell/utils.js:625:12)
at shellHelper (src/mongo/shell/utils.js:524:36)
at (shellhelp2):1:1 at src/mongo/shell/db.js:646
> rs.slaveOk();
> show collections;
system.indexes
test
>

证明从库是不能写，默认也是不可读的，需要执行rs.slaveOk();或者db.getMongo().setSlaveOk();并且只对当前session有效，所以每次连接从库都需要执行。

五、故障测试及解决方案

1.master宕机
1) 将master停机

$ mongo
MongoDB shell version: 3.0.6
connecting to: test
> use admin;
switched to db admin
> db.shutdownServer();
2015-09-23T16:02:16.983+0800 I NETWORK  DBClientCursor::init call() failed
server should be down...
2015-09-23T16:02:16.986+0800 I NETWORK  trying reconnect to 127.0.0.1:27017 (127.0.0.1) failed
2015-09-23T16:02:16.987+0800 I NETWORK  reconnect 127.0.0.1:27017 (127.0.0.1) ok
2015-09-23T16:02:16.992+0800 I NETWORK  Socket recv() errno:104 Connection reset by peer 127.0.0.1:27017
2015-09-23T16:02:16.992+0800 I NETWORK  SocketException: remote: 127.0.0.1:27017 error: 9001 socket exception [RECV_ERROR] server [127.0.0.1:27017]
2015-09-23T16:02:16.992+0800 I NETWORK  DBClientCursor::init call() failed
>
bye
$ ps aux | grep mongo | grep -v grep

2) 查看slave状态

$ mongo
MongoDB shell version: 3.0.6
connecting to: test
> rs.slaveOk();
> show dbs;
aaa    0.078GB
local  0.078GB
> use aaa;
switched to db aaa
> show collections;
system.indexes
test
> db.createCollection("test2");
{ "note" : "from execCommand", "ok" : 0, "errmsg" : "not master" }

可见slave还是只读，不能写

3) 解决方法

将slave停机，变为master后启动

> use admin;
switched to db admin
> db.shutdownServer();
2015-09-23T16:05:28.355+0800 I NETWORK  DBClientCursor::init call() failed
server should be down...
2015-09-23T16:05:28.357+0800 I NETWORK  trying reconnect to 127.0.0.1:27017 (127.0.0.1) failed
2015-09-23T16:05:28.357+0800 W NETWORK  Failed to connect to 127.0.0.1:27017, reason: errno:111 Connection refused
2015-09-23T16:05:28.358+0800 I NETWORK  reconnect 127.0.0.1:27017 (127.0.0.1) failed failed couldn't connect to server 127.0.0.1:27017 (127.0.0.1), connection attempt  failed
>
bye
$ ps aux | grep mongo | grep -v grep
$ sudo vim /data/mongodb/conf/db0/mongodb.conf
master = true
#slave = true
#source = 172.30.2.201:27017
$ sudo /opt/mongodb/bin/mongod --config /data/mongodb/conf/db0/mongodb.conf
about to fork child process, waiting until server is ready for connections.
forked process: 36988
child process started successfully, parent exiting
$ mongo
MongoDB shell version: 3.0.6
connecting to: test
> show dbs;
aaa     0.078GB
local  50.054GB
> use aaa;
switched to db aaa
> db.createCollection("test2");
{ "ok" : 1 }
> show collections;
system.indexes
system.profile
test
test2
> db.isMaster();
{
"ismaster" : true,
"maxBsonObjectSize" : 16777216,
"maxMessageSizeBytes" : 48000000,
"maxWriteBatchSize" : 1000,
"localTime" : ISODate("2015-09-23T08:07:41.812Z"),
"maxWireVersion" : 3,
"minWireVersion" : 0,
"ok" : 1
}
>
bye

这样slave就变为新的master了

2.slave宕机
1) 制作新的slave

$ sudo rm -fr /data/mongodb/data/db0/*
$ sudo vim /data/mongodb/conf/db0/mongodb.conf
#master = true
slave = true
source = 172.30.2.202:27017
$ sudo /opt/mongodb/bin/mongod --config /data/mongodb/conf/db0/mongodb.conf
about to fork child process, waiting until server is ready for connections.
forked process: 48779
child process started successfully, parent exiting
$ mongo
MongoDB shell version: 3.0.6
connecting to: test
> db.isMaster();
{
"ismaster" : false,
"maxBsonObjectSize" : 16777216,
"maxMessageSizeBytes" : 48000000,
"maxWriteBatchSize" : 1000,
"localTime" : ISODate("2015-09-23T08:14:51.283Z"),
"maxWireVersion" : 3,
"minWireVersion" : 0,
"ok" : 1
}

2) 将新的slave停机

> use admin;
switched to db admin
> db.shutdownServer();
2015-09-23T16:15:31.329+0800 I NETWORK  DBClientCursor::init call() failed
server should be down...
2015-09-23T16:15:31.331+0800 I NETWORK  trying reconnect to 127.0.0.1:27017 (127.0.0.1) failed
2015-09-23T16:15:31.332+0800 I NETWORK  reconnect 127.0.0.1:27017 (127.0.0.1) ok
2015-09-23T16:15:31.333+0800 I NETWORK  Socket recv() errno:104 Connection reset by peer 127.0.0.1:27017
2015-09-23T16:15:31.333+0800 I NETWORK  SocketException: remote: 127.0.0.1:27017 error: 9001 socket exception [RECV_ERROR] server [127.0.0.1:27017]
2015-09-23T16:15:31.333+0800 I NETWORK  DBClientCursor::init call() failed
>
bye
$ ps aux | grep mongo | grep -v grep

3) 查看新master状况

$ mongo
MongoDB shell version: 3.0.6
connecting to: test
> show dbs;
aaa     0.078GB
local  50.054GB
> use aaa;
switched to db aaa
> db.createCollection("test3");
{ "ok" : 1 }
> show collections;
system.indexes
system.profile
test
test2
test3
>
bye

4) 结论

可见master是可以正常读写的，只需要再制作从库即可