mongodb replica set 多服务器 高可用 配置 详解

张映 发表于 2013-12-16

分类目录: nosql

标签:, , , ,

mongodb的多服务器配置,以前写过一篇文章,是master-slave模式的,请参考:mongodb 主从 配置 详解。master-slave模式,不能自动实现故障转移和恢复。所以推荐大家使用mongodb的replica set,来实现多服务器的高可用。给我的感觉是replica set好像自带了heartbeat功能,挺强大的。

一,三台服务器,1主,2从

服务器1:127.0.0.1:27017

服务器2:127.0.0.1:27018

服务器3:127.0.0.1:27019

1,创建数据库目录

[root@localhost ~]# mkdir /var/lib/{mongodb_2,mongodb_3}

在一台机子上面模拟,三台服务器,所以把DB目录分开了。

2,创建配置文件

[root@localhost ~]# cat /etc/mongodb.conf |awk '{if($0 !~ /^$/ && $0 !~ /^#/) {print $0}}'  //主服务器配置
port = 27017        //监听端口
fork = true         //后台运行
pidfilepath = /var/run/mongodb/mongodb.pid    //进程PID文件
logpath = /var/log/mongodb/mongodb.log        //日志文件
dbpath =/var/lib/mongodb           //db存放目录
journal = true                   //存储模式
nohttpinterface = true           //禁用http
directoryperdb=true              //一个数据库一个文件夹
logappend=true                  //追加方式写日志
replSet=repmore                 //集群名称,自定义
oplogSize=1000                  //oplog大小

[root@localhost ~]# cat /etc/mongodb_2.conf |awk '{if($0 !~ /^$/ && $0 !~ /^#/) {print $0}}'  //从服务器
port = 27018
fork = true
pidfilepath = /var/run/mongodb/mongodb_2.pid
logpath = /var/log/mongodb/mongodb_2.log
dbpath =/var/lib/mongodb_2
journal = true
nohttpinterface = true
directoryperdb=true
logappend=true
replSet=repmore
oplogSize=1000

[root@localhost ~]# cat /etc/mongodb_3.conf |awk '{if($0 !~ /^$/ && $0 !~ /^#/) {print $0}}'  //从服务器
port = 27019
fork = true
pidfilepath = /var/run/mongodb/mongodb_3.pid
logpath = /var/log/mongodb/mongodb_3.log
dbpath =/var/lib/mongodb_3
journal = true
nohttpinterface = true
oplogSize = 1000
directoryperdb=true
logappend=true
replSet=repmore

在这里要注意一点,不要把认证开起来了,不然查看rs.status();时,主从服务器间,无法连接,"lastHeartbeatMessage" : "initial sync couldn't connect to 127.0.0.1:27017"

3,启动三台服务器

mongod -f /etc/mongodb.conf
mongod -f /etc/mongodb_2.conf
mongod -f /etc/mongodb_3.conf

注意:初次启动时,主服务器比较快的,从服务器有点慢。

二,配置并初始化replica set

1,配置replica set节点

> config = {_id:"repmore",members:[{_id:0,host:'127.0.0.1:27017',priority :2},{_id:1,host:'127.0.0.1:27018',priority:1},{_id:2,host:'127.0.0.1:27019',priority:1}]}

2,初始化replica set

> rs.initiate(config);
{
    "info" : "Config now saved locally.  Should come online in about a minute.",
    "ok" : 1
}

3,查看replica set各节点状态

repmore:PRIMARY> rs.status();
{
    "set" : "repmore",
    "date" : ISODate("2013-12-16T21:01:51Z"),
    "myState" : 2,
    "syncingTo" : "127.0.0.1:27017",
    "members" : [
        {
            "_id" : 0,
            "name" : "127.0.0.1:27017",
            "health" : 1,
            "state" : 1,
            "stateStr" : "PRIMARY",
            "uptime" : 33,
            "optime" : Timestamp(1387227638, 1),
            "optimeDate" : ISODate("2013-12-16T21:00:38Z"),
            "lastHeartbeat" : ISODate("2013-12-16T21:01:50Z"),
            "lastHeartbeatRecv" : ISODate("2013-12-16T21:01:50Z"),
            "pingMs" : 0,
            "syncingTo" : "127.0.0.1:27018"
        },
        {
            "_id" : 1,
            "name" : "127.0.0.1:27018",
            "health" : 1,
            "state" : 2,
            "stateStr" : "SECONDARY",
            "uptime" : 1808,
            "optime" : Timestamp(1387227638, 1),
            "optimeDate" : ISODate("2013-12-16T21:00:38Z"),
            "errmsg" : "syncing to: 127.0.0.1:27017",
            "self" : true
        },
        {
            "_id" : 2,
            "name" : "127.0.0.1:27019",
            "health" : 1,
            "state" : 2,
            "stateStr" : "SECONDARY",
            "uptime" : 1806,
            "optime" : Timestamp(1387227638, 1),
            "optimeDate" : ISODate("2013-12-16T21:00:38Z"),
            "lastHeartbeat" : ISODate("2013-12-16T21:01:50Z"),
            "lastHeartbeatRecv" : ISODate("2013-12-16T21:01:51Z"),
            "pingMs" : 0,
            "lastHeartbeatMessage" : "syncing to: 127.0.0.1:27018",
            "syncingTo" : "127.0.0.1:27018"
        }
    ],
    "ok" : 1
}

在这里要注意,rs.initiate初始化也是要一定时间的,刚执行完rs.initiate,我就查看状态,从服务器的stateStr不是SECONDARY,而是stateStr" : "STARTUP2",等一会就好了。

三,replica set主,从测试

1,主服务器测试

repmore:PRIMARY> show dbs;
local    1.078125GB
repmore:PRIMARY> use test
switched to db test
repmore:PRIMARY> db.test.insert({'name':'tank','phone':'12345678'});
repmore:PRIMARY> db.test.find();
{ "_id" : ObjectId("52af64549d2f9e75bc57cda7"), "name" : "tank", "phone" : "12345678" }

2,从服务器测试

[root@localhost mongodb]# mongo 127.0.0.1:27018   //连接
MongoDB shell version: 2.4.6
connecting to: 127.0.0.1:27018/test
repmore:SECONDARY> show dbs;
local    1.078125GB
test    0.203125GB
repmore:SECONDARY> db.test.find();     //无权限查看
error: { "$err" : "not master and slaveOk=false", "code" : 13435 }
repmore:SECONDARY> rs.slaveOk();       //从库开启
repmore:SECONDARY> db.test.find();     //从库可看到主库刚插入的数据
{ "_id" : ObjectId("52af64549d2f9e75bc57cda7"), "name" : "tank", "phone" : "12345678" }
repmore:SECONDARY> db.test.insert({'name':'zhangying','phone':'12345678'});   //从库只读,无插入权限
not master

到这儿,我们的replica set就配置好了。

四,故障测试

前面我说过,mongodb replica set有故障转移功能,下面就模拟一下,这个过程

1,故障转移

1.1,关闭主服务器

[root@localhost mongodb]# ps aux |grep mongod    //查看所有的mongod
root     16977  0.2  1.1 3153692 44464 ?       Sl   04:31   0:02 mongod -f /etc/mongodb.conf
root     17032  0.2  1.1 3128996 43640 ?       Sl   04:31   0:02 mongod -f /etc/mongodb_2.conf
root     17092  0.2  0.9 3127976 38324 ?       Sl   04:31   0:02 mongod -f /etc/mongodb_3.conf
root     20400  0.0  0.0 103248   860 pts/2    S+   04:47   0:00 grep mongod
[root@localhost mongodb]# kill 16977  //关闭主服务器进程
[root@localhost mongodb]# ps aux |grep mongod
root     17032  0.2  1.1 3133124 43836 ?       Sl   04:31   0:02 mongod -f /etc/mongodb_2.conf
root     17092  0.2  0.9 3127976 38404 ?       Sl   04:31   0:02 mongod -f /etc/mongodb_3.conf
root     20488  0.0  0.0 103248   860 pts/2    S+   04:47   0:00 grep mongod

1.2,在主库执行命令

repmore:PRIMARY> show dbs;
Tue Dec 17 04:48:02.392 DBClientCursor::init call() failed

1.3,从库查看状态,如下图,

replica set 故障测试

replica set 故障测试

以前的从库变主库了,故障转移成功

2,故障恢复

mongod -f /etc/mongodb.conf

启动刚被关闭的主服务器,然后在登录到主服务器,查看状态rs.status();已恢复到最原始的状态了。



转载请注明
作者:海底苍鹰
地址:http://blog.51yip.com/nosql/1580.html