mongo 集群配置

master-slave

主的配置文件 /Users/olifer/middle/mongo/master-slave/master/mongod.conf

bind_ip = 127.0.0.1port = 27017dbpath = /Users/olifer/middle/mongo/master-slave/master/data/master = true

从的配置文件

bind_ip = 127.0.0.1port = 27018dbpath = /Users/olifer/middle/mongo/master-slave/slave/data/slave = truesource = 127.0.0.1:27017

启动主服务器,并且看到下面的日志，说明配置成功。

mongod -f ~/middle/mongo/master-slave/master/mongod.conf2017-12-05T15:35:22.479+0800 I JOURNAL  [initandlisten] journal dir=/Users/olifer/middle/mongo/master-slave/master/data/journal2017-12-05T15:35:22.479+0800 I JOURNAL  [initandlisten] recover : no journal files present, no recovery needed2017-12-05T15:35:22.500+0800 I JOURNAL  [durability] Durability thread started2017-12-05T15:35:22.500+0800 I JOURNAL  [journal writer] Journal writer thread started2017-12-05T15:35:22.501+0800 I CONTROL  [initandlisten] MongoDB starting : pid=6368 port=27017 dbpath=/Users/olifer/middle/mongo/master-slave/master/data/ master=1 64-bit host=oliferdeMacBook-Pro.local2017-12-05T15:35:22.501+0800 I CONTROL  [initandlisten] db version v3.0.72017-12-05T15:35:22.501+0800 I CONTROL  [initandlisten] git version: nogitversion2017-12-05T15:35:22.501+0800 I CONTROL  [initandlisten] build info: Darwin yosemitevm.local 14.5.0 Darwin Kernel Version 14.5.0: Wed Jul 29 02:26:53 PDT 2015; root:xnu-2782.40.9~1/RELEASE_X86_64 x86_64 BOOST_LIB_VERSION=1_492017-12-05T15:35:22.501+0800 I CONTROL  [initandlisten] allocator: system2017-12-05T15:35:22.501+0800 I CONTROL  [initandlisten] options: { config: "/Users/olifer/middle/mongo/master-slave/master/mongod.conf", master: true, net: { bindIp: "127.0.0.1", port: 27017 }, storage: { dbPath: "/Users/olifer/middle/mongo/master-slave/master/data/" } }2017-12-05T15:35:22.505+0800 I INDEX    [initandlisten] allocating new ns file /Users/olifer/middle/mongo/master-slave/master/data/local.ns, filling with zeroes...2017-12-05T15:35:22.570+0800 I STORAGE  [FileAllocator] allocating new datafile /Users/olifer/middle/mongo/master-slave/master/data/local.0, filling with zeroes...2017-12-05T15:35:22.570+0800 I STORAGE  [FileAllocator] creating directory /Users/olifer/middle/mongo/master-slave/master/data/_tmp2017-12-05T15:35:22.786+0800 I STORAGE  [FileAllocator] done allocating datafile /Users/olifer/middle/mongo/master-slave/master/data/local.0, size: 64MB,  took 0.215 secs2017-12-05T15:35:22.972+0800 I REPL     [initandlisten] ******2017-12-05T15:35:22.972+0800 I REPL     [initandlisten] creating replication oplog of size: 192MB...2017-12-05T15:35:22.972+0800 I STORAGE  [FileAllocator] allocating new datafile /Users/olifer/middle/mongo/master-slave/master/data/local.1, filling with zeroes...2017-12-05T15:35:24.061+0800 I STORAGE  [FileAllocator] done allocating datafile /Users/olifer/middle/mongo/master-slave/master/data/local.1, size: 256MB,  took 1.088 secs2017-12-05T15:35:24.117+0800 I REPL     [initandlisten] ******2017-12-05T15:35:24.119+0800 I NETWORK  [initandlisten] waiting for connections on port 27017

启动从服务器,并且看到下面的日志，说明配置成功。

mongod -f ~/middle/mongo/master-slave/slave/mongod.conf2017-12-05T15:37:11.518+0800 I JOURNAL  [initandlisten] journal dir=/Users/olifer/middle/mongo/master-slave/slave/data/journal2017-12-05T15:37:11.518+0800 I JOURNAL  [initandlisten] recover : no journal files present, no recovery needed2017-12-05T15:37:11.535+0800 I JOURNAL  [durability] Durability thread started2017-12-05T15:37:11.535+0800 I JOURNAL  [journal writer] Journal writer thread started2017-12-05T15:37:11.535+0800 I CONTROL  [initandlisten] MongoDB starting : pid=7304 port=27018 dbpath=/Users/olifer/middle/mongo/master-slave/slave/data/ slave=1 64-bit host=oliferdeMacBook-Pro.local2017-12-05T15:37:11.535+0800 I CONTROL  [initandlisten] db version v3.0.72017-12-05T15:37:11.535+0800 I CONTROL  [initandlisten] git version: nogitversion2017-12-05T15:37:11.535+0800 I CONTROL  [initandlisten] build info: Darwin yosemitevm.local 14.5.0 Darwin Kernel Version 14.5.0: Wed Jul 29 02:26:53 PDT 2015; root:xnu-2782.40.9~1/RELEASE_X86_64 x86_64 BOOST_LIB_VERSION=1_492017-12-05T15:37:11.535+0800 I CONTROL  [initandlisten] allocator: system2017-12-05T15:37:11.535+0800 I CONTROL  [initandlisten] options: { config: "/Users/olifer/middle/mongo/master-slave/slave/mongod.conf", net: { bindIp: "127.0.0.1", port: 27018 }, slave: true, source: "127.0.0.1:27017", storage: { dbPath: "/Users/olifer/middle/mongo/master-slave/slave/data/" } }2017-12-05T15:37:11.536+0800 I INDEX    [initandlisten] allocating new ns file /Users/olifer/middle/mongo/master-slave/slave/data/local.ns, filling with zeroes...2017-12-05T15:37:11.605+0800 I STORAGE  [FileAllocator] allocating new datafile /Users/olifer/middle/mongo/master-slave/slave/data/local.0, filling with zeroes...2017-12-05T15:37:11.605+0800 I STORAGE  [FileAllocator] creating directory /Users/olifer/middle/mongo/master-slave/slave/data/_tmp2017-12-05T15:37:11.824+0800 I STORAGE  [FileAllocator] done allocating datafile /Users/olifer/middle/mongo/master-slave/slave/data/local.0, size: 64MB,  took 0.218 secs2017-12-05T15:37:11.890+0800 I NETWORK  [initandlisten] waiting for connections on port 270182017-12-05T15:37:12.894+0800 I REPL     [replslave] repl: syncing from host:127.0.0.1:270172017-12-05T15:37:17.930+0800 I REPL     [replslave] repl: sleep 2 sec before next pass2017-12-05T15:37:19.935+0800 I REPL     [replslave] repl: syncing from host:127.0.0.1:270172017-12-05T15:37:24.958+0800 I REPL     [replslave] repl: syncing from host:127.0.0.1:270172017-12-05T15:37:33.188+0800 I REPL     [replslave] repl: syncing from host:127.0.0.1:270172017-12-05T15:37:43.193+0800 I REPL     [replslave] repl: syncing from host:127.0.0.1:270172017-12-05T15:37:53.198+0800 I REPL     [replslave] repl: syncing from host:127.0.0.1:27017

我们可以通过slave的日志中可以看到，slave一直同master保持同步数据的联系。

我们通过客户端链接上服务端

mongo 127.0.0.1:27017  #主mongo 127.0.0.1:27018  #从

在主上操作

> db.test.insert({"name":"linyang"})
WriteResult({ "nInserted" : 1 })
> db.test.find({});
{ "_id" : ObjectId("5a264e0ecb6f7d3713c516a7"), "name" : "linyang" }

往里面插入了一条记录，可以再从的日志中发现同步的记录

2017-12-05T15:43:10.387+0800 I INDEX    [replslave] allocating new ns file /Users/olifer/middle/mongo/master-slave/slave/data/test.ns, filling with zeroes...2017-12-05T15:43:10.450+0800 I STORAGE  [FileAllocator] allocating new datafile /Users/olifer/middle/mongo/master-slave/slave/data/test.0, filling with zeroes...2017-12-05T15:43:10.646+0800 I STORAGE  [FileAllocator] done allocating datafile /Users/olifer/middle/mongo/master-slave/slave/data/test.0, size: 64MB,  took 0.196 secs2017-12-05T15:43:10.702+0800 I REPL     [replslave] resync: dropping database test2017-12-05T15:43:10.709+0800 I JOURNAL  [replslave] journalCleanup...2017-12-05T15:43:10.710+0800 I JOURNAL  [replslave] removeJournalFiles2017-12-05T15:43:10.713+0800 I JOURNAL  [replslave] journalCleanup...2017-12-05T15:43:10.713+0800 I JOURNAL  [replslave] removeJournalFiles2017-12-05T15:43:10.715+0800 I REPL     [replslave] resync: cloning database test to get an initial copy2017-12-05T15:43:10.718+0800 I INDEX    [replslave] allocating new ns file /Users/olifer/middle/mongo/master-slave/slave/data/test.ns, filling with zeroes...2017-12-05T15:43:10.803+0800 I STORAGE  [FileAllocator] allocating new datafile /Users/olifer/middle/mongo/master-slave/slave/data/test.0, filling with zeroes...2017-12-05T15:43:11.056+0800 I STORAGE  [FileAllocator] done allocating datafile /Users/olifer/middle/mongo/master-slave/slave/data/test.0, size: 64MB,  took 0.252 secs2017-12-05T15:43:11.107+0800 I INDEX    [replslave] build index on: test.test properties: { v: 1, key: { _id: 1 }, name: "_id_", ns: "test.test" }2017-12-05T15:43:11.107+0800 I INDEX    [replslave]      building index using bulk method2017-12-05T15:43:11.107+0800 I INDEX    [replslave] build index done.  scanned 1 total records. 0 secs2017-12-05T15:43:11.107+0800 I STORAGE  [replslave] copying indexes for: { name: "test", options: {} }2017-12-05T15:43:11.108+0800 I REPL     [replslave] resync: done with initial clone for db: test

我们再从的client中查看有没有信息。

mongo 127.0.0.1:27018
MongoDB shell version: 3.0.7
connecting to: 127.0.0.1:27018/test> db.test.find({});
{ "_id" : ObjectId("5a264e0ecb6f7d3713c516a7"), "name" : "linyang" }
>

也确认同步过来了。然后再从上插入记录

> db.test.insert({"age":34});
WriteResult({ "writeError" : { "code" : undefined, "errmsg" : "not master" } })
>

从上不能插入记录。如果主挂掉了，从又不能写数据，那么是否mongo集群就挂掉了。答案当然是否定的，因为，mongo还有副本集。

副本集

我们使用三台mongo来模拟.
第一台的配置文件 /Users/olifer/middle/mongo/replica/a/mongod.conf

dbpath = /Users/olifer/middle/mongo/replica/a/data
port = 8001bind_ip = 127.0.0.1replSet = child/127.0.0.1:8002

第二台的配置文件 /Users/olifer/middle/mongo/replica/b/mongod.conf

dbpath = /Users/olifer/middle/mongo/replica/a/data
port = 8002bind_ip = 127.0.0.1replSet = child/127.0.0.1:8003

第三台的配置文件 /Users/olifer/middle/mongo/replica/c/mongod.conf

dbpath = /Users/olifer/middle/mongo/replica/a/data
port = 8003bind_ip = 127.0.0.1replSet = child/127.0.0.1:8001

启动服务器

mongod -f ~/middle/mongo/replica/a/mongod.conf
mongod -f ~/middle/mongo/replica/b/mongod.conf
mongod -f ~/middle/mongo/replica/c/mongod.conf

启动成功后，进入三个服务的任何一个客户端

mongo 127.0.0.1:8002

> config = {_id: 'child', members: [
  {    "_id":1,    "host":"127.0.0.1:8001"
  },
  {     "_id":2,     "host":"127.0.0.1:8002"
  },
  {     "_id":3,     "host":"127.0.0.1:8003"
   }
   ]
 }
{    "_id" : "child",    "members" : [
        {            "_id" : 1,            "host" : "127.0.0.1:8001"
        },
        {            "_id" : 2,            "host" : "127.0.0.1:8002"
        },
        {            "_id" : 3,            "host" : "127.0.0.1:8003"
        }
    ]
}
> rs.initiate(config);
{ "ok" : 1 }
child:SECONDARY>

配置完后发现前缀发生了改变，我们进入另外两个渠道的客户端

mongo 127.0.0.1:8001
MongoDB shell version: 3.0.7
connecting to: 127.0.0.1:8001/testchild:PRIMARY> 

mongo 127.0.0.1:8003
MongoDB shell version: 3.0.7
connecting to: 127.0.0.1:8003/testchild:SECONDARY>

其中child:PRIMARY>表示活跃节点。其余为备份节点。注意:只有活跃节点才能进行查询数据库的信息操作，备份节点不能进行会报错,在任意客户端执行rs.status() 来查看所有状态

child:PRIMARY> rs.status()
{    "set" : "child",    "date" : ISODate("2017-12-05T08:56:44.523Z"),    "myState" : 1,    "members" : [
        {            "_id" : 1,            "name" : "127.0.0.1:8001",            "health" : 1,            "state" : 1,            "stateStr" : "PRIMARY",            "uptime" : 620,            "optime" : Timestamp(1512463818, 1),            "optimeDate" : ISODate("2017-12-05T08:50:18Z"),            "electionTime" : Timestamp(1512463821, 1),            "electionDate" : ISODate("2017-12-05T08:50:21Z"),            "configVersion" : 1,            "self" : true
        },
        {            "_id" : 2,            "name" : "127.0.0.1:8002",            "health" : 1,            "state" : 2,            "stateStr" : "SECONDARY",            "uptime" : 387,            "optime" : Timestamp(1512463818, 1),            "optimeDate" : ISODate("2017-12-05T08:50:18Z"),            "lastHeartbeat" : ISODate("2017-12-05T08:56:43.756Z"),            "lastHeartbeatRecv" : ISODate("2017-12-05T08:56:43.756Z"),            "pingMs" : 0,            "configVersion" : 1
        },
        {            "_id" : 3,            "name" : "127.0.0.1:8003",            "health" : 1,            "state" : 2,            "stateStr" : "SECONDARY",            "uptime" : 387,            "optime" : Timestamp(1512463818, 1),            "optimeDate" : ISODate("2017-12-05T08:50:18Z"),            "lastHeartbeat" : ISODate("2017-12-05T08:56:43.756Z"),            "lastHeartbeatRecv" : ISODate("2017-12-05T08:56:43.756Z"),            "pingMs" : 0,            "configVersion" : 1
        }
    ],    "ok" : 1
}

搭建成功后，我们来验证一下。

在活跃的节点创建数据,在备份库拉取数据

child:PRIMARY> db.repl.insert({"name":"123"});
WriteResult({ "nInserted" : 1 })

child:SECONDARY> db.repl.find();
{ "_id" : ObjectId("5a26602f5cefb1fdb377843b"), "name" : "123" }

功能正常。

从服务器不能写

child:SECONDARY> db.repl.insert({"age":444});
WriteResult({ "writeError" : { "code" : undefined, "errmsg" : "not master" } })

关闭活跃的节点，从服务器会通过选举得到最新的活跃的节点

当我关闭原来活跃的服务器8001后，通过内部选举，8002成了最新的活跃节点没从前缀也可以看出

child:SECONDARY>child:PRIMARY>

现在看一下最新的状态

child:PRIMARY> rs.status();
{    "set" : "child",    "date" : ISODate("2017-12-05T09:09:31.666Z"),    "myState" : 1,    "members" : [
        {            "_id" : 1,            "name" : "127.0.0.1:8001",            "health" : 0,            "state" : 8,            "stateStr" : "(not reachable/healthy)",            "uptime" : 0,            "optime" : Timestamp(0, 0),            "optimeDate" : ISODate("1970-01-01T00:00:00Z"),            "lastHeartbeat" : ISODate("2017-12-05T09:09:31.193Z"),            "lastHeartbeatRecv" : ISODate("2017-12-05T09:07:12.773Z"),            "pingMs" : 0,            "lastHeartbeatMessage" : "Failed attempt to connect to 127.0.0.1:8001; couldn't connect to server 127.0.0.1:8001 (127.0.0.1), connection attempt failed",            "configVersion" : -1
        },
        {            "_id" : 2,            "name" : "127.0.0.1:8002",            "health" : 1,            "state" : 1,            "stateStr" : "PRIMARY",            "uptime" : 1308,            "optime" : Timestamp(1512464432, 2),            "optimeDate" : ISODate("2017-12-05T09:00:32Z"),            "electionTime" : Timestamp(1512464835, 1),            "electionDate" : ISODate("2017-12-05T09:07:15Z"),            "configVersion" : 1,            "self" : true
        },
        {            "_id" : 3,            "name" : "127.0.0.1:8003",            "health" : 1,            "state" : 2,            "stateStr" : "SECONDARY",            "uptime" : 1152,            "optime" : Timestamp(1512464432, 2),            "optimeDate" : ISODate("2017-12-05T09:00:32Z"),            "lastHeartbeat" : ISODate("2017-12-05T09:09:30.963Z"),            "lastHeartbeatRecv" : ISODate("2017-12-05T09:09:30.963Z"),            "pingMs" : 0,            "configVersion" : 1
        }
    ],    "ok" : 1
}

说明内部的切换确实成功了。还剩下一个分片，我们下次再讲。

作者：数齐
链接：https://www.jianshu.com/p/1f5dd7492228