##集群规划

IP

端口

角色

10.1.200.118

27017

mongos

10.1.200.120

27017

config server

10.1.200.125

27017

shard

10.1.200.143

27017

shard（primary）

10.1.200.146

27017

shard

sharding集群具体流程如下图：

MongoDB集群 主从节点切换 mongodb集群角色_开发工具

分析：

1、sharding集群的三类角色：mongos、config server、shard： mongos：类似于负载均衡器的作用，可以部署多个。 config server：配置服务器用来存储mongos的配置信息，可以部署成一个replica set集群。 shard：片键，可以部署成一个replica set集群。

2、replica set集群是使用集群中的每一台服务器来承载所有对整个集群的流量，容易造成服务器压力过大。而sharding集群是使用负载均衡的思想，将对整个集群的流量分摊到每一台服务器上，这样可以大大降低整个集群的负荷，提高系统的吞吐量。

3、分片的方式有三种：range，hash，tag： 1）range：就是下面要讲述的。 2）hash：和redis cluster一样。 3）tag：手工指定分片。

##安装和配置

参照单实例。

##各个节点配置文件修改

1、10.1.200.118节点：

[root@centos4 logs]# vim /etc/mongod.conf 
# 数据库文件位置
dbpath=/usr/local/mongodb/db
# 日志文件位置
logpath=/usr/local/mongodb/logs/mongodb.log
# 以追加方式写入日志
logappend=true
# 绑定服务IP，若绑定127.0.0.1，则只能本机访问，不指定默认本地所有IP  
bind_ip=10.1.200.118
# 默认端口为27017
port=27017

2、10.1.200.120节点：

[root@centos5 db]# vim /etc/mongod.conf 
# 数据库文件位置
dbpath=/usr/local/mongodb/db
# 日志文件位置
logpath=/usr/local/mongodb/logs/mongodb.log
# 以追加方式写入日志
logappend=true
# 绑定服务IP，若绑定127.0.0.1，则只能本机访问，不指定默认本地所有IP
bind_ip=10.1.200.120
# 默认端口为27017
port=27017

# sharding集群中，设置是否为config server
configsvr=true

3、10.1.200.125、10.1.200.143、10.1.200.146三个节点：

[root@centos6 db]# vim /etc/mongod.conf 
# 数据库文件位置
dbpath=/usr/local/mongodb/db
# 日志文件位置
logpath=/usr/local/mongodb/logs/mongodb.log
# 以追加方式写入日志
logappend=true
# 绑定服务IP，若绑定127.0.0.1，则只能本机访问，不指定默认本地所有IP
bind_ip=10.1.200.125
# 默认端口为27017
port=27017

##启动10.1.200.120节点的mongodb服务

[root@centos5 db]# service mongodb start
db path is: /usr/local/mongodb/db
/usr/local/mongodb/bin/mongod
Starting mongod: /usr/local/mongodb/bin/mongod --config /etc/mongod.conf

[root@centos5 db]#

##启动10.1.200.118节点的mongos服务

[root@centos4 logs]# mongos --configdb 10.1.200.120:27017 
2016-08-13T15:29:41.652+0800 W SHARDING [main] Running a sharded cluster with fewer than 3 config servers should only be done for testing purposes and is not recommended for production.
2016-08-13T15:29:41.655+0800 I CONTROL  [main] ** WARNING: You are running this process as the root user, which is not recommended.
2016-08-13T15:29:41.655+0800 I CONTROL  [main] 
2016-08-13T15:29:41.655+0800 I SHARDING [mongosMain] MongoS version 3.2.8 starting: pid=5090 port=27017 64-bit host=centos4 (--help for usage)
2016-08-13T15:29:41.655+0800 I CONTROL  [mongosMain] db version v3.2.8
2016-08-13T15:29:41.655+0800 I CONTROL  [mongosMain] git version: ed70e33130c977bda0024c125b56d159573dbaf0
2016-08-13T15:29:41.655+0800 I CONTROL  [mongosMain] allocator: tcmalloc
2016-08-13T15:29:41.655+0800 I CONTROL  [mongosMain] modules: none
2016-08-13T15:29:41.655+0800 I CONTROL  [mongosMain] build environment:
2016-08-13T15:29:41.655+0800 I CONTROL  [mongosMain]     distarch: x86_64
2016-08-13T15:29:41.655+0800 I CONTROL  [mongosMain]     target_arch: x86_64
2016-08-13T15:29:41.655+0800 I CONTROL  [mongosMain] options: { sharding: { configDB: "10.1.200.120:27017" } }
2016-08-13T15:29:41.655+0800 I SHARDING [mongosMain] Updating config server connection string to: 10.1.200.120:27017
2016-08-13T15:29:41.659+0800 I SHARDING [LockPinger] creating distributed lock ping thread for 10.1.200.120:27017 and process centos4:27017:1471073381:1047189914 (sleeping for 30000ms)
2016-08-13T15:29:41.662+0800 I SHARDING [LockPinger] cluster 10.1.200.120:27017 pinged successfully at 2016-08-13T15:29:41.660+0800 by distributed lock pinger '10.1.200.120:27017/centos4:27017:1471073381:1047189914', sleeping for 30000ms
2016-08-13T15:29:41.670+0800 I NETWORK  [HostnameCanonicalizationWorker] Starting hostname canonicalization worker
2016-08-13T15:29:41.670+0800 I SHARDING [Balancer] about to contact config servers and shards
2016-08-13T15:29:41.671+0800 I SHARDING [Balancer] config servers and shards contacted successfully
2016-08-13T15:29:41.671+0800 I SHARDING [Balancer] balancer id: centos4:27017 started
2016-08-13T15:29:41.689+0800 I NETWORK  [mongosMain] waiting for connections on port 27017
2016-08-13T15:29:41.693+0800 I SHARDING [Balancer] distributed lock 'balancer/centos4:27017:1471073381:1047189914' acquired for 'doing balance round', ts : 57aecc6597c942c8b7784c79
2016-08-13T15:29:41.694+0800 I SHARDING [Balancer] distributed lock 'balancer/centos4:27017:1471073381:1047189914' unlocked. 
2016-08-13T15:29:41.719+0800 I NETWORK  [mongosMain] connection accepted from 10.6.101.46:26223 #1 (1 connection now open)

注意：此时10.1.200.118节点的角色是sharding集群的mongos，所以需要开启对应的mongos服务，而不是mongod服务。

##启动另外三个节点的mongod服务：

[root@centos6 db]# service mongodb start
db path is: /usr/local/mongodb/db
/usr/local/mongodb/bin/mongod
Starting mongod: /usr/local/mongodb/bin/mongod --config /etc/mongod.conf

[root@centos6 db]#

##集群配置 1、客户端连接10.1.200.118节点，如下操作添加10.1.200.125、10.1.200.143和10.1.200.146三个节点到集群中：

sh.addShard("10.1.200.125:27017")
sh.addShard("10.1.200.143:27017")
sh.addShard("10.1.200.146:27017")
/* 1 */
{
    "shardAdded" : "shard0000",
    "ok" : 1.0
}
/* 1 */
{
    "shardAdded" : "shard0001",
    "ok" : 1.0
}
/* 1 */
{
    "shardAdded" : "shard0002",
    "ok" : 1.0
}

2、配置在具体某个数据库上启用分片功能：

sh.enableSharding("flightswitch")
/* 1 */
{
    "ok" : 1.0
}

如上配置之后在flightswitch数据库上启用分片功能。

3、在某个数据库的某个collection上启用分片功能，片键是主键id：

sh.shardCollection("flightswitch.flightsegment",{"_id":1})
/* 1 */
{
    "collectionsharded" : "flightswitch.flightsegment",
    "ok" : 1.0
}

注意：我们最后的分片是对某一个collection的数据进行分片的。

4、查看sharding集群的状态：

sh.status()
--- Sharding Status --- 
  sharding version: {
	"_id" : 1,
	"minCompatibleVersion" : 5,
	"currentVersion" : 6,
	"clusterId" : ObjectId("57aec8ba87cc3e8611b4713e")
}
  shards:
	{  "_id" : "shard0000",  "host" : "10.1.200.125:27017" }
	{  "_id" : "shard0001",  "host" : "10.1.200.143:27017" }
	{  "_id" : "shard0002",  "host" : "10.1.200.146:27017" }
  active mongoses:
	"3.2.8" : 1
  balancer:
	Currently enabled:  yes
	Currently running:  no
	Failed balancer rounds in last 5 attempts:  0
	Migration Results for the last 24 hours: 
		No recent migrations
  databases:
	{  "_id" : "flightswitch",  "primary" : "shard0001",  "partitioned" : true }
		flightswitch.flightsegment
			shard key: { "_id" : 1 }
			unique: false
			balancing: true
			chunks:
				shard0001	1
			{ "_id" : { "$minKey" : 1 } } -->> { "_id" : { "$maxKey" : 1 } } on : shard0001 Timestamp(1, 0)

从集群状态可以看出来此时集群的主片键是shard0001，也就是10.1.200.143节点。 至此sharding集群已经创建完成。

##集群测试

1、客户端连接10.1.200.118节点，向之前配置分片功能的collection中插入document数据：

for(var i=1; i<=100; i++ ){
    db.flightsegment.insert({"originCity":"SHA" + i})
}
Inserted 1 record(s) in 55ms
Inserted 1 record(s) in 10ms
Inserted 1 record(s) in 11ms
Inserted 1 record(s) in 9ms
Inserted 1 record(s) in 9ms
Inserted 1 record(s) in 7ms
Inserted 1 record(s) in 7ms
Inserted 1 record(s) in 8ms
Inserted 1 record(s) in 7ms
Inserted 1 record(s) in 7ms
Inserted 1 record(s) in 9ms
Inserted 1 record(s) in 7ms
Inserted 1 record(s) in 6ms
Inserted 1 record(s) in 7ms
Inserted 1 record(s) in 7ms
Inserted 1 record(s) in 10ms
Inserted 1 record(s) in 7ms
Inserted 1 record(s) in 12ms
……

2、此时再查看集群的状态：

sh.status()
--- Sharding Status --- 
  sharding version: {
	"_id" : 1,
	"minCompatibleVersion" : 5,
	"currentVersion" : 6,
	"clusterId" : ObjectId("57aec8ba87cc3e8611b4713e")
}
  shards:
	{  "_id" : "shard0000",  "host" : "10.1.200.125:27017" }
	{  "_id" : "shard0001",  "host" : "10.1.200.143:27017" }
	{  "_id" : "shard0002",  "host" : "10.1.200.146:27017" }
  active mongoses:
	"3.2.8" : 1
  balancer:
	Currently enabled:  yes
	Currently running:  no
	Failed balancer rounds in last 5 attempts:  0
	Migration Results for the last 24 hours: 
		2 : Success
  databases:
	{  "_id" : "flightswitch",  "primary" : "shard0001",  "partitioned" : true }
		flightswitch.flightsegment
			shard key: { "_id" : 1 }
			unique: false
			balancing: true
			chunks:
				shard0000	1
				shard0001	1
				shard0002	1
			{ "_id" : { "$minKey" : 1 } } -->> { "_id" : ObjectId("57aed47d2d291818f5dc78fb") } on : shard0000 Timestamp(2, 0) 
			{ "_id" : ObjectId("57aed47d2d291818f5dc78fb") } -->> { "_id" : ObjectId("57aed47d2d291818f5dc7911") } on : shard0002 Timestamp(3, 0) 
			{ "_id" : ObjectId("57aed47d2d291818f5dc7911") } -->> { "_id" : { "$maxKey" : 1 } } on : shard0001 Timestamp(3, 1)

可以看到三个节点上都有数据了，说明数据已经被分片存储了。