MongoDB数据库中如何配置autoshard

mongodb支持自动分片功能，这无疑给DBA和开发人员带来很大便利，我们无需要手动按一定规则进行拆分DB，当数据量增大后，我们只需要扩充设备即可。
mongodb shard配置文档如下：

auo shard 试验设备列表
主机列表
主机名 ip 用途 监听端口 web访问url
localhost 192.168.100.245 config server 27019 http://192.168.100.245:28019/
TEST40 192.168.100.212 shard 1 27018 http://192.168.100.212:28018/
chn-xg-5-5c1 192.168.100.210 shard 2 27018 http://192.168.100.210:28018/
band-all 192.168.100.229 mongos process/mongos Router 27017 http://192.168.100.229:27017/
分析发现config server、shard、mongos的监听端口和web端口均是第二位的差异。监听端口第二位是7，web端口是8，其它全部相同。
mongod --fork --logpath /var/log/mongodb.log --logappend
mogod默认启动为前台进程，当连接session断开，mongod程序自动关闭

一、配置shard 各个模块

1、启动shard1
[root@TEST40 ~]# mongod --shardsvr --fork --logpath /var/log/mongodb.log --logappend
[1] 28521
[root@TEST40 ~]# Tue Aug 17 03:01:57 MongoDB starting : pid=28521 port=27018 dbpath=/data/db/ 32-bit 
** NOTE: when using MongoDB 32 bit, you are limited to about 2 gigabytes of data
**       see http://blog.mongodb.org/post/137788967/32-bit-limitations
Tue Aug 17 03:01:57 db version v1.6.0, pdfile version 4.5
Tue Aug 17 03:01:57 git version: 2c7f164b653f0d703947572ede064aed41cc2185
Tue Aug 17 03:01:57 sys info: Linux domU-12-31-39-01-70-B4 2.6.21.7-2.fc8xen #1 SMP Fri Feb 15 12:39:36 EST 2008 i686 BOOST_LIB_VERSION=1_37
Tue Aug 17 03:01:57 [initandlisten] waiting for connections on port 27018
Tue Aug 17 03:01:57 [websvr] web admin interface listening on port 28018

2、启动shard2，shard2的dbpath与默认路径不同，因此需要制定dbpath
[root@chn-xg-5-5c1 cache2]# mongod --shardsvr --dbpath /data/cache2/db --fork --logpath /var/log/mongodb.log --logappend
[1] 31264
[root@chn-xg-5-5c1 cache2]# Tue Aug 17 03:11:48 MongoDB starting : pid=31264 port=27018 dbpath=/data/cache2/db 32-bit 
** NOTE: when using MongoDB 32 bit, you are limited to about 2 gigabytes of data
**       see http://blog.mongodb.org/post/137788967/32-bit-limitations
Tue Aug 17 03:11:48 db version v1.6.0, pdfile version 4.5
Tue Aug 17 03:11:48 git version: 2c7f164b653f0d703947572ede064aed41cc2185
Tue Aug 17 03:11:48 sys info: Linux domU-12-31-39-01-70-B4 2.6.21.7-2.fc8xen #1 SMP Fri Feb 15 12:39:36 EST 2008 i686 BOOST_LIB_VERSION=1_37
Tue Aug 17 03:11:48 [initandlisten] waiting for connections on port 27018
Tue Aug 17 03:11:48 [websvr] web admin interface listening on port 28018
[root@chn-xg-5-5c1 cache2]# 

3、配置config server，config server需要dbpath和repairpath
dbpath默认为/data/configdb，repairpath默认为/data/db/，这与shard是不同的
[root@localhost data]# mongod  --configsvr --fork --logpath /var/log/mongodb.log --logappend
[1] 2144
[root@localhost data]# Tue Aug 17 03:13:54 MongoDB starting : pid=2144 port=27019 dbpath=/data/configdb 32-bit 
** NOTE: when using MongoDB 32 bit, you are limited to about 2 gigabytes of data
**       see http://blog.mongodb.org/post/137788967/32-bit-limitations
Tue Aug 17 03:13:54 db version v1.6.0, pdfile version 4.5
Tue Aug 17 03:13:54 git version: 2c7f164b653f0d703947572ede064aed41cc2185
Tue Aug 17 03:13:54 sys info: Linux domU-12-31-39-01-70-B4 2.6.21.7-2.fc8xen #1 SMP Fri Feb 15 12:39:36 EST 2008 i686 BOOST_LIB_VERSION=1_37
Tue Aug 17 03:13:54 [initandlisten] diagLogging = 1
Tue Aug 17 03:13:54 [initandlisten] waiting for connections on port 27019
Tue Aug 17 03:13:54 [websvr] web admin interface listening on port 28019
[root@localhost data]# 
4、配置mongos router
[root@band-all local]# mongos --configdb 192.168.100.245:27019 --fork --logpath /var/log/mongodb.log --logappend
[1] 6384
[root@band-all local]# Tue Aug 17 03:46:12 mongos db version v1.6.0, pdfile version 4.5 starting (--help for usage)
Tue Aug 17 03:46:12 git version: 2c7f164b653f0d703947572ede064aed41cc2185
Tue Aug 17 03:46:12 sys info: Linux domU-12-31-39-01-70-B4 2.6.21.7-2.fc8xen #1 SMP Fri Feb 15 12:39:36 EST 2008 i686 BOOST_LIB_VERSION=1_37
Tue Aug 17 03:46:12 [websvr] web admin interface listening on port 28017
Tue Aug 17 03:46:12 waiting for connections on port 27017
[root@band-all local]# 


二、配置shard 集群

1、登陆mongos server，添加shard
[root@localhost data]# mongo 192.168.100.229:27017/admin
MongoDB shell version: 1.6.0
connecting to: 192.168.100.229:27017/admin
Tue Aug 17 05:59:38 [conn4] building new index on { _id: 1 } for config.databases
Tue Aug 17 05:59:38 [conn4] Buildindex config.databases idxNo:0 { name: "_id_", ns: "config.databases", key: { _id: 1 } }
Tue Aug 17 05:59:38 [conn4] done for 0 records 0.025secs
添加shard 1
>  db.runCommand( { addshard : "192.168.100.210:27018" } );
{ "shardAdded" : "shard0000", "ok" : 1 }
添加shard 2
创建shard时，两个shard不能有相同的数据库
>  db.runCommand( { addshard : "192.168.100.212:27018" } );
{
        "ok" : 0,
        "errmsg" : "trying to add shard 192.168.100.212:27018 because local database recommender exists in another shard0000:192.168.100.210:27018"
}
可以在192.168.100.212:27018通过 db.dropDatabase() 删除recommender
db.runCommand( { addshard : "192.168.100.212:27018" } );
查看shard节点列表
> db.runCommand( { listshards : 1 } );
{
        "shards" : [
                {
                        "_id" : "shard0000",
                        "host" : "192.168.100.210:27018"
                },
                {
                        "_id" : "shard0001",
                        "host" : "192.168.100.212:27018"
                }
        ],
        "ok" : 1
}
> 
将数据库设置为shard模式
> db.runCommand( { enablesharding : "test" } );
{ "ok" : 1 }
> db.runCommand( { enablesharding : "recommender" } );
{ "ok" : 1 }
> 

2、登陆mongos查看对集合进行分片
[root@band-all data]# mongo 192.168.100.229:27017/admin
MongoDB shell version: 1.6.0
connecting to: 192.168.100.229:27017/admin
> show dbs
admin
local
> use admin
switched to db admin
> db.runCommand({listshards:1})
{
        "shards" : [
                {
                        "_id" : "shard0000",
                        "host" : "192.168.100.210:27018"
                },
                {
                        "_id" : "shard0001",
                        "host" : "192.168.100.212:27018"
                }
        ],
        "ok" : 1
}
> db.printShardingStatus(); 
--- Sharding Status --- 
  sharding version: { "_id" : 1, "version" : 3 }
  shards:
      { "_id" : "shard0000", "host" : "192.168.100.210:27018" }
      { "_id" : "shard0001", "host" : "192.168.100.212:27018" }
  databases:
        { "_id" : "admin", "partitioned" : false, "primary" : "config" }
> cOnfig= connect("192.168.100.245:27019") //登陆config server
connecting to: 192.168.100.245:27019
192.168.100.245:27019
> test = db.getSisterDB("test")       //新增数据库test
> db.runCommand( { enablesharding : "test" } ) //将test 数据库进行分片
> db.runCommand( { shardcollection : "test.people", key : {name : 1} } )     //将数据集people进行分片
>  db.printShardingStatus(); 
--- Sharding Status --- 
  sharding version: { "_id" : 1, "version" : 3 }
  shards:
      { "_id" : "shard0000", "host" : "192.168.100.210:27018" }
      { "_id" : "shard0001", "host" : "192.168.100.212:27018" }
  databases:
        { "_id" : "admin", "partitioned" : false, "primary" : "config" }
        { "_id" : "test", "partitioned" : true, "primary" : "shard0000" }
                test.people chunks:
                        { "name" : { $minKey : 1 } } -->> { "name" : { $maxKey : 1 } } on : shard0000 { "t" : 1000, "i" : 0 }
> use test
switched to db test
> show collections
people
system.indexes
批量插入数据
for (var i = 1; i <= 200000; i++) db.people.save({name:"fly_people",sex:"female",age:25});
for (var i = 1; i <= 20000000; i++) db.people.save({name:"houxuefeng",sex:"male",age:30});
> db.people.stats()
{
        "sharded" : true,
        "ns" : "test.people",
        "count" : 462008,
        "size" : 33264600,
        "avgObjSize" : 72.00005194715243,
        "storageSize" : 902522624,
        "nindexes" : 2,
        "nchunks" : 4,
        "shards" : {
                "shard0000" : {
                        "ns" : "test.people",
                        "count" : 402006,
                        "size" : 28944448,
                        "avgObjSize" : 72.00003980040098,
                        "storageSize" : 891340544,
                        "numExtents" : 22,
                        "nindexes" : 2,
                        "lastExtentSize" : 155079936,
                        "paddingFactor" : 1,
                        "flags" : 1,
                        "totalIndexSize" : 58966016,
                        "indexSizes" : {
                                "_id_" : 18694144,
                                "name_1" : 40271872
                        },
                        "ok" : 1
                },
                "shard0001" : {
                        "ns" : "test.people",
                        "count" : 60002,
                        "size" : 4320152,
                        "avgObjSize" : 72.00013332888904,
                        "storageSize" : 11182080,
                        "numExtents" : 6,
                        "nindexes" : 2,
                        "lastExtentSize" : 8388608,
                        "paddingFactor" : 1,
                        "flags" : 1,
                        "totalIndexSize" : 5275648,
                        "indexSizes" : {
                                "_id_" : 2498560,
                                "name_1" : 2777088
                        },
                        "ok" : 1
                }
        },
        "ok" : 1
}
> 
至此MongoDB auto shard配置完毕

备注：
问题1、
当test.people数据量达到一定量时，这也就是32位限制，新数据无法入库，查看数据库文件也不在增加
>   for (var i = 1; i <= 20000; i++) db.people.save({name:"houxuefeng",sex:"male",age:30});
can't map file memory - mongo requires 64 bit build for larger datasets
> 
> db.people.stats();
{
        "ns" : "test.people",
        "count" : 10128817,
        "size" : 729274896,
        "avgObjSize" : 72.00000710843132,
        "storageSize" : 891340544,
        "numExtents" : 22,
        "nindexes" : 2,
        "lastExtentSize" : 155079936,
        "paddingFactor" : 1,
        "flags" : 1,
        "totalIndexSize" : 1241788192,
        "indexSizes" : {
                "_id_" : 420325264,
                "name_1" : 821462928
        },
        "ok" : 1
}
> 
[root@chn-xg-5-5c1 db]# ll -lh
总用量 3.0G
-rwxr-xr-x  1 root root    6  8月 18 17:42 mongod.lock
drwxr-xr-x  3 root root 4.0K  8月 19 10:48 moveChunk
-rw-------  1 root root  64M  8月 18 17:56 test.0
-rw-------  1 root root 128M  8月 18 17:56 test.1
-rw-------  1 root root 256M  8月 19 10:46 test.2
-rw-------  1 root root 512M  8月 19 10:48 test.3
-rw-------  1 root root 512M  8月 19 10:52 test.4
-rw-------  1 root root 512M  8月 19 10:58 test.5
-rw-------  1 root root 512M  8月 19 11:04 test.6
-rw-------  1 root root 512M  8月 19 11:12 test.7
-rw-------  1 root root  16M  8月 18 17:56 test.ns
drwxr-xr-x  2 root root 4.0K  8月 18 17:58 _tmp
[root@chn-xg-5-5c1 db]# 
问题2、
创建shard时，两个shard不能有相同的数据库
>  db.runCommand( { addshard : "192.168.100.212:27018" } );
{
        "ok" : 0,
        "errmsg" : "trying to add shard 192.168.100.212:27018 because local database recommender exists in another shard0000:192.168.100.210:27018"
}
可以在192.168.100.212:27018通过 db.dropDatabase() 删除recommender
问题3、
replica set 配置完毕后才可以添加为shard，否则会出错
> db.runCommand( { addshard : "foo/192.168.100.212:27018,192.168.100.212:27019,192.168.100.212:27020" } );
{ "shardAdded" : "shard0001", "ok" : 1 }
>