图解MongoDB原理(二)

根据上篇文章《图解MongoDB原理》的基本介绍，本文以具体的操作为例继续深入解析mongodb。

演示的所有服务都是在一台机器上，我的目录是F:\JAVA\MongoDB\cluster，文件结构如下：

说明：1个配置服务结点，1个路由服务结点，三个分片结点。

启动流程

1、启动config服务

F:\JAVA\MongoDB\cluster\mongo\mongod -configsvr -dbpath F:\JAVA\MongoDB\cluster\data\config -port 20000 -rest

2、启动mongos服务

F:\JAVA\MongoDB\cluster\mongo\mongos -configdb 127.0.0.1:20000 -port 30000 -chunkSize 5

3、启动所有分片服务

F:\JAVA\MongoDB\cluster\mongo\mongod -shardsvr -master -port 27001 -dbpath F:\JAVA\MongoDB\cluster\data\shard1 -logpath F:\JAVA\MongoDB\cluster\data\shard1\log -logappend

F:\JAVA\MongoDB\cluster\mongo\mongod -shardsvr -master -port 27002 -dbpath F:\JAVA\MongoDB\cluster\data\shard2 -logpath F:\JAVA\MongoDB\cluster\data\shard2\log -logappend

F:\JAVA\MongoDB\cluster\mongo\mongod -shardsvr -master -port 27003 -dbpath F:\JAVA\MongoDB\cluster\data\shard3 -logpath F:\JAVA\MongoDB\cluster\data\shard3\log -logappend

4、连接mongos服务

F:\JAVA\MongoDB\cluster\mongos>mongo.exe 127.0.0.1:30000
MongoDB shell version: 2.4.8
connecting to: 127.0.0.1:30000/test

5、切换为admin用户

mongos> use admin
switched to db admin

6、设置分片：

mongos> db.runCommand({addshard:"127.0.0.1:27001"})
{ "shardAdded" : "shard0000", "ok" : 1 }
mongos> db.runCommand({addshard:"127.0.0.1:27002"})
{ "shardAdded" : "shard0001", "ok" : 1 }
mongos> db.runCommand({addshard:"127.0.0.1:27003"})
{ "shardAdded" : "shard0002", "ok" : 1 }

7、创建用户名和密码：

mongos> db.addUser("vince","vince")
{
        "user" : "vince",
        "readOnly" : false,
        "pwd" : "507acb958e6a63558641743dadb4014e",
        "_id" : ObjectId("548be0b77bee4afe9ca1453b")
}

8、对数据库和表设置分片：

mongos> db.runCommand({enablesharding:"message"})
{ "ok" : 1 }
mongos> db.runCommand({shardcollection:"message.history",key:{_id:1}})
{ "collectionsharded" : "message.history", "ok" : 1 }

9、查询所有分片：

mongos> db.runCommand({listshards:1})
{
        "shards" : [
                {
                        "_id" : "shard0000",
                        "host" : "127.0.0.1:27031"
                },
                {
                        "_id" : "shard0001",
                        "host" : "127.0.0.1:27032"
                }
        ],
        "ok" : 1
}

常见错误

没有设置分片时创建用户报错：

mongos> db.addUser("vince","vince")
Sat Dec 13 14:38:56.116 error: {
        "$err" : "error creating initial database config information :: caused b
y :: can't find a shard to put new db on",
        "code" : 10185
} at src/mongo/shell/query.js:128

删除分片时上个任务未完成时报错：

mongos> db.runCommand({removeshard:"127.0.0.1:27032"})
{ "ok" : 0, "errmsg" : "Can't have more than one draining shard at a time" }

MongoVUE连接测试

连接界面如下：

查看分片：

每个chunk 的大小（size）（单位为MB）：

单个分片存储数据效果：

增加分片shard2和shard3后的Balancer操作过程：

分片后的数据：

查看所有的数据库，其中partitioned为True表示这个数据库设置了分片。

config数据库的结构：

查看服务状态，可以看到MongoDB支持的最大连接数为20000：

mongod启动参数

当启动参数报错时会有提示，还有参数的英文解释：

General options:
  -h [ --help ]               show this usage information
  --version                   show version information
  -f [ --config ] arg         configuration file specifying additional options
  -v [ --verbose ]            be more verbose (include multiple times for more
                              verbosity e.g. -vvvvv)
  --quiet                     quieter output
  --port arg                  specify port number - 27017 by default
  --bind_ip arg               comma separated list of ip addresses to listen on
                              - all local ips by default
  --maxConns arg              max number of simultaneous connections - 20000 by
                              default
  --logpath arg               log file to send write to instead of stdout - has
                              to be a file, not directory
  --logappend                 append to logpath instead of over-writing
  --pidfilepath arg           full path to pidfile (if not set, no pidfile is
                              created)
  --keyFile arg               private key for cluster authentication
  --setParameter arg          Set a configurable parameter
  --auth                      run with security
  --cpu                       periodically show cpu and iowait utilization
  --dbpath arg                directory for datafiles - defaults to \data\db\
  --diaglog arg               0=off 1=W 2=R 3=both 7=W+some reads
  --directoryperdb            each database will be stored in a separate
                              directory
  --ipv6                      enable IPv6 support (disabled by default)
  --journal                   enable journaling
  --journalCommitInterval arg how often to group/batch commit (ms)
  --journalOptions arg        journal diagnostic options
  --jsonp                     allow JSONP access via http (has security
                              implications)
  --noauth                    run without security
  --nohttpinterface           disable http interface
  --nojournal                 disable journaling (journaling is on by default
                              for 64 bit)
  --noprealloc                disable data file preallocation - will often hurt
                              performance
  --noscripting               disable scripting engine
  --notablescan               do not allow table scans
  --nssize arg (=16)          .ns file size (in MB) for new databases
  --profile arg               0=off 1=slow, 2=all
  --quota                     limits each database to a certain number of files
                              (8 default)
  --quotaFiles arg            number of files allowed per db, requires --quota
  --repair                    run repair on all dbs
  --repairpath arg            root directory for repair files - defaults to
                              dbpath
  --rest                      turn on simple rest api
  --slowms arg (=100)         value of slow for profile and console log
  --smallfiles                use a smaller default file size
  --syncdelay arg (=60)       seconds between disk syncs (0=never, but not
                              recommended)
  --sysinfo                   print some diagnostic system information
  --upgrade                   upgrade db if needed

Windows Service Control Manager options:
  --install                install Windows service
  --remove                 remove Windows service
  --reinstall              reinstall Windows service (equivalent to --remove
                           followed by --install)
  --serviceName arg        Windows service name
  --serviceDisplayName arg Windows service display name
  --serviceDescription arg Windows service description
  --serviceUser arg        account for service execution
  --servicePassword arg    password used to authenticate serviceUser

Replication options:
  --oplogSize arg       size to use (in MB) for replication op log. default is
                        5% of disk space (i.e. large is good)

Master/slave options (old; use replica sets instead):
  --master              master mode
  --slave               slave mode
  --source arg          when slave: specify master as <server:port>
  --only arg            when slave: specify a single database to replicate
  --slavedelay arg      specify delay (in seconds) to be used when applying
                        master ops to slave
  --autoresync          automatically resync if slave data is stale

Replica set options:
  --replSet arg           arg is <setname>[/<optionalseedhostlist>]
  --replIndexPrefetch arg specify index prefetching behavior (if secondary)
                          [none|_id_only|all]

Sharding options:
  --configsvr           declare this is a config db of a cluster; default port
                        27019; default dir /data/configdb
  --shardsvr            declare this is a shard db of a cluster; default port
                        27018

mongos启动参数

General options:
  -h [ --help ]         show this usage information
  --version             show version information
  -f [ --config ] arg   configuration file specifying additional options
  -v [ --verbose ]      be more verbose (include multiple times for more
                        verbosity e.g. -vvvvv)
  --quiet               quieter output
  --port arg            specify port number - 27017 by default
  --bind_ip arg         comma separated list of ip addresses to listen on - all
                        local ips by default
  --maxConns arg        max number of simultaneous connections - 20000 by
                        default
  --logpath arg         log file to send write to instead of stdout - has to be
                        a file, not directory
  --logappend           append to logpath instead of over-writing
  --pidfilepath arg     full path to pidfile (if not set, no pidfile is
                        created)
  --keyFile arg         private key for cluster authentication
  --setParameter arg    Set a configurable parameter
  --nohttpinterface     disable http interface

Windows Service Control Manager options:
  --install                install Windows service
  --remove                 remove Windows service
  --reinstall              reinstall Windows service (equivalent to --remove
                           followed by --install)
  --serviceName arg        Windows service name
  --serviceDisplayName arg Windows service display name
  --serviceDescription arg Windows service description
  --serviceUser arg        account for service execution
  --servicePassword arg    password used to authenticate serviceUser

Sharding options:
  --configdb arg        1 or 3 comma separated config servers
  --localThreshold arg  ping time (in ms) for a node to be considered local
                        (default 15ms)
  --test                just run unit tests
  --upgrade             upgrade meta data version
  --chunkSize arg       maximum amount of data per chunk
  --ipv6                enable IPv6 support (disabled by default)
  --jsonp               allow JSONP access via http (has security implications)
  --noscripting         disable scripting engine

分片+主从安装命令

主从的部署是不具有故障自动转移功能的，而复本集的部署支持故障转移。

主数据结点：
mongod -shardsvr -master -port 27001 -dbpath F:\JAVA\MongoDB\cluster\shard\data
从数据结点：
mongod -shardsvr -slave -source 127.0.0.1:27001 -port 27002 -dbpath F:\JAVA\MongoDB\cluster\shards\data

分片+复本集安装

复本集安装要对每个分片进行设置，用mongo直接连接分片，执行rs.initiate(cfg)进行设置，设置好后需要分别将每个分片加入config中。相关的函数有：

        rs.status()                     { replSetGetStatus : 1 } checks repl set status
        rs.initiate()                   { replSetInitiate : null } initiates set with default settings
        rs.initiate(cfg)                { replSetInitiate : cfg } initiates setwith configuration cfg
        rs.conf()                       get the current configuration object from local.system.replset
        rs.reconfig(cfg)                updates the configuration of a running replica set with cfg (disconnects)
        rs.add(hostportstr)             add a new member to the set with default attributes (disconnects)
        rs.add(membercfgobj)            add a new member to the set with extra attributes (disconnects)
        rs.addArb(hostportstr)          add a new member which is arbiterOnly:true (disconnects)
        rs.stepDown([secs])             step down as primary (momentarily) (disconnects)
        rs.syncFrom(hostportstr)        make a secondary to sync from the givenmember
        rs.freeze(secs)                 make a node ineligible to become primary for the time specified
        rs.remove(hostportstr)          remove a host from the replica set (disconnects)
        rs.slaveOk()                    shorthand for db.getMongo().setSlaveOk()

启动命令

路由结点：
mongos -configdb 127.0.0.1:20000 -port 30000 -chunkSize 5

配置结点：

mongod -configsvr -dbpath F:\JAVA\MongoDB\cluster\config\data -port 20000 -rest

分片1：
mongod -shardsvr -replSet shard1/127.0.0.1:27051,127.0.0.1:27041 -port 27031 -dbpath F:\JAVA\MongoDB\cluster\repa_shard1\data

分片1复本集b：
mongod -shardsvr -replSet shard1/127.0.0.1:27051 -port 27041 -dbpath F:\JAVA\MongoDB\cluster\repb_shard1\data

分片1复本集c：
mongod -shardsvr -replSet shard1/127.0.0.1:27041 -port 27051 -dbpath F:\JAVA\MongoDB\cluster\repc_shard1\data

分片2：
mongod -shardsvr -replSet shard2/127.0.0.1:27052,127.0.0.1:27042 -port 27032 -dbpath F:\JAVA\MongoDB\cluster\repa_shard2\data

分片2复本集b：
mongod -shardsvr -replSet shard2/127.0.0.1:27052 -port 27042 -dbpath F:\JAVA\MongoDB\cluster\repb_shard2\data

分片2复本集c：
mongod -shardsvr -replSet shard2/127.0.0.1:27042 -port 27052 -dbpath F:\JAVA\MongoDB\cluster\repc_shard2\data

WEB控制台

分片服务与配置服务的mongod加上-rest后会有web控制台，具体的地址在启动输出中会有，如下