-
集群的模块介绍:
从MongoDB官方给的集群架构了解,整个集群主要有4个模块:Config Server、mongs、 shard、replica set:
Config Server:用来存放集群的元数据,也就是存放所有分片的的配置数据,mongos第一次启动就需要连接configServer读取相关数据,当configServer有数据进行更新时,也会主动推送消息到所有的mongos上,在3.0.7版本中,官方是建议配置3份的Config Server,以便挂掉两台时,业务还能够正常运转。
mongs:Mongodb集群的的请求入口,能否自动实现数据的分布式分发,生产环境中建议部署在应用服务器上。
shard:分片就比如是将一张大表分散在几个不同的shard中,实现数据分布式存储。
replica set:主要是对每个分片进行冗余,生产环境中,一般将将副本集配置在三个节点上,两份副本、一份仲裁。
-
环境规划:
-
端口及安装路径规划:
用途 | IP | 端口 | 备注 | 安装路径 |
ConfigeServer | 172.16.16.120 | 30001 | /db/configS | |
172.16.16.121 | 30001 | /db/configS | ||
172.16.16.122 | 30001 | /db/configS | ||
share1 | 172.16.16.124 | 40001 | Shard1主节点 | /db/shard1 |
172.16.16.125 | 40001 | Shard1副本节点 | /db/shard1 | |
172.16.16.126 | 40001 | Shard1仲裁节点 | /db/shard1 | |
share2 | 172.16.16.125 | 40002 | Shard2主节点 | /db/shard2 |
172.16.16.126 | 40002 | Shard2副本节点 | /db/shard2 | |
172.16.16.131 | 40002 | Shard2仲裁节点 | /db/shard2 | |
share3 | 172.16.16.126 | 40003 | Shard3主节点 | /db/shard3 |
172.16.16.131 | 40003 | Shard3副本节点 | /db/shard3 | |
172.16.16.124 | 40003 | Shard3仲裁节点 | /db/shard3 | |
share4 | 172.16.16.121 | 40004 | Shard4主节点 | /db/shard4 |
172.16.16.124 | 40004 | Shard4副本节点 | /db/shard4 | |
172.16.16.125 | 40004 | Shard4仲裁节点 | /db/shard4 | |
mongos | 172.16.16.124 | 50001 | 生产环境中一般直接部署在应用端 | /db/mongos |
172.16.16.125 | 50001 | /db/mongos | ||
172.16.16.126 | 50001 | /db/mongos | ||
172.16.16.131 | 50001 | /db/mongos |
-
搭建步骤:
-
下载mongodb(https://www.mongodb.org/downloads ) ,目前的最新版本是 3.0.7
opt]# tar zxvf mongodb-linux-x86_64-rhel55-3.0.7.gz opt]# mv mongodb-linux-x86_64-rhel55-3.0.7 /usr/local/mongodb opt]# useradd mongo opt]# passwd mongo Changing password for user mongo. New UNIX password: BAD PASSWORD: it is too simplistic/systematic Retype new UNIX password: passwd: all authentication tokens updated successfully. opt]# chown -R mongo:mongo /usr/local/mongodb/ opt]# chown -R mongo:mongo /db
-
创建相关模块存放路径:
-
创建configeServer目录(172.16.16.120/121/122):
#mkdir -p /db/configS/data & mkdir -p /db/configS/log (存放ConfigServer的数据、日志)
-
创建shard1目录(172.16.16.124/125/126):
#mkdir -p /db/shard1/data & mkdir -p /db/shard1/log (存放shard1的数据、日志)
-
创建shard2目录(172.16.16.125/126/131 ):
#mkdir -p /db/shard2/data & mkdir -p /db/shard2/log (存放shard2的数据、日志)
-
创建shard3目录(172.16.16.126/131/124 ):
#mkdir -p /db/shard3/data & mkdir -p /db/shard3/log (存放shard3的数据、日志)
-
创建shard4目录(172.16.16.131/124/125 ):
#mkdir -p /db/shard4/data & mkdir -p /db/shard4/log (存放shard4的数据、日志)
-
创建mongos目录(172.16.16.124/125/126/131)
#mkdir -p /db/mongos/log (由于mongos只做路由使用,不存数据,所以只需要建立log目录)
-
模块配置及启动:
-
configServer(172.16.16.120/121/122)配置及服务启动:
编写 /usr/local/mongodb/conf/configServer.conf,将参数都放在该文件中:
#vim /usr/local/mongodb/conf/configServer.conf #!/bin/bash systemLog: destination: file path: "/db/configS/log/configServer.log" #日志存储位置 logAppend: true storage: journal: #journal配置 enabled: true dbPath: "/db/configS/data" #数据文件存储位置 directoryPerDB: true #是否一个库一个文件夹 engine: wiredTiger #数据引擎 wiredTiger: #WT引擎配置 engineConfig: cacheSizeGB: 6 #设置为6G,默认为物理内存的一半 directoryForIndexes: true #是否将索引也按数据库名单独存储 journalCompressor: zlib collectionConfig: #表压缩配置 blockCompressor: zlib indexConfig: #索引配置 prefixCompression: true net: #端口配置 port: 30001 #另外两台需要分别修改为30002、30003 processManagement: #配置启动管理方式 fork: true sharding: #分片配置 clusterRole: configsvr #分片角色
启动configServer:
conf]$ /usr/local/mongodb/bin/mongod -f /usr/local/mongodb/conf/configServer.conf
-
mongos(172.16.16.124/125/126/131)配置及服务启动:
编写mongos.conf,将参数都放在该文件中(4台配置文件都一样):
#vim /usr/local/mongodb/conf/mongos.conf #!/bin/bash systemLog: destination: file path: "/db/mongos/log/mongos.log" logAppend: true net: port: 50001 sharding: configDB: 172.16.16.120:30001,172.16.16.121:30001,172.16.16.122:30001 processManagement: fork: true
启动mongos:应保证集群中设备的时间都是一样的,否则启动mongos会报错,若不相同,可先搭建一套NTP服务器
conf]$ /usr/local/mongodb/bin/mongos -f /usr/local/mongodb/conf/mongos.conf
-
shard1分片+副本集配置及服务启动(172.16.16.124/125/126 ):
#vim /usr/local/mongodb/conf/shard1.conf #!/bin/bash systemLog: destination: file path: "/db/shard1/log/shard1.log" #日志存储位置 logAppend: true storage: journal: #journal配置 enabled: true dbPath: "/db/shard1/data" #数据文件存储位置 directoryPerDB: true #是否一个库一个文件夹 engine: wiredTiger #数据引擎 wiredTiger: #WT引擎配置 engineConfig: cacheSizeGB: 6 #设置为6G,默认为物理内存的一半 directoryForIndexes: true #是否将索引也按数据库名单独存储 journalCompressor: zlib collectionConfig: #表压缩配置 blockCompressor: zlib indexConfig: #索引配置 prefixCompression: true net: #端口配置 port: 40001 processManagement: #配置启动管理方式 fork: true sharding: #分片配置 clusterRole: shardsvr replication: replSetName: shard1 #配置副本集名称
启动shard1 mongod:
conf]$ /usr/local/mongodb/bin/mongod -f /usr/local/mongodb/conf/shard1.conf
-
shard2分片+副本集配置及服务启动(172.16.16.125/126/131 ):
#vim /usr/local/mongodb/conf/shard2.conf #!/bin/bash systemLog: destination: file path: "/db/shard2/log/shard2.log" #日志存储位置 logAppend: true storage: journal: #journal配置 enabled: true dbPath: "/db/shard2/data" #数据文件存储位置 directoryPerDB: true #是否一个库一个文件夹 engine: wiredTiger #数据引擎 wiredTiger: #WT引擎配置 engineConfig: cacheSizeGB: 6 #设置为6G,默认为物理内存的一半 directoryForIndexes: true #是否将索引也按数据库名单独存储 journalCompressor: zlib collectionConfig: #表压缩配置 blockCompressor: zlib indexConfig: #索引配置 prefixCompression: true net: #端口配置 port: 40002 processManagement: #配置启动管理方式 fork: true sharding: #分片配置 clusterRole: shardsvr replication: #oplogSizeMB: replSetName: shard2 #配置副本集名称
启动shard2 mongod:
conf]$ /usr/local/mongodb/bin/mongod -f /usr/local/mongodb/conf/shard2.conf
-
shard3分片+副本集配置及服务启动(172.16.16.126/131/124 ):
#vim /usr/local/mongodb/conf/shard3.conf #!/bin/bash systemLog: destination: file path: "/db/shard3/log/shard3.log" #日志存储位置 logAppend: true storage: journal: #journal配置 enabled: true dbPath: "/db/shard3/data" #数据文件存储位置 directoryPerDB: true #是否一个库一个文件夹 engine: wiredTiger #数据引擎 wiredTiger: #WT引擎配置 engineConfig: cacheSizeGB: 6 #设置为6G,默认为物理内存的一半 directoryForIndexes: true #是否将索引也按数据库名单独存储 journalCompressor: zlib collectionConfig: #表压缩配置 blockCompressor: zlib indexConfig: #索引配置 prefixCompression: true net: #端口配置 port: 40003 processManagement: #配置启动管理方式 fork: true sharding: #分片配置 clusterRole: shardsvr replication: #oplogSizeMB: replSetName: shard3 #配置副本集名称
启动shara3 mongod:
conf]$ /usr/local/mongodb/bin/mongod -f /usr/local/mongodb/conf/shard3.conf
-
shard4分片+副本集配置及服务启动(172.16.16.131/124/125 ):
#vim /usr/local/mongodb/conf/shard4.conf #!/bin/bash systemLog: destination: file path: "/db/shard4/log/shard4.log" #日志存储位置 logAppend: true storage: journal: #journal配置 enabled: true dbPath: "/db/shard4/data" #数据文件存储位置 directoryPerDB: true #是否一个库一个文件夹 engine: wiredTiger #数据引擎 wiredTiger: #WT引擎配置 engineConfig: cacheSizeGB: 6 #设置为6G,默认为物理内存的一半 directoryForIndexes: true #是否将索引也按数据库名单独存储 journalCompressor: zlib collectionConfig: #表压缩配置 blockCompressor: zlib indexConfig: #索引配置 prefixCompression: true net: #端口配置 port: 40004 processManagement: #配置启动管理方式 fork: true sharding: #分片配置 clusterRole: shardsvr replication: #oplogSizeMB: replSetName: shard4 #复制集名
启动shara4 mongod:
conf]$ /usr/local/mongodb/bin/mongod -f /usr/local/mongodb/conf/shard4.conf
-
集群配置:
-
副本集配置(在每个 shard的主节点上进行配置及初始化,否则会初始化失败 ):
-
shard1的副本集配置(主节点、副本节点、仲裁节点):
bin]$ ./mongo 172.16.16.124:40001 MongoDB shell version: 3.0.7 connecting to: 172.16.16.124:40001/test > use admin switched to db admin > config = { _id:"shard1", members:[ {_id:0,host:"172.16.16.124:40001"}, {_id:1,host:"172.16.16.125:40001"}, {_id:2,host:"172.16.16.126:40001",arbiterOnly:true}] } #以下为输出 { "_id" : "shard1", "members" : [ { "_id" : 0, "host" : "172.16.16.124:40001" }, { "_id" : 1, "host" : "172.16.16.125:40001" }, { "_id" : 2, "host" : "172.16.16.126:40001", "arbiterOnly" : true } ] } > rs.initiate(config); #初始化配置 { "ok" : 1 }
-
shard2的副本集配置(主节点、副本节点、仲裁节点):
bin]$ ./mongo 172.16.16.125:40002 MongoDB shell version: 3.0.7 connecting to: 172.16.16.125:40002/test > use admin switched to db admin > config = { _id:"shard2", members:[ {_id:0,host:"172.16.16.125:40002"}, {_id:1,host:"172.16.16.126:40002"}, {_id:2,host:"172.16.16.131:40002",arbiterOnly:true}] } #以下为输出 { "_id" : "shard2", "members" : [ { "_id" : 0, "host" : "172.16.16.125:40002" }, { "_id" : 1, "host" : "172.16.16.126:40002" }, { "_id" : 2, "host" : "172.16.16.131:40002", "arbiterOnly" : true } ] } > rs.initiate(config); #初始化配置 { "ok" : 1 }
-
shard3的副本集配置(主节点、副本节点、仲裁节点):
bin]$ ./mongo 172.16.16.126:40003 MongoDB shell version: 3.0.7 connecting to: 172.16.16.126:40003/test > use admin switched to db admin > config = { _id:"shard3", members:[ {_id:0,host:"172.16.16.126:40003"}, {_id:1,host:"172.16.16.131:40003"}, {_id:2,host:"172.16.16.124:40003",arbiterOnly:true}] } #以下为输出 { "_id" : "shard3", "members" : [ { "_id" : 0, "host" : "172.16.16.126:40003" }, { "_id" : 1, "host" : "172.16.16.131:40003" }, { "_id" : 2, "host" : "172.16.16.124:40003", "arbiterOnly" : true } ] } > rs.initiate(config); #初始化配置 { "ok" : 1 }
-
shard4的副本集配置(主节点、副本节点、仲裁节点):
bin]$ ./mongo 172.16.16.131:40004 MongoDB shell version: 3.0.7 connecting to: 172.16.16.131:40004/test > use admin switched to db admin > config = { _id:"shard4", members:[ {_id:0,host:"172.16.16.131:40004"}, {_id:1,host:"172.16.16.124:40004"}, {_id:2,host:"172.16.16.125:40004",arbiterOnly:true}] } #以下为输出 { "_id" : "shard4", "members" : [ { "_id" : 0, "host" : "172.16.16.131:40004" }, { "_id" : 1, "host" : "172.16.16.124:40004" }, { "_id" : 2, "host" : "172.16.16.125:40004", "arbiterOnly" : true } ] } > rs.initiate(config); #初始化配置 { "ok" : 1 }
-
分片配置
bin]$ ./mongo 172.16.16.124:50001 mongos> use admin switched to db admin mongos> db.runCommand({addshard:"shard1/172.16.16.124:40001,172.16.16.125:40001,172.16.16.126:40001"}); { "shardAdded" : "shard1", "ok" : 1 } mongos>db.runCommand({addshard:"shard2/172.16.16.125:40002,172.16.16.126:40002,172.16.16.131:40002"}); { "shardAdded" : "shard2", "ok" : 1 } mongos>db.runCommand({addshard:"shard3/172.16.16.126:40003,172.16.16.131:40003,172.16.16.124:40003"}); { "shardAdded" : "shard3", "ok" : 1 } mongos>db.runCommand({addshard:"shard4/172.16.16.131:40004,172.16.16.124:40004,172.16.16.125:40004"}); { "shardAdded" : "shard4", "ok" : 1 }
-
查看配置是否生效(仲裁不被列出 ):
mongos> db.runCommand( { listshards : 1 } ); { "shards" : [ { "_id" : "shard1", "host" : "shard1/172.16.16.124:40001,172.16.16.125:40001" }, { "_id" : "shard2", "host" : "shard2/172.16.16.125:40002,172.16.16.126:40002" }, { "_id" : "shard3", "host" : "shard3/172.16.16.126:40003,172.16.16.131:40003" }, { "_id" : "shard4", "host" : "shard4/172.16.16.124:40004,172.16.16.131:40004" } ], "ok" : 1 }
以上就完成了MongoDB shard+replica模式的集群搭建,接下来做业务测试。
-
集群测试
默认情况下,库和集合没有自动分片的,若有数据写入,只会往一个shard中存储,做个测试验证:
bin]$ ./mongo 172.16.16.131:50001 MongoDB shell version: 3.0.7 connecting to: 172.16.16.131:50001/test mongos> use ljaidb switched to db ljaidb mongos> for (var i=1;i<=10000;i++) db.ljaitable.save({"name":"ljai","age":27,"addr":"fuzhou"}) WriteResult({ "nInserted" : 1 }) mongos> db.ljaitable.stats() { "sharded" : false, "primary" : "shard1", "ns" : "ljaidb.ljaitable", "count" : 10000, "size" : 670000, "avgObjSize" : 67, "storageSize" : 49152, "capped" : false, "wiredTiger" : { "metadata" : { "formatVersion" : 1 } mongos> db.printShardingStatus() --- Sharding Status --- sharding version: { "_id" : 1, "minCompatibleVersion" : 5, "currentVersion" : 6, "clusterId" : ObjectId("5625fc29e3c17fdff8517b73") } shards: { "_id" : "shard1", "host" : "shard1/172.16.16.124:40001,172.16.16.125:40001" } { "_id" : "shard2", "host" : "shard2/172.16.16.125:40002,172.16.16.126:40002" } { "_id" : "shard3", "host" : "shard3/172.16.16.126:40003,172.16.16.131:40003" } { "_id" : "shard4", "host" : "shard4/172.16.16.124:40004,172.16.16.131:40004" } balancer: Currently enabled: yes Currently running: yes Balancer lock taken at Tue Oct 20 2015 21:01:26 GMT+0800 (CST) by DataServer-04:50001:1445330413:1804289383:Balancer:846930886 Failed balancer rounds in last 5 attempts: 0 Migration Results for the last 24 hours: No recent migrations databases: { "_id" : "admin", "partitioned" : false, "primary" : "config" } { "_id" : "test", "partitioned" : false, "primary" : "shard1" } { "_id" : "ljaidb", "partitioned" : false, "primary" : "shard1" }
可以看出ljaidb库并没有分片,且数据都在shard1上,登录其他shard1上查看:
bin]$ ./mongo 172.16.16.124:40001 MongoDB shell version: 3.0.7 connecting to: 172.16.16.124:40001/test shard1:PRIMARY> show dbs ljaidb 0.000GB local 0.000GB shard1:PRIMARY> use ljaidb switched to db ljaidb shard1:PRIMARY> show tables ljaitable shard1:PRIMARY> db.ljaitable.find().count() 10000
验证shard2、shard3、shard4上都没有ljaidb这个库:
bin]$ ./mongo 172.16.16.125:40002 MongoDB shell version: 3.0.7 connecting to: 172.16.16.125:40002/test shard2:PRIMARY> show dbs local 0.000GB
-
指定数据库和集合进行分片:
为了让某个数据库与集合自动分片生效,对数据库(lymdb)及数据库(lymtable)中的表进行分片配置:
bin]$ ./mongo 172.16.16.124:50001 MongoDB shell version: 3.0.7 connecting to: 172.16.16.124:50001/test mongos> use admin switched to db admin mongos> db.runCommand( { enablesharding :"lymdb"}); { "ok" : 1 } mongos> db.runCommand( { shardcollection : "lymdb.lymtable",key : {_id: 1} } ) { "collectionsharded" : "lymdb.lymtable", "ok" : 1 }
-
通过java或者python驱动,连接mongo集群测试:
java连接代码:
1 import java.util.ArrayList; 2 import java.util.List; 3 4 import com.mongodb.BasicDBObject; 5 import com.mongodb.DB; 6 import com.mongodb.DBCollection; 7 import com.mongodb.DBObject; 8 import com.mongodb.MongoClient; 9 import com.mongodb.ServerAddress; 10 11 public class TestMongoDBShards { 12 13 public static void main(String[] args) { 14 try { 15 List<ServerAddress> addresses = new ArrayList<ServerAddress>(); 16 ServerAddress address1 = new ServerAddress("172.16.16.124" , 50001); 17 ServerAddress address2 = new ServerAddress("172.16.16.125" , 50001); 18 ServerAddress address3 = new ServerAddress("172.16.16.126" , 50001); 19 ServerAddress address4 = new ServerAddress("172.16.16.131" , 50001); 20 addresses.add(address1); 21 addresses.add(address2); 22 addresses.add(address3); 23 24 MongoClient client = new MongoClient(addresses); 25 DB db = client.getDB( "lymdb" ); 26 DBCollection coll = db.getCollection( "lymtable" ); 27 28 // BasicDBObject object = new BasicDBObject(); 29 // object.append( "id" , 1); 30 31 // DBObject dbObject = coll.findOne(object); 32 33 for(int i=1;i<=1000000;i++) { 34 DBObject saveData=new BasicDBObject(); 35 saveData.put("id", i); 36 saveData.put("userName", "baiwan" + i); 37 saveData.put("age", "26"); 38 saveData.put("gender", "m"); 39 40 41 coll.save(saveData); 42 } 43 44 45 46 47 // System. out .println(dbObject); 48 49 } catch (Exception e) { 50 e.printStackTrace(); 51 } 52 // TODO Auto-generated method stub 53 54 } 55 56 }
python连接代码:
1 #encoding=UTF-8 2 import datetime 3 4 ISOTIMEFORMAT = ‘%Y-%m-%d %X‘ 5 6 from pymongo import MongoClient 7 conn = MongoClient("172.16.16.124",50001) 8 db = conn.funodb 9 def dateDiffInSeconds(date1,date2): 10 timedelta = date2 - date1 11 return timedelta.days*24*3600 +timedelta.seconds 12 db.funotable.drop() 13 date1 = datetime.datetime.now() 14 for i in range(0,1000000): db.funotable.insert({"name":"ljai","age":i,"addr":"fuzhou"}) 15 c = db.funotable.find().count() 16 print("count is ",c) 17 date2 = datetime.datetime.now() 18 print(date1) 19 print(date2) 20 print("消耗:",dateDiffInSeconds(date1,date2),"seconds") 21 conn.close()
-
测试是否自动分片:
mongos> db.lymtable.getShardDistribution() Shard shard1 at shard1/172.16.16.124:40001,172.16.16.125:40001 data : 96.46MiB docs : 1216064 chunks : 4 estimated data per chunk : 24.11MiB estimated docs per chunk : 304016 Shard shard2 at shard2/172.16.16.125:40002,172.16.16.126:40002 data : 44.9MiB docs : 565289 chunks : 4 estimated data per chunk : 11.22MiB estimated docs per chunk : 141322 Shard shard3 at shard3/172.16.16.126:40003,172.16.16.131:40003 data : 99.39MiB docs : 1259979 chunks : 4 estimated data per chunk : 24.84MiB estimated docs per chunk : 314994 Shard shard4 at shard4/172.16.16.124:40004,172.16.16.131:40004 data : 76.46MiB docs : 958668 chunks : 4 estimated data per chunk : 19.11MiB estimated docs per chunk : 239667 Totals data : 317.22MiB docs : 4000000 chunks : 16 Shard shard1 contains 30.4% data, 30.4% docs in cluster, avg obj size on shard : 83B Shard shard2 contains 14.15% data, 14.13% docs in cluster, avg obj size on shard : 83B Shard shard3 contains 31.33% data, 31.49% docs in cluster, avg obj size on shard : 82B Shard shard4 contains 24.1% data, 23.96% docs in cluster, avg obj size on shard : 83B
可以看出,插入400万条数据,都有分布各个shard上,但是不够均匀,需要进一步研究分片的配置。