MongoDB基础请参考:http://blog.51cto.com/kaliarch/2044423
MongoDB(replica set)请参考:http://blog.51cto.com/kaliarch/2044618
一、概述
1.1 背景
为解决mongodb在replica set每个从节点上面的数据库均是对数据库的全量拷贝,从节点压力在高并发大数据量的场景下存在很大挑战,同时考虑到后期mongodb集群的在数据压力巨大时的扩展性,应对海量数据引出了分片机制。
1.2 分片概念
分片是将数据库进行拆分,将其分散在不同的机器上的过程,无需功能强大的服务器就可以存储更多的数据,处理更大的负载,在总数据中,将集合切成小块,将这些块分散到若干片中,每个片只负载总数据的一部分,通过一个知道数据与片对应关系的组件mongos的路由进程进行操作。
1.3 基础组件
其利用到了四个组件:mongos,config server,shard,replica set
mongos:数据库集群请求的入口,所有请求需要经过mongos进行协调,无需在应用层面利用程序来进行路由选择,mongos其自身是一个请求分发中心,负责将外部的请求分发到对应的shard服务器上,mongos作为统一的请求入口,为防止mongos单节点故障,一般需要对其做HA。
config server:配置服务器,存储所有数据库元数据(分片,路由)的配置。mongos本身没有物理存储分片服务器和数据路由信息,只是存缓存在内存中来读取数据,mongos在第一次启动或后期重启时候,就会从config server中加载配置信息,如果配置服务器信息发生更新会通知所有的mongos来更新自己的状态,从而保证准确的请求路由,生产环境中通常也需要多个config server,防止配置文件存在单节点丢失问题。
shard:在传统意义上来讲,如果存在海量数据,单台服务器存储1T压力非常大,无论考虑数据库的硬盘,网络IO,又有CPU,内存的瓶颈,如果多台进行分摊1T的数据,到每台上就是可估量的较小数据,在mongodb集群只要设置好分片规则,通过mongos操作数据库,就可以自动把对应的操作请求转发到对应的后端分片服务器上。
replica set:在总体mongodb集群架构中,对应的分片节点,如果单台机器下线,对应整个集群的数据就会出现部分缺失,这是不能发生的,因此对于shard节点需要replica set来保证数据的可靠性,生产环境通常为2个副本+1个仲裁。
1.4 架构图
二、安装部署
2.1 基础环境
为了节省服务器,采用多实例配置,三个mongos,三个config server,单个服务器上面运行不通角色的shard(为了后期数据分片均匀,将三台shard在各个服务器上充当不同的角色。),在一个节点内采用replica set保证高可用,对应主机与端口信息如下:
主机名
IP地址
组件mongos
组件config server
shard
mongodb-1
172.20.6.10
端口:20000
端口:21000
主节点: 22001
副本节点:22002
仲裁节点:22003
mongodb-2
172.20.6.11
端口:20000
端口:21000
仲裁节点:22001
主节点: 22002
副本节点:22003
mongodb-3
172.20.6.12
端口:20000
端口:21000
副本节点:22001
仲裁节点:22002
主节点: 22003
2.2、安装部署
2.2.1 软件下载目录创建
wget -c https://fastdl.mongodb.org/linux/mongodb-linux-x86_64-rhel62-3.4.10.tgz
tar -zxvf mongodb-linux-x86_64-rhel62-3.4.10.tgz
ln -sv mongodb-linux-x86_64-rhel62-3.4.10 mongodb
echo "PATH=$PAHT:/usr/local/mongodb/bin">/etc/profile.d/mongodb.sh
source /etc/profile.d/mongodb.sh
2.2.2 创建目录
分别在mongodb-1/mongodb-2/mongodb-3创建目录及日志文件
mkdir -p /data/mongodb/mongos/{log,conf}
mkdir -p /data/mongodb/mongoconf/{data,log,conf}
mkdir -p /data/mongodb/shard1/{data,log,conf}
mkdir -p /data/mongodb/shard2/{data,log,conf}
mkdir -p /data/mongodb/shard3/{data,log,conf}
touch /data/mongodb/mongos/log/mongos.log
touch /data/mongodb/mongoconf/log/mongoconf.log
touch /data/mongodb/shard1/log/shard1.log
touch /data/mongodb/shard2/log/shard2.log
touch /data/mongodb/shard3/log/shard3.log
2.2.3 配置config server副本集
在mongodb3.4版本后要求配置服务器也创建为副本集,在此副本集名称:replconf
在三台服务器上配置config server副本集配置文件,并启动服务
cat>/data/mongodb/mongoconf/conf/mongoconf.conf<
logpath=/data/mongodb/mongoconf/log/mongoconf.log #定义config server日志文件
logappend=true
port = 21000
maxConns = 1000 #链接数
journal = true #日志开启
journalCommitInterval = 200
fork = true #后台执行
syncdelay = 60
oplogSize = 1000
configsvr = true #配置服务器
replSet=replconf #config server配置集replconf
EOF
mongod -f /data/mongodb/mongoconf/conf/mongoconf.conf #三台服务器均启动config server
任意登录一台服务器进行配置服务器副本集初始化
use admin
config = {_id:"replconf",members:[
{_id:0,host:"172.20.6.10:21000"},
{_id:1,host:"172.20.6.11:21000"},
{_id:2,host:"172.20.6.12:21000"},]
}
rs.initiate(config);
查看集群状态:
replconf:OTHER> rs.status()
{
"set" : "replconf",
"date" : ISODate("2017-12-04T07:42:09.054Z"),
"myState" : 1,
"term" : NumberLong(1),
"configsvr" : true,
"heartbeatIntervalMillis" : NumberLong(2000),
"optimes" : {
"lastCommittedOpTime" : {
"ts" : Timestamp(1512373328, 1),
"t" : NumberLong(1)
},
"readConcernMajorityOpTime" : {
"ts" : Timestamp(1512373328, 1),
"t" : NumberLong(1)
},
"appliedOpTime" : {
"ts" : Timestamp(1512373328, 1),
"t" : NumberLong(1)
},
"durableOpTime" : {
"ts" : Timestamp(1512373328, 1),
"t" : NumberLong(1)
}
},
"members" : [
{
"_id" : 0,
"name" : "172.20.6.10:21000",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY",
"uptime" : 221,
"optime" : {
"ts" : Timestamp(1512373328, 1),
"t" : NumberLong(1)
},
"optimeDate" : ISODate("2017-12-04T07:42:08Z"),
"infoMessage" : "could not find member to sync from",
"electionTime" : Timestamp(1512373296, 1),
"electionDate" : ISODate("2017-12-04T07:41:36Z"),
"configVersion" : 1,
"self" : true
},
{
"_id" : 1,
"name" : "172.20.6.11:21000",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 42,
"optime" : {
"ts" : Timestamp(1512373318, 1),
"t" : NumberLong(1)
},
"optimeDurable" : {
"ts" : Timestamp(1512373318, 1),
"t" : NumberLong(1)
},
"optimeDate" : ISODate("2017-12-04T07:41:58Z"),
"optimeDurableDate" : ISODate("2017-12-04T07:41:58Z"),
"lastHeartbeat" : ISODate("2017-12-04T07:42:08.637Z"),
"lastHeartbeatRecv" : ISODate("2017-12-04T07:42:07.648Z"),
"pingMs" : NumberLong(0),
"syncingTo" : "172.20.6.10:21000",
"configVersion" : 1
},
{
"_id" : 2,
"name" : "172.20.6.12:21000",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY",
"uptime" : 42,
"optime" : {
"ts" : Timestamp(1512373318, 1),
"t" : NumberLong(1)
},
"optimeDurable" : {
"ts" : Timestamp(1512373318, 1),
"t" : NumberLong(1)
},
"optimeDate" : ISODate("2017-12-04T07:41:58Z"),
"optimeDurableDate" : ISODate("2017-12-04T07:41:58Z"),
"lastHeartbeat" : ISODate("2017-12-04T07:42:08.637Z"),
"lastHeartbeatRecv" : ISODate("2017-12-04T07:42:07.642Z"),
"pingMs" : NumberLong(0),
"syncingTo" : "172.20.6.10:21000",
"configVersion" : 1
}
],
"ok" : 1
}
此时config server集群已经配置完成,mongodb-1为primary,mongdb-2/mongodb-3为secondary
2.2.4 配置shard集群
三台服务器均进行shard集群配置
shard1配置
cat >/data/mongodb/shard1/conf/shard.conf <
dbpath=/data/mongodb/shard1/data
logpath = /data/mongodb/shard1/log/shard1.log
port = 22001
logappend = true
nohttpinterface = true
fork = true
oplogSize = 4096
journal = true
#engine = wiredTiger
#cacheSizeGB = 38G
smallfiles=true
shardsvr=true #shard服务器
replSet=shard1 #副本集名称shard1
EOF
mongod -f /data/mongodb/shard1/conf/shard.conf #启动shard服务
查看此时服务已经正常启动,shard1的22001端口已经正常监听,接下来登录mongodb-1服务器进行shard1副本集初始化
mongo 172.20.6.10:22001
use admin
config = {_id:"shard1",members:[
{_id:0,host:"172.20.6.10:22001"},
{_id:1,host:"172.20.6.11:22001",arbiterOnly:true},
{_id:2,host:"172.20.6.12:22001"},]
}
rs.initiate(config);
查看集群状态,只列出了部分信息:
{
"_id" : 0,
"name" : "172.20.6.10:22001",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY", #mongodb-1为primary
"uptime" : 276,
"optime" : {
"ts" : Timestamp(1512373911, 1),
"t" : NumberLong(1)
},
"optimeDate" : ISODate("2017-12-04T07:51:51Z"),
"infoMessage" : "could not find member to sync from",
"electionTime" : Timestamp(1512373879, 1),
"electionDate" : ISODate("2017-12-04T07:51:19Z"),
"configVersion" : 1,
"self" : true
},
{
"_id" : 1,
"name" : "172.20.6.11:22001",
"health" : 1,
"state" : 7,
"stateStr" : "ARBITER", #mongodb-2为arbiter
"uptime" : 45,
"lastHeartbeat" : ISODate("2017-12-04T07:51:53.597Z"),
"lastHeartbeatRecv" : ISODate("2017-12-04T07:51:51.243Z"),
"pingMs" : NumberLong(0),
"configVersion" : 1
},
{
"_id" : 2,
"name" : "172.20.6.12:22001",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY", #mongodb-3为secondary
"uptime" : 45,
"optime" : {
"ts" : Timestamp(1512373911, 1),
"t" : NumberLong(1)
},
此时shard1 副本集已经配置完成,mongodb-1为primary,mongodb-2为arbiter,mongodb-3为secondary。
同样的操作进行shard2配置和shard3配置
注意:进行shard2的副本集初始化,在mongodb-2, 初始化shard3副本集在mongodb-3上进行操作。
shard2配置文件
cat >/data/mongodb/shard2/conf/shard.conf <
dbpath=/data/mongodb/shard2/data
logpath = /data/mongodb/shard2/log/shard2.log
port = 22002
logappend = true
nohttpinterface = true
fork = true
oplogSize = 4096
journal = true
#engine = wiredTiger
#cacheSizeGB = 38G
smallfiles=true
shardsvr=true
replSet=shard2
EOF
mongod -f /data/mongodb/shard2/conf/shard.conf
shard3配置文件
cat >/data/mongodb/shard3/conf/shard.conf <
dbpath=/data/mongodb/shard3/data
logpath = /data/mongodb/shard3/log/shard3.log
port = 22003
logappend = true
nohttpinterface = true
fork = true
oplogSize = 4096
journal = true
#engine = wiredTiger
#cacheSizeGB = 38G
smallfiles=true
shardsvr=true
replSet=shard3
EOF
mongod -f /data/mongodb/shard2/conf/shard.conf
在mongodb-2上进行shard2副本集初始化
mongo 172.20.6.11:22002 #登录mongodb-2
use admin
config = {_id:"shard2",members:[
{_id:0,host:"172.20.6.10:22002"},
{_id:1,host:"172.20.6.11:22002"},
{_id:2,host:"172.20.6.12:22002",arbiterOnly:true},]
}
rs.initiate(config);
查看shard2副本集状态
{
"_id" : 0,
"name" : "172.20.6.10:22002",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY", #mongodb-2为secondary
"uptime" : 15,
"optime" : {
"ts" : Timestamp(1512374668, 1),
"t" : NumberLong(1)
},
"optimeDurable" : {
"ts" : Timestamp(1512374668, 1),
"t" : NumberLong(1)
},
"optimeDate" : ISODate("2017-12-04T08:04:28Z"),
"optimeDurableDate" : ISODate("2017-12-04T08:04:28Z"),
"lastHeartbeat" : ISODate("2017-12-04T08:04:30.527Z"),
"lastHeartbeatRecv" : ISODate("2017-12-04T08:04:28.492Z"),
"pingMs" : NumberLong(0),
"syncingTo" : "172.20.6.11:22002",
"configVersion" : 1
},
{
"_id" : 1,
"name" : "172.20.6.11:22002",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY", #mongodb-2为primary
"uptime" : 211,
"optime" : {
"ts" : Timestamp(1512374668, 1),
"t" : NumberLong(1)
},
"optimeDate" : ISODate("2017-12-04T08:04:28Z"),
"infoMessage" : "could not find member to sync from",
"electionTime" : Timestamp(1512374666, 1),
"electionDate" : ISODate("2017-12-04T08:04:26Z"),
"configVersion" : 1,
"self" : true
},
{
"_id" : 2,
"name" : "172.20.6.12:22002", #mongodb-3为arbiter
"health" : 1,
"state" : 7,
"stateStr" : "ARBITER",
"uptime" : 15,
"lastHeartbeat" : ISODate("2017-12-04T08:04:30.527Z"),
"lastHeartbeatRecv" : ISODate("2017-12-04T08:04:28.384Z"),
"pingMs" : NumberLong(0),
"configVersion" : 1
}
登录mongodb-3进行shard3副本集初始化
mongo 172.20.6.12:22003 #登录mongodb-3
use admin
config = {_id:"shard3",members:[
{_id:0,host:"172.20.6.10:22003",arbiterOnly:true},
{_id:1,host:"172.20.6.11:22003"},
{_id:2,host:"172.20.6.12:22003"},]
}
rs.initiate(config);
查看shard3副本集状态
{
"_id" : 0,
"name" : "172.20.6.10:22003",
"health" : 1,
"state" : 7,
"stateStr" : "ARBITER", #mongodb-1为arbiter
"uptime" : 18,
"lastHeartbeat" : ISODate("2017-12-04T08:07:37.488Z"),
"lastHeartbeatRecv" : ISODate("2017-12-04T08:07:36.224Z"),
"pingMs" : NumberLong(0),
"configVersion" : 1
},
{
"_id" : 1,
"name" : "172.20.6.11:22003",
"health" : 1,
"state" : 2,
"stateStr" : "SECONDARY", #mongodb-2为secondary
"uptime" : 18,
"optime" : {
"ts" : Timestamp(1512374851, 1),
"t" : NumberLong(1)
},
"optimeDurable" : {
"ts" : Timestamp(1512374851, 1),
"t" : NumberLong(1)
},
"optimeDate" : ISODate("2017-12-04T08:07:31Z"),
"optimeDurableDate" : ISODate("2017-12-04T08:07:31Z"),
"lastHeartbeat" : ISODate("2017-12-04T08:07:37.488Z"),
"lastHeartbeatRecv" : ISODate("2017-12-04T08:07:36.297Z"),
"pingMs" : NumberLong(0),
"syncingTo" : "172.20.6.12:22003",
"configVersion" : 1
},
{
"_id" : 2,
"name" : "172.20.6.12:22003",
"health" : 1,
"state" : 1,
"stateStr" : "PRIMARY", #mongodb-3为primary
"uptime" : 380,
"optime" : {
"ts" : Timestamp(1512374851, 1),
"t" : NumberLong(1)
},
"optimeDate" : ISODate("2017-12-04T08:07:31Z"),
"infoMessage" : "could not find member to sync from",
"electionTime" : Timestamp(1512374849, 1),
"electionDate" : ISODate("2017-12-04T08:07:29Z"),
"configVersion" : 1,
"self" : true
}
此时shard集群全部已经配置完毕。
2.2.5 配置路由服务器mongos
目前三台服务器的配置服务器和分片服务器均已启动,配置三台mongos服务器
由于mongos服务器的配置是从内存中加载,所以自己没有存在数据目录configdb连接为配置服务器集群
cat >/data/mongodb/mongos/conf/mongos.conf<
--logpath=/data/mongodb/mongos/log/mongos.log
logappend=true
port = 20000
maxConns = 1000
configdb=replconf/172.20.6.10:21000,172.20.6.11:21000,172.20.6.12:21000 #制定config server集群
fork = true
EOF
mongos -f /data/mongodb/mongos/conf/mongos.conf #启动mongos服务
目前config server集群/shard集群/mongos服务均已启动,但此时为设置分片,还不能使用分片功能。需要登录mongos启用分片
登录任意一台mongos
mongo 172.20.6.10:20000
use admin
db.runCommand({addshard:"shard1/172.20.6.10:22001,172.20.6.11:22001,172.20.6.12:22001"})
db.runCommand({addshard:"shard2/172.20.6.10:22002,172.20.6.11:22002,172.20.6.12:22002"})
db.runCommand({addshard:"shard3/172.20.6.10:22003,172.20.6.11:22003,172.20.6.12:22003"})
查看集群
三、 测试
目前配置服务、路由服务、分片服务、副本集服务都已经串联起来了,此时进行数据插入,数据能够自动分片。连接在mongos上让指定的数据库、指定的集合分片生效。
注意:设置分片需要在admin数据库进行
use admin
db.runCommand( { enablesharding :"kaliarchdb"}); #开启kaliarch库分片功能
db.runCommand( { shardcollection : "kaliarchdb.table1",key : {_id:"hashed"} } ) #指定数据库里需要分片的集合tables和片键_id
设置kaliarchdb的 table1 表需要分片,根据 _id 自动分片到 shard1 ,shard2,shard3 上面去。
查看分片信息
测试插入数据
use kaliarchdb;
for (var i = 1; i <= 100000; i++) db.table1.save({_id:i,"test1":"testval1"});
查看分片情况:(省去部分信息)
db.table1.stats()
{
"sharded" : true,
"capped" : false,
"ns" : "kaliarchdb.table1",
"count" : 100000, #总count
"size" : 3800000,
"storageSize" : 1335296,
"totalIndexSize" : 4329472,
"indexSizes" : {
"_id_" : 1327104,
"_id_hashed" : 3002368
},
"avgObjSize" : 38,
"nindexes" : 2,
"nchunks" : 6,
"shards" : {
"shard1" : {
"ns" : "kaliarchdb.table1",
"size" : 1282690,
"count" : 33755, #shard1的count数
"avgObjSize" : 38,
"storageSize" : 450560,
"capped" : false,
......
"shard2" : {
"ns" : "kaliarchdb.table1",
"size" : 1259434,
"count" : 33143, #shard2的count数
"avgObjSize" : 38,
"storageSize" : 442368,
"capped" : false,
.......
"shard3" : {
"ns" : "kaliarchdb.table1",
"size" : 1257876,
"count" : 33102, #shard3的count数
"avgObjSize" : 38,
"storageSize" : 442368,
"capped" : false,
此时架构中的mongos,config server,shard集群均已经搭建部署完毕,在实际生成环境话需要对前端的mongos做高可用来提示整体高可用。