前言
副本集部署是对数据的冗余和增加读请求的处理能力，却不能提高写请求的处理能力；关键问题是随着数据增加，单机硬件配置会成为性能的瓶颈。而分片集群可以很好的解决这一问题，通过水平扩展来提升性能。分片部署依赖三个组件：mongos(路由)，config(配置服务)，shard（分片）
shard:每个分片存储被分片的部分数据，同时每个分片又可以部署成副本集
mongos:作为查询路由器，为客户端与分片集群之间通讯的提供访问接口
config server:配置服务器存储这个集群的元数据和配置信息
部署规划
1. 服务分布
3台机器：mongodbA 192.168.128.128，mongodbB 192.168.128.129，mongodbB 192.168.128.130
3个分片，分片和配置服务部署成副本集，如下图：
副本集部署时，当节点数超过3个时候，需要配置仲裁节点；这里为了演示节点分布，所以配置了仲裁节点。
在实际生产环境部署的时候，可以根据机器资源灵活配置。
2. 目录规划
mongos：
/data/mongo-cluster/mongos/log
config：
/data/mongo-cluster/config/data
/data/mongo-cluster/config/log
shard1：
/data/mongo-cluster/shard1/data
/data/mongo-cluster/shard1/log
shard2：
/data/mongo-cluster/shard2/data
/data/mongo-cluster/shard2/log
shard3：
/data/mongo-cluster/shard3/data
/data/mongo-cluster/shard3/log
3. 端口规划
mongos： 21000，22000，23000 config： 31000，32000，33000 shard1： 41010，41020，41030 shard2： 42010，42020，42030 shard3： 43010，43020，43030
准备工作
1. 下载安装包
参考网上教程
本文档安装版本：
db version v3.6.4
git version: d0181a711f7e7f39e60b5aeb1dc7097bf6ae5856
OpenSSL version: OpenSSL 1.0.2g 1 Mar 2016
allocator: tcmalloc
modules: enterprise
build environment:
distmod: ubuntu1604
distarch: x86_64
target_arch: x86_64
2. 部署目录
mkdir /data/mongo-cluster -p mkdir /data/mongo-cluster/mongos/log -p mkdir /data/mongo-cluster/shard1/data -p mkdir /data/mongo-cluster/shard1/log -p
mkdir /data/mongo-cluster/shard2/data -p
mkdir /data/mongo-cluster/shard2/log -p
mkdir /data/mongo-cluster/shard3/data -p
mkdir /data/mongo-cluster/shard3/log -p
mkdir /data/mongo-cluster/config/log -p
mkdir /data/mongo-cluster/config/data -p
mkdir /data/mongo-cluster/keyfile -p
3. 创建配置文件
cd /data/mongo-cluster
A. mongod 配置文件 mongo_node.conf
mongo_node.conf 作为mongod实例共享的配置文件，内容如下：
# mongod.conf
# for documentation of all options, see:
# http://docs.mongodb.org/manual/reference/configuration-options/
# Where and how to store data.
storage:
# dbPath: /var/lib/mongodb
journal:
enabled: true
engine: wiredTiger
# mmapv1:
# wiredTiger:
# where to write logging data.
systemLog:
destination: file
logAppend: true
# path: /var/log/mongodb/mongod.log
# network interfaces
#net:
# port: 27017
# bindIp: 127.0.0.1,192.168.147.128
# bindIpAll: true
# how the process runs
processManagement:
timeZoneInfo: /usr/share/zoneinfo
fork: true
security:
authorization: "enabled"
#operationProfiling:
#replication:
#sharding:
## Enterprise-Only Options:
#auditLog:
#snmp:
B. mongos 配置文件 mongos.conf
systemLog:
destination: file
logAppend: true
processManagement:
fork: true
4. 创建keyfile文件
cd /data/mongo-cluster mkdir keyfile
openssl rand -base64 756 > mongo.key chmod 400 mongo.key
集群搭建
mongoCluster.sh内容：
#!/bin/sh
WORK_DIR=/data/mongo-cluster
KEYFILE=$WORK_DIR/keyfile/mongo.key
CONFFILE=$WORK_DIR/mongo_node.conf
MONGOS_CONFFILE=$WORK_DIR/mongos.conf
MONGOD=/usr/bin/mongod
MONGOS=/usr/bin/mongos
mongos_start_cmd()
{
$MONGOS --port 21000 --bind_ip 127.0.0.1,192.168.128.130 --configdb configReplSet/192.168.128.130:31000,192.168.128.129:32000,192.168.128.128:33000 --keyFile $KEYFILE --pidfilepath $WORK_DIR/mongos.pid --logpath $WORK_DIR/mongos/log/mongo.log --config $MONGOS_CONFFILE
return $?
}
config_start_cmd()
{
$MONGOD --port 31000 --bind_ip 127.0.0.1,192.168.128.130 --configsvr --replSet configReplSet --keyFile $KEYFILE --dbpath $WORK_DIR/config/data --pidfilepath $WORK_DIR/mongo_config.pid --logpath $WORK_DIR/config/log/mongo.log --config $CONFFILE
return $?
}
shard_start_cmd()
{
$MONGOD --port 41010 --bind_ip 127.0.0.1,192.168.128.130 --shardsvr --replSet shard1 --keyFile $KEYFILE --dbpath $WORK_DIR/shard1/data --pidfilepath $WORK_DIR/mongo_shard1.pid --logpath $WORK_DIR/shard1/log/mongo.log --config $CONFFILE &&\
$MONGOD --port 42010 --bind_ip 127.0.0.1,192.168.128.130 --shardsvr --replSet shard2 --keyFile $KEYFILE --dbpath $WORK_DIR/shard2/data --pidfilepath $WORK_DIR/mongo_shard2.pid --logpath $WORK_DIR/shard2/log/mongo.log --config $CONFFILE &&\
$MONGOD --port 43010 --bind_ip 127.0.0.1,192.168.128.130 --shardsvr --replSet shard3 --keyFile $KEYFILE --dbpath $WORK_DIR/shard3/data --pidfilepath $WORK_DIR/mongo_shard3.pid --logpath $WORK_DIR/shard3/log/mongo.log --config $CONFFILE
}
mongos_server()
{
case $1 in
start)
mongos_start_cmd
;;
stop)
echo "stoping mongos server..."
cat mongos.pid | xargs kill -9
;;
restart)
echo "stoping mongos server..."
cat mongos.pid | xargs kill -9
echo "setup mongos server..."
mongos_start_cmd
;;
status)
pid=`cat mongos.pid`
if ps -ef | grep -v "grep" | grep $pid > /dev/null 2>&1; then
status="running"
else
status="stoped"
fi
echo "mongos server[$pid] status: ${status}"
;;
*)
echo "Usage: /etc/init.d/samba {start|stop|reload|restart|force-reload|status}"
;;
esac
}
config_server()
{
case $1 in
start)
config_start_cmd
;;
stop)
echo "stoping config server..."
cat mongo_config.pid | xargs kill -9
;;
restart)
echo "stoping config server..."
cat mongo_config.pid | xargs kill -9;
echo "setup config server..."
config_start_cmd
;;
status)
pid=`cat mongo_config.pid`
if ps -ef | grep -v "grep" | grep $pid > /dev/null 2>&1; then
status="running"
else
status="stoped"
fi
echo "config server[$pid] status: ${status}"
;;
*)
echo "Usage: `basename $0` config {start|stop|reload|restart|force-reload|status}"
;;
esac
}
shard_server()
{
case $1 in
start)
shard_start_cmd
;;
stop)
cat mongo_shard1.pid | xargs kill -9;
cat mongo_shard2.pid | xargs kill -9;
cat mongo_shard3.pid | xargs kill -9;
;;
restart)
cat mongo_shard1.pid | xargs kill -9;
cat mongo_shard2.pid | xargs kill -9;
cat mongo_shard3.pid | xargs kill -9;
shard_start_cmd
;;
status)
pid=`cat mongo_shard1.pid`
if ps -ef | grep -v "grep" | grep $pid > /dev/null 2>&1; then
status="running"
else
status="stoped"
fi
echo "shard1 server[$pid] status: ${status}"
#shard2
pid=`cat mongo_shard2.pid`
if ps -ef | grep -v "grep" | grep $pid > /dev/null; then
status="running"
else
status="stoped"
fi
echo "shard2 server[$pid] status: ${status}"
#shard3
pid=`cat mongo_shard3.pid`
if ps -ef | grep -v "grep" | grep $pid > /dev/null; then
status="running"
else
status="stoped"
fi
echo "shard3 server[$pid] status: ${status}"
;;
*)
echo "Usage: /etc/init.d/samba {start|stop|reload|restart|force-reload|status}"
;;
esac
}
all_server ()
{
case $1 in
start)
mongos_server start
config_server start
shard_server start
;;
stop)
mongos_server stop
config_server stop
shard_server stop
cat mongo_config.pid | xargs kill -9
cat mongo_shard1.pid | xargs kill -9
cat mongo_shard2.pid | xargs kill -9
cat mongo_shard3.pid | xargs kill -9
;;
restart)
mongos_server restart
config_server restart
shard_server restart
;;
status)
mongos_server status
config_server status
shard_server status
exit 0
;;
*)
echo "Usage: $0 {all|mongos|config|shard} {start|stop|reload|restart|force-reload|status}"
exit 1
;;
esac
}
#set -e
case $1 in
all)
all_server $2
;;
mongos)
mongos_server $2
;;
config)
config_server $2
;;
shard)
shard_server $2
;;
*)
echo "Usage: `basename $0` {all|mongos|config|shard} {start|stop|reload|restart|force-reload|status}"
;;
esac
对于不同主机，脚本中绑定的IP地址要做相应的修改
1. config集群部署
分别在mongodbA，mongodbB，mongodbC机器上，
执行./mongoCluster.sh config start
启动成功，会显示如下信息：
连接其中一个config服务，执行副本集初始化
mongo --port 31000 --host 127.0.0.1
>cfg={
_id:"configReplSet",
configsvr: true,
members:[
{_id:0, host:'192.168.128.128:33000'},
{_id:1, host:'192.168.128.129:32000'},
{_id:2, host:'192.168.128.130:31000'}
]};
>rs.initiate(cfg);
2. shard集群部署
分别在mongodbA，mongodbB，mongodbC机器上：
执行./mongoCluster.sh shard start
启动成功，会显示如下信息：
连接其中一个Shard进程，执行副本集初始化
mongo --port 41010 --host 127.0.0.1
>cfg={
_id:"shard1",
members:[
{_id:0, host:'192.168.128.128:41030'，arbiterOnly:true},
{_id:1, host:'192.168.128.129:41020'},
{_id:2, host:'192.168.128.130:41010'}
]};
rs.initiate(cfg);
同样shard2，shard3执行相同的操作：
shard2
cfg={
_id:"shard2",
members:[
{_id:0, host:'192.168.128.128:42030'},
{_id:1, host:'192.168.128.129:42020',arbiterOnly:true},
{_id:2, host:'192.168.128.130:42010'}
]};
rs.initiate(cfg);
shard3
cfg={
_id:"shard3",
members:[
{_id:0, host:'192.168.128.128:43030'},
{_id:1, host:'192.168.128.129:43020'},
{_id:2, host:'192.168.128.130:43010',arbiterOnly:true}
]};
rs.initiate(cfg);
3. mongos部署
分别在mongodbA，mongodbB，mongodbC机器上：
执行./mongoCluster.sh mongos start
启动成功，会显示如下信息：
接入其中一个mongos实例，执行添加分片操作：
./bin/mongo --port 21000 --host 127.0.0.1
>sh.addShard("shard1/192.168.128.130:41010")
{ "shardAdded" : "shard1", "ok" : 1 }
>sh.addShard("shard2/192.168.128.129:42010")
{ "shardAdded" : "shard2", "ok" : 1 }
>sh.addShard("shard2/192.168.128.128:43030")
{ "shardAdded" : "shard3", "ok" : 1 }
4. 初始化用户
初始化mongos用户
接入其中一个mongos实例，添加管理员用户
>use admin
>db.createUser({
user:'admin',pwd:'Admin@01',
roles:[
{role:'clusterAdmin',db:'admin'},
{role:'userAdminAnyDatabase',db:'admin'},
{role:'dbAdminAnyDatabase',db:'admin'},
{role:'readWriteAnyDatabase',db:'admin'}
]})
当前admin用户具有集群管理权限、所有数据库的操作权限。
需要注意的是，在第一次创建用户之后，localexception不再有效，接下来的所有操作要求先通过鉴权。
查看集群状态，sh.status()
初始化shard集群用户
分片集群中的访问都会通过mongos入口，而鉴权数据是存储在config副本集中的，即config实例中system.users数据库存储了集群用户及角色权限配置。mongos与shard实例则通过内部鉴权(keyfile机制)完成，因此shard实例上可以通过添加本地用户以方便操作管理。在一个副本集上，只需要在Primary节点上添加用户及权限，相关数据会自动同步到Secondary节点。
>use admin
>db.createUser({
user:'admin',pwd:'Admin123',
roles:[
{role:'clusterAdmin',db:'admin'},
{role:'userAdminAnyDatabase',db:'admin'},
{role:'dbAdminAnyDatabase',db:'admin'},
{role:'readWriteAnyDatabase',db:'admin'}
]})
数据割接
集合分片一般都是针对大数据量的集合；小集合分片没有必要进行分片，如果片键选择不当反而会影响查询效率
初始化数据分片
首先创建数据库用户uhome、为数据库实例uhome启动分片
use uhome
db.createUser({user:'uhome',pwd:'Uhome123',roles:[{role:'dbOwner',db:'uhome'}]})
sh.enableSharding("uhome")
以customer为例，
use uhome
db.createCollection("customer")
sh.shardCollection("uhome.
customer
", {community_id:"hashed"}, false)
导入数据：
mongoimport --port 21000 --db uhome --collection customer -u uhome -p Uhome123 --type csv --headerline --ignoreBlanks --file ./customer-mongo.csv
查看一下数据分布
db.customer.getShardDistribution()
可以看出来，分布还算均匀
参考：