OneCache-针对redis代理服务

OneCache-针对redis代理服务

OneCache是一个基于Redis协议的分布式缓存中间件，使用中间层将多个Redis实例组成一个分布式的缓存集群。并且引入Redis组的概念，可以将Redis的Master和Slave分到同一个组里，进行自动故障切换和流理分担。

1.下载OneCache

下载http://www.onexsoft.com/software/onecache_rhel5_linux64.tar.gz
完成后进行上传解压安装。

2.安装、修改配置文件

接下来我们先来了解下需要搭建后的实际效果。我们使用5个redis节点进行分布式的搭建，将5个redis节点分为4个组，其中group1中有2个redis节点一主一从，如下。

组	ip端口	主从
group1	172.21.3.195:7000	主
group1	172.21.3.195:7004	从
group2	172.21.3.195:7001	主
group3	172.21.3.195:7002	主
group4	172.21.3.195:7003	主

安装redis可以参考我之前的redis集群安装配置前面的第一小节中redis安装配置，在此就不做详细说明了。
需要注意各个节点的端口以及从节点配置主从

#从节点上172.21.3.195 7004上修改配置文件redis.conf中配置
slaveof 172.21.3.195 7000

修改onecache.xml配置文件

<onecache port="8221" thread_num="15" hash_value_max="80"  >
 <vip if_alias_name="eml:0" vip_address="172.21.3.195" enable="0"></vip>
 <group name="group1" hash_min="0" hash_max="20">
        <host host_name="host1" ip="172.21.3.195" port="7000" master="1"></host>
        <host host_name="host1" ip="172.21.3.195" port="7004" master="0"></host>
 </group>
 <group name="group2" hash_min="21" hash_max="40">
        <host host_name="host1" ip="172.21.3.195" port="7001" master="1"></host>
 </group>
 <group name="group3" hash_min="41" hash_max="60">
        <host host_name="host1" ip="172.21.3.195" port="7002" master="1"></host>
 </group>
 <group name="group4" hash_min="61" hash_max="79">
        <host host_name="host1" ip="172.21.3.195" port="7003" master="1"></host>
 </group>
</onecache>

3.启动服务

• 启动各个redis节点
  bash
# /usr/local/redis-0/bin/redis-server /usr/local/redis-0/conf/redis-7000.conf
# /usr/local/redis-1/bin/redis-server /usr/local/redis-1/conf/redis-7001.conf
# /usr/local/redis-2/bin/redis-server /usr/local/redis-2/conf/redis-7002.conf
# /usr/local/redis-3/bin/redis-server /usr/local/redis-3/conf/redis-7003.conf
# /usr/local/redis-4/bin/redis-server /usr/local/redis-4/conf/redis-7004.conf


• 启动OneCache

  
  # ./onecache.bin 
  
  [2017-07-12 09:26:59] Message: Create the thread pool(size=15)...
  [2017-07-12 09:26:59] Message: Thread pool created
  [2017-07-12 09:26:59] Message: Connect to redis server(172.21.3.195:7000)...
  [2017-07-12 09:26:59] Message: Connected. conn pool size=50
  [2017-07-12 09:26:59] Message: Connect to redis server(172.21.3.195:7004)...
  [2017-07-12 09:26:59] Message: Connected. conn pool size=50
  [2017-07-12 09:26:59] Message: Connect to redis server(172.21.3.195:7001)...
  [2017-07-12 09:26:59] Message: Connected. conn pool size=50
  [2017-07-12 09:26:59] Message: Connect to redis server(172.21.3.195:7002)...
  [2017-07-12 09:26:59] Message: Connected. conn pool size=50
  [2017-07-12 09:26:59] Message: Connect to redis server(172.21.3.195:7003)...
  [2017-07-12 09:26:59] Message: Connected. conn pool size=50
  [2017-07-12 09:26:59] Message: Start the OneCache on port 8221. PID=2680
  

  查看测试
  “`bash

  /usr/local/redis-0/bin/redis-cli -h 172.21.3.195 -p 8221

  172.21.3.195:8221> set abc 123
  OK
  172.21.3.195:8221> set 999 999
  OK
  172.21.3.195:8221> set qqq 999
  OK
  172.21.3.195:8221> status
  [OneCache]
  Version=4.0.0.0
  Port=8221
  StartTime=2017-7-12 09:26:59
  UpTime=00:06:33
  GroupCount=4

[Backends]
[GROUP] [IP] [PORT] [CONNPOOL] [MASTER] [ACTIVE] [REQUESTS] [RECV SIZE] [SEND SIZE] [SEND>1KB] [SEND>1MB] [COMMANDS]
group1 172.21.3.195 7000 50 Y Y 1 0.02KB 0.01KB 0 0 [GET]:1
group1 172.21.3.195 7004 50 N Y 0 0KB 0KB 0 0 NULL
group2 172.21.3.195 7001 50 Y Y 0 0KB 0KB 0 0 NULL
group3 172.21.3.195 7002 50 Y Y 1 0.02KB 0.01KB 0 0 [GET]:1
group4 172.21.3.195 7003 50 Y Y 0 0KB 0KB 0 0 NULL

[Clients]
[NUM] [IP] [CONNECTS] [REQUESTS] [RECV>1KB] [RECV>1MB] [LAST CONNECT] [COMMANDS]
1 172.21.3.195 0 3 0 0 1970-1-1 08:00:00 [GET]:2
“`
目前 OneCache 的默认策略是：master(s)负责写操作，slaver(s)负责读操作，采用均匀分配的方式去处理每个请求，所以也降低了单个节
点的负载率。OneCache 能识别非活动的节点，从而跳过该节点，并能够正确寻找下一个可用的节点，如无节点可用，OneCache 将会返回错
误的答复。

下面我们来测试下。先关闭group1中的主节点，在来从节点是否可用

# /usr/local/redis-0/bin/redis-cli -h 172.21.3.195 -p 7000  shutdown
#  /usr/local/redis-0/bin/redis-cli -h 172.21.3.195 -p 8221
 172.21.3.195:8221> status
[OneCache]
Version=4.0.0.0
Port=8221
StartTime=2017-7-12 09:26:59
UpTime=00:49:39
GroupCount=4

[Backends]
[GROUP]                [IP]    [PORT] [CONNPOOL] [MASTER] [ACTIVE]      [REQUESTS]        [RECV SIZE]       [SEND SIZE]    [SEND>1KB]  [SEND>1MB]   [COMMANDS]
group1         172.21.3.195      7000         50     Y       N                   6             0.15KB            0.05KB             0           0   [GET]:4    [SET]:2 
group1         172.21.3.195      7004         50     N       Y                   1             0.02KB            0.01KB             0           0   [GET]:1 
group2         172.21.3.195      7001         50     Y       Y                   0                0KB               0KB             0           0   NULL
group3         172.21.3.195      7002         50     Y       Y                   1             0.02KB            0.01KB             0           0   [GET]:1 
group4         172.21.3.195      7003         50     Y       Y                   1             0.03KB            0.00KB             0           0   [SET]:1 

[Clients]
[NUM]               [IP]   [CONNECTS]  [REQUESTS]  [RECV>1KB]  [RECV>1MB]      [LAST CONNECT]   [COMMANDS]
1           172.21.3.195            0          17           5           0   1970-1-1 08:00:00   [GET]:6    [SET]:4 
172.21.3.195:8221> get qqq
"12345"
172.21.3.195:8221> set qqq 111
OK

测试结果与预料的一样。