ansible-playbook实战之部署redis+sentinel+twemproxy

简介

twemproxy，也叫nutcraker。是一个twtter开源的一个redis 和memcache 快速/轻量级代理服务器；Twemproxy是一个快速的单线程代理程序，支持Memcached ASCII协议和更新的Redis协议
Twemproxy 通过引入一个代理层，可以将其后端的多台 Redis 或 Memcached 实例进行统一管理与分配，使应用程序只需要在 Twemproxy 上进行操作，而不用关心后面具体有多少个真实的 Redis 或 Memcached 存储。
本次使用redis+sentinel+twemproxy实现如下功能：
1.前端使用twemproxy （主备节点）做代理，将其后端的多台Redis实例分片进行统一管理与分配；
2.每一个分片节点的redis slave 都是redis master的副本且只读；
3.redis sentinel 持续不断的监控每个分片节点的master，当master出现故障且不可用状态时，sentinel 会通知/启动自动故障转移等动作；
4.sentinel 可以在发生故障转移动作后触发相应脚本（通过 client-reconfig-script 参数配置 ），脚本获取到最新的master来修改 twemproxy 配置并重启 twemproxy；

架构图

集群名 角色 IP 端口 组件 redis1 master 10.20.10.43 6379 redis+sentinel slave 10.20.10.44 6380 redis2 master 10.20.10.44 6379 redis+sentinel slave 10.20.10.45 6380 redis3 master 10.20.10.45 6379 redis+sentinel slave 10.20.10.43 6380 代理 proxy1 10.20.10.46 22121 twemproxy+sentinel proxy2 10.20.10.47 22121 proxy3 10.20.10.48 22121

如上图所示：
1.在43/44/45上分别安装redis+sentinel，并且3台机器之间两两做主从，以保证任何一台机器出问题不会影响整个集群。
2.在46/47/48上分别安装twemproxy+sentinel，当3组redis主从中，sentinle会监控每组主从中的master，当master失效sentinel会进行failover，并且sentinel会通过 client-reconfig-script 来触发client-reconfig.sh，将twemproxy代理的群组切换为新的master并重启，以使twemproxy对外正常提供服务。
此处由于在每台服务器上同时存在主从、proxy与redis隔离，为便于后续自动化安装，我们统一定义master的端口为6379、slave的端口为6380，proxy上的sentinel单独安装。

安装思路

1.变量定义
我们将需要安装的服务器分为3个角色：master、slave、proxy，在master和slave上安装redis+sentinel；在proxy上安装sentinel+twemproxy；由于master、slave和proxy可能使用应用用户启动而非root用户启动，因此我们将安装路径、启动用户及其他都提前定义到变量中；
2.安装脚本
我们使用redis_install.sh在master和slave上安装redis、sentinel并启动服务；在proxy安装sentinel并启动服务；
我们使用twemproxy.sh在proxy上安装twemproxy，并启动应用；另我们还使用此脚本设置sentinel的 client-reconfig-script 来设置切换脚本client-reconfig.sh

playbook的目录结构

├── hosts├── redis_cluster.yml└── roles    └── redis_cluster_install        ├── files        │   └── redis        │       ├── autoconf-2.69.tar.gz        │       ├── automake-1.15.tar.gz        │       ├── libtool-2.4.6.tar.gz        │       ├── redis-3.2.9.tar.gz        │       ├── redis.conf        │       ├── redis.conf.bak        │       ├── sentinel.conf        │       ├── sentinel.conf.bak        │       └── twemproxy-master.zip        ├── handlers        ├── meta        ├── tasks        │   └── main.yml        ├── templates        │   ├── client-reconfig.sh        │   ├── redis_install.sh        │   └── twemproxy_install.sh        └── vars            └── main.yml

说明：
1.file中的autoconf-2.69.tar.gz、automake-1.15.tar.gz、libtool-2.4.6.tar.gz、twemproxy-master.zip是安装twemproxy的源码文件；redis.conf、sentinel.conf是我们提前定义好的模板配置文件，以便我们进行变量替换；
2.template中的redis_install.sh是在master、slave、proxy上分别安装redis主从、sentinle的脚本；client-reconfig.sh是sentinel中client-reconfig-script 设置的切换脚本；twemproxy_install.sh是twemproxy的安装脚本

具体操作

1.创建资产hosts

vim hosts[redis_master]10.20.10.43 ansible_ssh_user=root ansible_ssh_pass=root123410.20.10.44 ansible_ssh_user=root ansible_ssh_pass=root123410.20.10.45 ansible_ssh_user=root ansible_ssh_pass=root1234[redis_slave1]10.20.10.43 ansible_ssh_user=root ansible_ssh_pass=root1234[redis_slave2]10.20.10.44 ansible_ssh_user=root ansible_ssh_pass=root1234[redis_slave3]10.20.10.45 ansible_ssh_user=root ansible_ssh_pass=root1234[redis_proxy]10.20.10.46 ansible_ssh_user=root ansible_ssh_pass=root1234[redis_proxy]10.20.10.47 ansible_ssh_user=root ansible_ssh_pass=root1234[redis_proxy]10.20.10.48 ansible_ssh_user=root ansible_ssh_pass=root1234

注意：此处的slave1、slave、slave3没有特别含义，可随意指定；具体的对应关系我们是在下面的角色文件中定义。
2.创建角色文件

vim redis_cluster.yml#角色为master- hosts: redis_master  remote_user: root  gather_facts: False  roles:    - {role: redis_cluster_install,redis_role: master}#角色为slave，隶属集群3- hosts: redis_slave1  remote_user: root  gather_facts: False  roles:    - {role: redis_cluster_install,redis_role: slave,cluster_no: 3}#角色为slave，隶属集群1- hosts: redis_slave2  remote_user: root  gather_facts: False  roles:    - {role: redis_cluster_install,redis_role: slave,cluster_no: 1}#角色为slave，隶属集群2- hosts: redis_slave3  remote_user: root  gather_facts: False  roles:    - {role: redis_cluster_install,redis_role: slave,cluster_no: 2}#角色为proxy- hosts: redis_proxy  remote_user: root  gather_facts: False  roles:    - {role: redis_cluster_install,redis_role: proxy}

其中：
slave1的ip为10.20.10.43，我们根据架构图中的表格定义属于数据集群redis3，因此cluster_no为3
slave1的ip为10.20.10.44，我们根据架构图中的表格定义属于数据集群redis1，因此cluster_no为1
slave1的ip为10.20.10.45，我们根据架构图中的表格定义属于数据集群redis3，因此cluster_no为2
3.创建变量文件

vim roles/redis_cluster/vars/main.yml#master和slave上的启动用户及安装目录、源码目录;app为应用用户家目录#redis server dirredis_user: appinstall_dir: /home/ap/appsource_dir: /home/ap/app/src/#proxy上的启动用户及安装目录、源码目录；appwb为应用用户家目录#proxy server dirproxy_user: appwbproxy_install_dir: /home/ap/appwbproxy_source_dir: /home/ap/appwb/src/#redis主从的端口redis_master_port: 6379redis_slave_port: 6380maxmemory: 2gb#sentinel的端口及quorum数量sen_port: 26379sen_quorum: 3#集群列表cluster1:- masterip: 10.20.10.43  sen_mastername: redis1cluster2:- masterip: 10.20.10.44  sen_mastername: redis2cluster3:- masterip: 10.20.10.45  sen_mastername: redis3#twemproxy端口tw_port: 22121

如下我们根据角色文件中的cluster_no，分别列举了3组集群的名称和主机ip，以用于twemproxy

4.创建任务文件

#在master和slave上拷贝一份安装源码用于安装redis，在此以master为主- name: copy redis dir to client  copy: src=redis dest={{source_dir}} owner=root group=root  when: redis_role == "master"#在proxy上贝一份安装源码用于安装redis- name: copy redis dir to client  copy: src=redis dest={{proxy_source_dir}} owner=root group=root  when: redis_role == "proxy"#在主或从上拷贝redis_install.sh- name: copy redis_install script to client  template: src=redis_install.sh dest={{source_dir}}/redis/ owner=root group=root mode=0775  when: redis_role == "master" or redis_role == "slave"#在proxy上拷贝redis_install.sh- name: copy redis_install script to client  template: src=redis_install.sh dest={{proxy_source_dir}}/redis/ owner=root group=root mode=0775  when: redis_role == "proxy"#在主或从上安装redis_install.sh- name: install redis and sentinel  shell: bash {{source_dir}}/redis/redis_install.sh  when: redis_role == "master" or redis_role == "slave"#在proxy上安装redis_install.sh- name: install redis and sentinel  shell: bash {{proxy_source_dir}}/redis/redis_install.sh  when: redis_role == "proxy"#在proxy上拷贝client-reconfig.sh- name: copy client-reconfig script to client  template: src=client-reconfig.sh dest={{proxy_install_dir}}/redis/ owner={{proxy_user}} group={{proxy_user}} mode=0775  when: redis_role == "proxy"#在proxy上拷贝twemproxy_install.sh- name: copy twemproxy_install script to client  template: src=twemproxy_install.sh dest={{proxy_source_dir}}/redis/ owner=root group=root mode=0775  when: redis_role == "proxy"#在proxy上安装twemproxy_install.sh- name: install twemproxy  shell: bash {{proxy_source_dir}}/redis/twemproxy_install.sh owner={{proxy_user}} group={{proxy_user}} mode=0775  when: redis_role == "proxy"

注意：我们为什么在每个task中都要指定相应的节点角色？
因为我们在redis_install.sh中是根据相应的master、slave、proxy角色来安装的，虽然redis_install.sh的名字一样，但是由于角色的不同，从ansible服务器端推送至客户端的redis_install.sh的脚本内容也是不同的。因此我们根据不同的角色推送，实现在同一个脚本中实现不同的角色安装配置。

5.创建模板脚本

（1）vim templates/redis_install.sh

#!/bin/bash#author: yanggd#content: install redis and sentinel#redis目录相关source_dir={{source_dir}}install_dir={{install_dir}}redis_user={{redis_user}}#proxy目录相关proxy_source_dir={{proxy_source_dir}}proxy_install_dir={{proxy_install_dir}}proxy_user={{proxy_user}}#redis内存maxmemory={{maxmemory}}#redis主从端口redis_master_port={{redis_master_port}}redis_slave_port={{redis_slave_port}}#sentinel端口及quorum数量sen_port={{sen_port}}sen_quorum={{sen_quorum}}#在master上安装redis{% if redis_role == "master" %}yum install make gcc gcc-c++ -yid $redis_user &> /dev/nullif [ $? -ne 0 ];then        useradd -d $install_dir $redis_userfi#install rediscd $source_dir/redistar -zxf redis-3.2.9.tar.gzcd redis-3.2.9make MALLOC=libcmake PREFIX=$install_dir/redis install#init redis dirmkdir $install_dir/redis/{data,conf,logs}{% endif %}#在proxy上安装redis{% if redis_role == "proxy" %}yum install make gcc gcc-c++ -yid $proxy_user &> /dev/nullif [ $? -ne 0 ];then        useradd -d $proxy_install_dir $proxy_userfi#install rediscd $proxy_source_dir/redistar -zxf redis-3.2.9.tar.gzcd redis-3.2.9make MALLOC=libcmake PREFIX=$proxy_install_dir/redis install#init redis dirmkdir $proxy_install_dir/redis/{data,conf,logs}#change install_dir ownerchown -R $proxy_user.$proxy_user $proxy_install_dir{% endif %}#获取监听的ip地址ip=`ifconfig eth1|grep "inet addr"|awk '{print $2}'|awk -F: '{print $2}'`#根据模板配置文件替换变量#modify redis confcd $install_dir/redis/conf#配置master配置文件{% if redis_role == "master" %}cp $source_dir/redis/redis.conf redis_$redis_master_port.confsed -i "s:bind:bind $ip:g" redis_$redis_master_port.confsed -i "s:port:port $redis_master_port:g" redis_$redis_master_port.confsed -i "s:pidfile:pidfile '$install_dir/redis/redis_$redis_master_port.pid':g" redis_$redis_master_port.confsed -i "s:logfile:logfile '$install_dir/redis/logs/redis_$redis_master_port.log':g" redis_$redis_master_port.confsed -i "s:dump:redis_$redis_master_port:g" redis_$redis_master_port.confsed -i "s:dir:dir '$install_dir/redis/data/':g" redis_$redis_master_port.confsed -i "s:memsize:$maxmemory:g" redis_$redis_master_port.conf#change install_dir ownerchown -R $redis_user.$redis_user $install_dir#modify kernelcat > /etc/sysctl.conf << EOFnet.core.somaxconn = 10240vm.overcommit_memory = 1EOFsysctl -pecho never > /sys/kernel/mm/transparent_hugepage/enabledecho "echo never > /sys/kernel/mm/transparent_hugepage/enabled" >> /etc/rc.local#启动master#start redis master su $redis_user -c "$install_dir/redis/bin/redis-server $install_dir/redis/conf/redis_$redis_master_port.conf"{% endif %}#配置slave配置文件{% if redis_role == "slave" %}cp $source_dir/redis/redis.conf redis_$redis_slave_port.confsed -i "s:bind:bind $ip:g" redis_$redis_slave_port.confsed -i "s:port:port $redis_slave_port:g" redis_$redis_slave_port.confsed -i "s:pidfile:pidfile '$install_dir/redis/redis_$redis_slave_port.pid':g" redis_$redis_slave_port.confsed -i "s:logfile:logfile '$install_dir/redis/logs/redis_$redis_slave_port.log':g" redis_$redis_slave_port.confsed -i "s:dump:redis_$redis_slave_port:g" redis_$redis_slave_port.confsed -i "s:dir:dir '$install_dir/redis/data/':g" redis_$redis_slave_port.confsed -i "s:memsize:$maxmemory:g" redis_$redis_slave_port.conf#根据角色中cluster_no找到对应的master{% if cluster_no == 1 %}{% for i in cluster1 %}echo "slaveof {{i.masterip}} $redis_master_port" >> redis_$redis_slave_port.conf{% endfor %}{% endif %}{% if cluster_no == 2 %}{% for i in cluster2 %}echo "slaveof {{i.masterip}} $redis_master_port" >> redis_$redis_slave_port.conf{% endfor %}{% endif %}{% if cluster_no == 3 %}{% for i in cluster3 %}echo "slaveof {{i.masterip}} $redis_master_port" >> redis_$redis_slave_port.conf{% endfor %}{% endif %}#change install_dir ownerchown -R $redis_user.$redis_user $install_dir#启动slave服务#start redis slavesu $redis_user -c "$install_dir/redis/bin/redis-server $install_dir/redis/conf/redis_$redis_slave_port.conf"{% endif %}#配置哨兵监控3组主从#modify sentinel conf{% if redis_role == "master" %}cd $install_dir/redis/confcp $source_dir/redis/sentinel.conf sentinel.conf{%for i in cluster1 %}sed -i "s:sen_port:$sen_port:g" sentinel.confsed -i "s:pidfile:pidfile '$install_dir/redis/sentinel.pid':g" sentinel.confsed -i "s:dir:dir '$install_dir/redis/data/':g" sentinel.confsed -i "s:logfile:logfile '$install_dir/redis/logs/sentinel.log':g" sentinel.confsed -i "s:master1_name:{{i.sen_mastername}}:g" sentinel.confsed -i "s:master1_host:{{i.masterip}}:g" sentinel.confsed -i "s:master_port:$redis_master_port:g" sentinel.confsed -i "s:quorum:$sen_quorum:g" sentinel.conf{% endfor %}{%for i in cluster2 %}sed -i "s:master2_name:{{i.sen_mastername}}:g" sentinel.confsed -i "s:master2_host:{{i.masterip}}:g" sentinel.confsed -i "s:master_port:$redis_master_port:g" sentinel.confsed -i "s:quorum:$sen_quorum:g" sentinel.conf{% endfor %}{%for i in cluster3 %}sed -i "s:master3_name:{{i.sen_mastername}}:g" sentinel.confsed -i "s:master3_host:{{i.masterip}}:g" sentinel.confsed -i "s:master_port:$redis_master_port:g" sentinel.confsed -i "s:quorum:$sen_quorum:g" sentinel.conf{% endfor %}#change install_dir ownerchown -R $redis_user.$redis_user $install_dir#启动哨兵服务#start sentinelsu $redis_user -c "$install_dir/redis/bin/redis-sentinel $install_dir/redis/conf/sentinel.conf"{% endif %}#在proxy上配置哨兵{% if redis_role == "proxy" %}cd $proxy_install_dir/redis/confcp $proxy_source_dir/redis/sentinel.conf sentinel.conf{%for i in cluster1 %}sed -i "s:sen_port:$sen_port:g" sentinel.confsed -i "s:pidfile:pidfile '$proxy_install_dir/redis/sentinel.pid':g" sentinel.confsed -i "s:dir:dir '$proxy_install_dir/redis/data/':g" sentinel.confsed -i "s:logfile:logfile '$proxy_install_dir/redis/logs/sentinel.log':g" sentinel.confsed -i "s:master1_name:{{i.sen_mastername}}:g" sentinel.confsed -i "s:master1_host:{{i.masterip}}:g" sentinel.confsed -i "s:master_port:$redis_master_port:g" sentinel.confsed -i "s:quorum:$sen_quorum:g" sentinel.conf{% endfor %}{%for i in cluster2 %}sed -i "s:master2_name:{{i.sen_mastername}}:g" sentinel.confsed -i "s:master2_host:{{i.masterip}}:g" sentinel.confsed -i "s:master_port:$redis_master_port:g" sentinel.confsed -i "s:quorum:$sen_quorum:g" sentinel.conf{% endfor %}{%for i in cluster3 %}sed -i "s:master3_name:{{i.sen_mastername}}:g" sentinel.confsed -i "s:master3_host:{{i.masterip}}:g" sentinel.confsed -i "s:master_port:$redis_master_port:g" sentinel.confsed -i "s:quorum:$sen_quorum:g" sentinel.conf{% endfor %}#change install_dir ownerchown -R $proxy_user.$proxy_user $proxy_install_dir#在proxy上启动哨兵su $proxy_user -c "$proxy_install_dir/redis/bin/redis-sentinel $proxy_install_dir/redis/conf/sentinel.conf"{% endif %}

（2）vim client-reconfig.sh

#!/bin/bash#content: modify twemproxy when sentinel failover#<master-name> <role> <state> <from-ip> <from-port> <to-ip> <to-port>proxy_install_dir={{proxy_install_dir}}monitor_name="$1"master_old_ip="$4"master_old_port="$5"master_new_ip="$6"master_new_port="$7"tw_bin=$proxy_install_dir/twemproxy/sbin/nutcrackertw_conf=$proxy_install_dir/twemproxy/conf/nutcracker.ymltw_log=$proxy_install_dir/twemproxy/logs/twemproxy.logtw_cmd="$tw_bin -c $tw_conf -o $tw_log -v 11 -d"#modify twemproxy confsed -i "s/${master_old_ip}:${master_old_port}/${master_new_ip}:${master_new_port}/g" $tw_conf#kill twemproxyps -ef|grep twemproxy|grep -v grep|awk '{print $2}'|xargs kill#start twemproxy$tw_cmdsleep 1ps -ef|grep twemproxy|grep -v grep

（3）vim twemproxy_install.sh

#!/bin/bash#author: yanggd#content: install twemproxy and add sentinel client-reconfig.sh#在proxy上安装twemproxyproxy_source_dir={{proxy_source_dir}}proxy_install_dir={{proxy_install_dir}}sen_port={{sen_port}}redis_master_port={{redis_master_port}}#redisproxy_user={{proxy_user}}redis_master_port={{redis_master_port}}#twemproxytw_port={{tw_port}}#install twemproxycd $proxy_source_dir/redistar -zxf autoconf-2.69.tar.gzcd autoconf-2.69./configuremake && make installcd ..tar -zxf automake-1.15.tar.gzcd automake-1.15./configuremake && make installcd ..tar -zxf libtool-2.4.6.tar.gzcd libtool-2.4.6./configuremake && make installcd ..unzip twemproxy-master.zipcd twemproxy-masteraclocalautoreconf -f -i -Wall,no-obsolete./configure --prefix=$proxy_install_dir/twemproxymake && make install#init twemproxymkdir -p $proxy_install_dir/twemproxy/{conf,logs}ip=`ifconfig eth1|grep "inet addr"|awk '{print $2}'|awk -F: '{print $2}'`#生成twemproxy的配置文件cat >> $proxy_install_dir/twemproxy/conf/nutcracker.yml <<EOFredis_proxy:  listen: $ip:$tw_port  hash: fnv1a_64  distribution: ketama  auto_eject_hosts: true  redis: true  server_retry_timeout: 2000  server_failure_limit: 3  servers:   - masterip1:redis_master_port:1 sen_mastername1   - masterip2:redis_master_port:1 sen_mastername2   - masterip3:redis_master_port:1 sen_mastername3EOF#配置文件中加入代理的3组集群的master、端口及集群名{% for i in cluster1 %}sed -i "s#masterip1#{{i.masterip}}#g" $proxy_install_dir/twemproxy/conf/nutcracker.ymlsed -i "s#redis_master_port#$redis_master_port#g" $proxy_install_dir/twemproxy/conf/nutcracker.ymlsed -i "s#sen_mastername1#${{i.sen_mastername}}#g" $proxy_install_dir/twemproxy/conf/nutcracker.yml{% endfor %}{% for i in cluster2 %}sed -i "s#masterip2#{{i.masterip}}#g" $proxy_install_dir/twemproxy/conf/nutcracker.ymlsed -i "s#sen_mastername2#${{i.sen_mastername}}#g" $proxy_install_dir/twemproxy/conf/nutcracker.yml{% endfor %}{% for i in cluster3 %}sed -i "s#masterip3#{{i.masterip}}#g" $proxy_install_dir/twemproxy/conf/nutcracker.ymlsed -i "s#sen_mastername3#${{i.sen_mastername}}#g" $proxy_install_dir/twemproxy/conf/nutcracker.yml{% endfor %}#change twemproxy ownerchown -R $proxy_user.$proxy_user $proxy_install_dir#start twemproxysu $proxy_user -c "$proxy_install_dir/twemproxy/sbin/nutcracker -c $proxy_install_dir/twemproxy/conf/nutcracker.yml -o $proxy_install_dir/twemproxy/logs/twemproxy.log -v 11 -d"#mv client-reconfig.shmv $proxy_install_dir/redis/client-reconfig.sh $proxy_install_dir/twemproxy/client-reconfig.sh#配置sentinel的client-reconfig-script#config sentinel$proxy_install_dir/redis/bin/redis-cli -h $ip -p $sen_port sentinel set ccbredis1 client-reconfig-script $proxy_install_dir/twemproxy/client-reconfig.sh$proxy_install_dir/redis/bin/redis-cli -h $ip -p $sen_port sentinel set ccbredis2 client-reconfig-script $proxy_install_dir/twemproxy/client-reconfig.sh$proxy_install_dir/redis/bin/redis-cli -h $ip -p $sen_port sentinel set ccbredis3 client-reconfig-script $proxy_install_dir/twemproxy/client-reconfig.sh

最后的sentinel的client-reconfig-script也可以在sentinel.conf中配置，如：

sentinel client-reconfig-script  redis1 $proxy_install_dir/twemproxy/client-reconfig.shsentinel client-reconfig-script  redis2 $proxy_install_dir/twemproxy/client-reconfig.shsentinel client-reconfig-script  redis3 $proxy_install_dir/twemproxy/client-reconfig.sh

6.安装

#检查语法ansible-playbook -C redis_cluster.yml#安装ansible-playbook redis_cluster.yml

安装完成后，整个集群也就随之启动了，我们可以在各个节点上检查配置，在此我们就不赘述了。

7.安装排错
ansible-playbook在安装过程中可能会报错中断，如果遇到这种问题，我们可以在相应的节点上的的安装目录查看相应的脚本，来帮助我们排查，如：

cd /home/ap/app/src/vim redis_install.sh

此时的redis_install.sh就是当前节点使用的安装脚本。

总结

此次自动化部署实现了3组redis主从和twemproxy的组合，我们可以通过变量的形式根据不同的环境要求进行安装，降低了我们手动部署的出错率。另外我们可以通过此次部署延伸出一主两从+haproxy+sentinel等各个组合方式。更重要的是通过此次部署，我们能够将其运用到其他环境的自动化部署的场景中。

最后是redis和sentinel的模板配置文件：
(1)redis.conf

bind 127.0.0.1protected-mode yesporttcp-backlog 511timeout 0tcp-keepalive 300daemonize yessupervised nopidfileloglevel warninglogfiledatabases 16#save 900 1#save 300 10#save 60 10000stop-writes-on-bgsave-error yesrdbcompression yesrdbchecksum yesdbfilename "dump.rdb"dirslave-serve-stale-data yesslave-read-only yesrepl-diskless-sync norepl-diskless-sync-delay 5repl-disable-tcp-nodelay noslave-priority 100appendonly noappendfilename "appendonly.aof"appendfsync everysecno-appendfsync-on-rewrite noauto-aof-rewrite-percentage 100auto-aof-rewrite-min-size 64mbaof-load-truncated yeslua-time-limit 5000slowlog-log-slower-than 10000slowlog-max-len 128latency-monitor-threshold 0notify-keyspace-events ""hash-max-ziplist-entries 512hash-max-ziplist-value 64list-max-ziplist-size -2list-compress-depth 0set-max-intset-entries 512zset-max-ziplist-entries 128zset-max-ziplist-value 64hll-sparse-max-bytes 3000activerehashing yesclient-output-buffer-limit normal 0 0 0client-output-buffer-limit slave 256mb 64mb 60client-output-buffer-limit pubsub 32mb 8mb 60hz 10aof-rewrite-incremental-fsync yesmaxclients 4096maxmemory memsizemaxmemory-policy allkeys-lru

(2)sentinel.conf

bind 0.0.0.0daemonize yesport sen_portloglevel noticepidfiledirlogfilesentinel monitor master1_name master1_host master_port quorumsentinel down-after-milliseconds master1_name 6000sentinel failover-timeout master1_name 18000sentinel parallel-syncs master1_name 1sentinel monitor master2_name master2_host master_port quorumsentinel down-after-milliseconds master2_name 6000sentinel failover-timeout master2_name 18000sentinel parallel-syncs master2_name 1sentinel monitor master3_name master3_host master_port quorumsentinel down-after-milliseconds master3_name 6000sentinel failover-timeout master3_name 18000sentinel parallel-syncs master3_name 1