用ansible-playbook安装redis的sentinel高可用集群
redis的高可用集群方案一般来说分为下面几种, redis集群 redis cluster http://www.redis.io/topics/cluster-tutorial , redis哨兵 redis sentinel http://www.redis.io/topics/sentinel keepalived+lvs+redis实现 codis集群 https://github.com/CodisLabs/codis
redis sentinel是redis自带的解决方法，轻量并简单，配置方法是: 1，先做一个主从 2，sentinel监控redis主从，如果发现主挂了，提升redis从为主，下面用ansible安装redis
# egrep -v "^$|^#" /etc/ansible/hosts
[server]
10.211.55.4 host_name=vm01
10.211.55.5 host_name=vm02
10.211.55.6 host_name=vm03
# cat /root/ansible/hosts
127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4
::1 localhost localhost.localdomain localhost6 localhost6.localdomain6
10.211.55.4 vm01
10.211.55.5 vm02
10.211.55.6 vm03
安装redis的ansible-playbook,获取ipv4 address,更改主机的hostname，解压安装redis，sed修改redis配置，最后启动redis服务
#jinja2: variable_start_string: "[%" , variable_end_string: "%]"
---
- hosts: server
remote_user: root
vars:
hostname: '[% host_name %]'
ipv4: ''
tasks:
- name: time sync
command: ntpdate ntp1.aliyun.com
- name: copy hosts file
copy: src=/root/ansible/hosts dest=/etc/hosts owner=root group=root mode=0644 force='yes'
- name: set hostname
shell: 'hostname '
register: host_name_change
- debug: msg=''
- name: close iptables
service: name=iptables enabled=yes state=stopped
- name: copy redis.tar.gz
copy: src=/root/ansible/redis-3.2.0.tar.gz dest=/tmp/ owner=root group=root mode=0644
- name: install redis
shell: cd /tmp/; tar -xzf /tmp/redis-3.2.0.tar.gz; cd /tmp/redis-3.2.0; make -j4; make install
- name: copy redis.conf to /etc dir
copy: src=/tmp/redis-3.2.0/redis.conf dest=/etc/ owner=root group=root mode=0644 force='yes'
- name: copy redis init file to /etc/init.d/
copy: src=/root/ansible/redis_init.sh dest=/etc/init.d/redis owner=root group=root mode=0755 force='yes'
- name: mkdir /var/redis
shell: mkdir -p /var/redis
- name: change redis config
shell: sed -i 's/daemonize\ no/daemonize\ yes/g' /etc/redis.conf ; sed -i '/bind 127.0.0.1/d' /etc/redis.conf;echo bind 127.0.0.1 >> /etc/redis.conf; sed -i 's/dir\ .\//dir\ \/var\/redis/g' /etc/redis.conf; sed -i 's/appendonly\ no/appendonly\ yes/g' /etc/redis.conf
- name: start redis service
service: name=redis enabled=yes state=restarted
- name: redis status
shell: echo info |redis-cli |head -n 19
register: result
- debug: msg=
redis的启动文件https://github.com/thinkeverytime/devops2u.com/blob/master/redis_init.sh
redis已经安装完成并启动了，下面vm01为redis主，vm02为redis从，vm03redis停止，作为sentinel
vm02的/etc/redis.conf 最后添加一行：slaveof 10.211.55.4 6379，并重启，作为vm01的从，从info replication可以获取主从关系的信息,也可以 role 命令查看master、slave、sentinel各个角色的信息
root@vm02 ~]# redis-cli -h 10.211.55.5
10.211.55.5:6379> info replication
# Replication
role:slave
master_host:10.211.55.4
master_port:6379
master_link_status:up
master_last_io_seconds_ago:5
master_sync_in_progress:0
slave_repl_offset:29
slave_priority:100
slave_read_only:1
connected_slaves:0
master_repl_offset:0
repl_backlog_active:0
repl_backlog_size:1048576
repl_backlog_first_byte_offset:0
repl_backlog_histlen:0
vm03 不用redis服务，启用sentinel
sentinel主要的配置就是下面一行，mymaster为哨兵的名称，后面是redis主的IP，最后面的“1”是因为现在monitor只有1台
sentinel monitor mymaster 10.211.55.4 6379 1
root@vm03 ~]# service redis stop
Stopping redis-server: [ OK ]
[root@vm03 ~]# cp /tmp/redis-3.2.0/sentinel.conf /etc/
[root@vm03 ~]# egrep -v "^$|^#" /etc/sentinel.conf
port 26379
dir /tmp
sentinel monitor mymaster 10.211.55.4 6379 1
sentinel down-after-milliseconds mymaster 30000
sentinel parallel-syncs mymaster 1
sentinel failover-timeout mymaster 180000
[root@vm03 ~]# redis-sentinel /etc/sentinel.conf &
[1] 20262
[root@vm03 ~]# _._
_.-``__ ''-._
_.-`` `. `_. ''-._ Redis 3.2.0 (00000000/0) 64 bit
.-`` .-```. ```\/ _.,_ ''-._
( ' , .-` | `, ) Running in sentinel mode
|`-._`-...-` __...-.``-._|'` _.-'| Port: 26379
| `-._ `._ / _.-' | PID: 20262
`-._ `-._ `-./ _.-' _.-'
|`-._`-._ `-.__.-' _.-'_.-'|
| `-._`-._ _.-'_.-' | http://redis.io
`-._ `-._`-.__.-'_.-' _.-'
|`-._`-._ `-.__.-' _.-'_.-'|
| `-._`-._ _.-'_.-' |
`-._ `-._`-.__.-'_.-' _.-'
`-._ `-.__.-' _.-'
`-._ _.-'
`-.__.-'
20262:X 11 May 15:21:18.437 # Sentinel ID is ea87ab5547e731d11cd95359f414475458e44f87
20262:X 11 May 15:21:18.437 # +monitor master mymaster 10.211.55.4 6379 quorum 1
20262:X 11 May 15:21:18.438 * +slave slave 10.211.55.5:6379 10.211.55.5 6379 @ mymaster 10.211.55.4 6379
在redis主随便设置些数据，从肯定是有的，也可以查看/var/redis/redis.log
# redis-cli
127.0.0.1:6379> set website devops2u.com
OK
127.0.0.1:6379> get website
"devops2u.com"
这时，把vm01的redis主停止，查看redis从和sentinel的变化
redis从状态的转变：
[root@vm02 ~]# redis-cli -h 10.211.55.5
10.211.55.5:6379> keys *
1) "website"
10.211.55.5:6379> info replication
# Replication
role:slave
master_host:10.211.55.4
master_port:6379
master_link_status:down
master_last_io_seconds_ago:-1
master_sync_in_progress:0
slave_repl_offset:26781
master_link_down_since_seconds:8
slave_priority:100
slave_read_only:1
connected_slaves:0
master_repl_offset:0
repl_backlog_active:0
repl_backlog_size:1048576
repl_backlog_first_byte_offset:0
repl_backlog_histlen:0
10.211.55.5:6379> info replication
# Replication
role:master
connected_slaves:0
master_repl_offset:0
repl_backlog_active:0
repl_backlog_size:1048576
repl_backlog_first_byte_offset:0
repl_backlog_histlen:0
redis sentinel的日志如下：
20262:X 11 May 15:28:17.289 # +sdown master mymaster 10.211.55.4 6379
20262:X 11 May 15:28:17.289 # +odown master mymaster 10.211.55.4 6379 #quorum 1/1
20262:X 11 May 15:28:17.289 # +new-epoch 1
20262:X 11 May 15:28:17.289 # +try-failover master mymaster 10.211.55.4 6379
20262:X 11 May 15:28:17.304 # +vote-for-leader ea87ab5547e731d11cd95359f414475458e44f87 1
20262:X 11 May 15:28:17.304 # +elected-leader master mymaster 10.211.55.4 6379
20262:X 11 May 15:28:17.304 # +failover-state-select-slave master mymaster 10.211.55.4 6379
20262:X 11 May 15:28:17.388 # +selected-slave slave 10.211.55.5:6379 10.211.55.5 6379 @ mymaster 10.211.55.4 6379
20262:X 11 May 15:28:17.388 * +failover-state-send-slaveof-noone slave 10.211.55.5:6379 10.211.55.5 6379 @ mymaster 10.211.55.4 6379
20262:X 11 May 15:28:17.455 * +failover-state-wait-promotion slave 10.211.55.5:6379 10.211.55.5 6379 @ mymaster 10.211.55.4 6379
20262:X 11 May 15:28:18.333 # +promoted-slave slave 10.211.55.5:6379 10.211.55.5 6379 @ mymaster 10.211.55.4 6379
20262:X 11 May 15:28:18.333 # +failover-state-reconf-slaves master mymaster 10.211.55.4 6379
20262:X 11 May 15:28:18.419 # +failover-end master mymaster 10.211.55.4 6379
20262:X 11 May 15:28:18.420 # +switch-master mymaster 10.211.55.4 6379 10.211.55.5 6379
20262:X 11 May 15:28:18.420 * +slave slave 10.211.55.4:6379 10.211.55.4 6379 @ mymaster 10.211.55.5 6379
20262:X 11 May 15:28:48.473 # +sdown slave 10.211.55.4:6379 10.211.55.4 6379 @ mymaster 10.211.55.5 6379
在转换过程中，或许会提示 READONLY You can’t write against a read only instance，是因为只读的redis实例不能写入数据
redis集群结构已经发生变化，想要修复的话下一步怎么办呢， 1，把vm01的redis配置最后添加 slaveof 10.211.55.5 6379 变为此时redis主的vm02 2，启动vm01的redis，info replication 查看状态
redis-sentinel查看状态:
info sentinel
client list
现在的结果是vm02为redis主，vm01为redis从，sentinel一样做一个守护神
程序因为一直连接的redis sentinel的ip和port，所以不会出现问题。
如果有多个redis集群，可以用sentinel相互做高可用，sentinel既能相互依存又节省主机，如下图：
sentinel_1/sentinel_2 sentinel_2/sentinel_1
| |
————————————————— ———————————————————
| \ / |
| X |
| / \ |
redis1主—————–——–—— —————————————–—–redis2主
| |
redis1从 redis2从
2016年05月11日 于 linux工匠 发表