概述

生产中为了避免单点故障，生产环境中redis升级为集群模式，需要对redis集群进行监控，一旦有节点出现故障便触发报警。Redis有自带的redis-cli客户端，通过cluster info命令能查询到集群的运行情况，我们可以写个shell脚本，通过zabbix来调用这个脚本实现集群的监控。

一、cluster info命令的使用

命令格式：

redis-cli -h [hostname] -p [port] -a [password] cluster info

1、查询集群运行状态，1为正常，0为故障

redis-cli -h xxx.xxx.xxx.xxx -a 'password' -p 7001 cluster info | grep "cluster_state"

2、redis集群已分配的槽

redis-cli -h xxx.xxx.xxx.xxx -a 'password' -p 7001 cluster info | grep -w "cluster_slots_assigned" | awk -F':' '{print $2}'

3、redis集群槽的状态是ok的数目

redis-cli -h xxx.xxx.xxx.xxx -a 'password' -p 7001 cluster info | grep -w "cluster_slots_ok" | awk -F':' '{print $2}'

4、redis集群可能失效的槽数目

redis-cli -h xxx.xxx.xxx.xxx -a 'password' -p 7001 cluster info | grep -w "cluster_slots_pfail" | awk -F':' '{print $2}'

5、redis集群已经失效的槽数目

redis-cli -h xxx.xxx.xxx.xxx -a 'password' -p 7001 cluster info | grep -w "cluster_slots_fail" | awk -F':' '{print $2}'

6、redis集群中所有节点个数

redis-cli -h xxx.xxx.xxx.xxx -a 'password' -p 7001 cluster info | grep -w "cluster_known_nodes" | awk -F':' '{print $2}'

7、redis集群中主节点个数

redis-cli -h xxx.xxx.xxx.xxx -a 'password' -p 7001 cluster info | grep -w "cluster_size" | awk -F':' '{print $2}'

8、redis集群中的currentEpoch值

(集群中的currentEpoch总是一致的,currentEpoch越高，代表节点的配置或者操作越新,集群中最大的那个node epoch

当集群的状态发生改变，某个节点为了执行一些动作需要寻求其他节点的同意时，就会增加 currentEpoch 的值)

redis-cli -h xxx.xxx.xxx.xxx -a 'password' -p 7001 cluster info | grep -w "cluster_current_epoch" | awk -F':' '{print $2}'

9、redis集群中二进制总线接收的消息数量

redis-cli -h xxx.xxx.xxx.xxx -a 'password' -p 7001 cluster info | grep -w "cluster_stats_messages_sent" | awk -F':' '{print $2}'

10、redis集群中二进制总线发送的消息数量

redis-cli -h xxx.xxx.xxx.xxx -a 'password' -p 7001 cluster info | grep -w "cluster_stats_messages_received" | awk -F':' '{print $2}'

11、redis端口存活监控

redis-cli -h xxx.xxx.xxx.xxx -a 'password' -p 7001 ping | grep -c PONG

二、创建监控脚本

(1) 、 监控redis 端口存活的脚本

[root@admin conf]# vim redis-port.sh

#!/bin/bash
#根据实际情况修改
REDISCLI="/usr/local/bin/redis-cli"
HOST="192.168.2.14"
port1=7001
port2=7002
port3=7003
port4=7004
port5=7005
port6=7006
if [[ $# == 1 ]];then
case $1 in
7001)
result=`$REDISCLI -h $HOST -p $port1 ping 2>/dev/null | grep -c PONG`
echo $result
;;
7002)
result=`$REDISCLI -h $HOST -p $port2 ping 2>/dev/null | grep -c PONG`
echo $result
;;
7003)
result=`$REDISCLI -h $HOST -p $port3 ping 2>/dev/null | grep -c PONG`
echo $result
;;
7004)
result=`$REDISCLI -h $HOST -p $port4 ping 2>/dev/null | grep -c PONG`
echo $result
;;
7005)
result=`$REDISCLI -h $HOST -p $port5 ping 2>/dev/null | grep -c PONG`
echo $result
;;
7006)
result=`$REDISCLI -h $HOST -p $port6 ping 2>/dev/null | grep -c PONG`
echo $result
;;
*)
echo -e "\033[33mUsage: $0 {7001|7002|7003}\033[0m"
;;
esac
fi

(2) 、 监控redis-cluster 集群的脚本

[root@admin conf]# vim redis-cluster.sh

#!/bin/bash
#根据实际情况修改
REDISCLI="/usr/local/bin/redis-cli"
HOST="192.168.2.14"
PORT=`ps -ef | grep -w redis-server | grep -v "grep" | awk '{print $9}' | awk -F ':' '{print $2}' | head -1`

if [[ $# == 1 ]];then
case $1 in
cluster_state)
result=`$REDISCLI -h $HOST -p $PORT cluster info 2>/dev/null | grep -w "cluster_state" | awk -F':' '{print $2}' | grep -c "ok"`
echo $result
;;
cluster_slots_assigned)
result=`$REDISCLI -h $HOST -p $PORT cluster info 2>/dev/null | grep -w "cluster_slots_assigned" | awk -F':' '{print $2}'`
echo $result
;;
cluster_slots_ok)
result=`$REDISCLI -h $HOST -p $PORT cluster info 2>/dev/null | grep -w "cluster_slots_ok" | awk -F':' '{print $2}'`
echo $result
;;
cluster_slots_pfail)
result=`$REDISCLI -h $HOST -p $PORT cluster info 2>/dev/null | grep -w "cluster_slots_pfail" | awk -F':' '{print $2}'`
echo $result
;;
cluster_slots_fail)
result=`$REDISCLI -h $HOST -p $PORT cluster info 2>/dev/null | grep -w "cluster_slots_fail" | awk -F':' '{print $2}'`
echo $result
;;
cluster_known_nodes)
result=`$REDISCLI -h $HOST -p $PORT cluster info 2>/dev/null | grep -w "cluster_known_nodes" | awk -F':' '{print $2}'`
echo $result
;;
cluster_size)
result=`$REDISCLI -h $HOST -p $PORT cluster info 2>/dev/null | grep -w "cluster_size" | awk -F':' '{print $2}'`
echo $result
;;
cluster_current_epoch)
result=`$REDISCLI -h $HOST -p $PORT cluster info 2>/dev/null | grep -w "cluster_current_epoch" | awk -F':' '{print $2}'`
echo $result
;;
cluster_stats_messages_sent)
result=`$REDISCLI -h $HOST -p $PORT cluster info 2>/dev/null | grep -w "cluster_stats_messages_sent" | awk -F':' '{print $2}'`
echo $result
;;
cluster_stats_messages_received)
result=`$REDISCLI -h $HOST -p $PORT cluster info 2>/dev/null | grep -w "cluster_stats_messages_received" | awk -F':' '{print $2}'`
echo $result
;;
*)
echo -e "\033[33mUsage: $0 {cluster_state|cluster_slots_assigned|cluster_slots_ok|cluster_slots_pfail|cluster_slots_fail|cluster_known_nodes|cluster size|cluster current epoch|cluster_stats_messages_sent|cluster_stats_messages_received}\033[0m"
;;
esac
fi

（3）赋予脚本可执行权限

[root@admin conf]# chmod +x redis-cluster.sh
[root@admin conf]# chmod +x redis-port.sh

( 4 ) 脚本测试

查看redis集群状态：

[root@admin conf]# ./redis-cluster.sh cluster_state
1

查看redis 7001端口状态：

[root@admin conf]# ./redis-port.sh 7001
1

三、创建redis集群监控配置文件

1、编写监控配置文件

[root@admin conf]# vi zabbix_agentd.conf

#添加两行自定义KEY

# redis-cluster 集群监控

UserParameter=Redis.Cluster[*],/home/zabbix/conf/redis-cluster.sh $1

# redis-port 端口监控

UserParameter=Redis.Port[*],/home/zabbix/conf/redis-port.sh $1

2、重启zabbix-agent

[root@admin ~]# /etc/init.d/zabbix-agent restart

3、登录zabbix server服务端机器上测试

[root@zabbix ~]# zabbix_get -s 192.168.2.14 -p 10050 -k "Redis.Cluster[cluster_state]"
1
[root@zabbix ~]# zabbix_get -s 192.168.2.14 -p 10050 -k "Redis.Port[7001]"
1

四、zabbix上创建redis监控模板

1、创建redis-cluster模板名称

2、创建监控项

例如：（根据自身情况决定）

名称：redis集群运行状态监测

类型： zabbix 客户端

键值： Redis.Cluster[cluster_state]

信息类型： 数字（无正负）

更新间隔: 30s

创建redis端口监控

3、创建触发器

例如：（根据自身情况决定）

名称：redis集群状态Failed

严重性： 一般严重

表达式：{Redis-cluster Service:Redis.Cluster[cluster_state].count(#3,1,"ne")}>2

五、监控效果图