监控redis cluster 主从实例是否切换，切换前后对应关系

需求：编写脚本实现对redis cluster 主从是否发生主从切换进行监控，若发生切换，输出切换前后主从对应关系。

初始化对比文件,仅执行一次，后续不需要在执行，除非该文件不存在。

cmd="redis-cli -c -h $1 -p $2"
$cmd cluster nodes > cluster_nodes.txt
cat cluster_nodes.txt | awk -F[ @] '/master/{print $1,$2}' | sort -nk2 -t ':'  > master.txt
cat cluster_nodes.txt | awk -F[ @] '/slave/{print $2,$5}' | sort -nk1 -t ':'  > slave.txt
cat master.txt | while read m
do
        master_ip_port=`echo $m | awk '{print $2}'`
        master_tag=`echo $m | awk '{print $1}'`
        cat slave.txt | while read s
        do
                slave_ip_port=`echo $s | awk '{print $1}'`
                slave_tag=`echo $s | awk '{print $2}'`
                if [[ ${master_tag} == ${slave_tag} ]];then
                　　echo "${master_ip_port}->${slave_ip_port}" >> master_slave_map_old
                fi
        done
done

判断是否发生切换，如果发生，输出切换前后的对应关系

cmd="redis-cli -c -h $1 -p $2"
$cmd cluster nodes > cluster_nodes.txt
cat cluster_nodes.txt | awk -F[ @] '/master/{print $1,$2}' | sort -nk2 -t ':'  > master.txt
cat cluster_nodes.txt | awk -F[ @] '/slave/{print $2,$5}' | sort -nk1 -t ':'  > slave.txt
cat master.txt | while read m
do
        master_ip_port=`echo $m | awk '{print $2}'`
        master_tag=`echo $m | awk '{print $1}'`
        cat slave.txt | while read s
        do
                slave_ip_port=`echo $s | awk '{print $1}'`
                slave_tag=`echo $s | awk '{print $2}'`
                if [[ ${master_tag} == ${slave_tag} ]];then
                　　echo "${master_ip_port}->${slave_ip_port}" >> master_slave_map_new
                fi
        done
done


ports=($($cmd cluster nodes | awk -F[ :@] '{print $3}' | sort |uniq))
len=${#ports[@]}

for ((i=0;i<=$len-1;i++))
do
    old=$(grep ${ports[i]} master_slave_map_old | md5sum | awk '{print $1}')
    new=$(grep ${ports[i]} master_slave_map_new | md5sum | awk '{print $1}')
    if [[ "${old}" != "${new}" ]];then
        echo -n "${ports[i]}主从实例切换前："
        grep ${ports[i]} master_slave_map_old
        echo -n "${ports[i]}主从实例切换后："
        grep ${ports[i]} master_slave_map_new
    fi
done

cat master_slave_map_new > master_slave_map_old
rm -rf master_slave_map_new

测试过程：

5.5.5.101:29001> info replication
# Replication
role:slave
master_host:5.5.5.102
master_port:29001
master_link_status:up
master_last_io_seconds_ago:0
master_sync_in_progress:0
slave_repl_offset:360775
slave_priority:100
slave_read_only:1
connected_slaves:0
master_replid:3c06fcc274b6fac6fe1d954f8b63687619e65cf5
master_replid2:0000000000000000000000000000000000000000
master_repl_offset:360775
second_repl_offset:-1
repl_backlog_active:1
repl_backlog_size:52428800
repl_backlog_first_byte_offset:1
repl_backlog_histlen:360775
lxd-vm1:29001> CLUSTER FAILOVER
OK
5.5.5.101:29001> info replication
# Replication
role:master
connected_slaves:1
slave0:ip=5.5.5.102,port=29001,state=online,offset=361741,lag=1
master_replid:a35374725759383211f8f76d6b9376d181beda17
master_replid2:3c06fcc274b6fac6fe1d954f8b63687619e65cf5
master_repl_offset:361755
second_repl_offset:360888
repl_backlog_active:1
repl_backlog_size:52428800
repl_backlog_first_byte_offset:1
repl_backlog_histlen:361755
lxd-vm1:29001> 

测试结果

[redis@lxd-vm3 ~]$ sh a2.sh 5.5.5.101 29001
29001主从实例切换前：5.5.5.102:29001->5.5.5.101:29001
29001主从实例切换后：5.5.5.101:29001->5.5.5.102:29001
[redis@lxd-vm3 ~]$