Zabbix自定义key监控docker

一、需求场景
在某些docker容器中运行着java，容器提供API供监控调用。需要监控服务运行是否正常和容器内JVM使用情况。

二、想法
编写监控脚本，使用自定义key的方式添加到Zabbix中。

三、实践
1、编写服务监控脚本

#!/bin/bash
interval_time=600
service_port_file=/script/tmp/service_port.txt

#Gener_service_port
tee > ${service_port_file} << EOF
apigate=8603
config=8602
eureka=8601
file=8606
schedule=8677
service=8605
EOF

#Accept input
container_id=$($(which docker) ps|grep "$1"|awk ‘{print $1}‘)
service_name=$(echo $1 | awk -F ‘.‘ ‘{print $1}‘|awk -F ‘-‘ ‘{print $NF}‘)
service_port=$(grep ${service_name} ${service_port_file}|awk -F ‘=‘ ‘{print $NF}‘)

#Get created
created_time=$(docker inspect ${container_id} --format ‘{{.Created}}‘)
created_timestamp=$(date +%s -d ${created_time})
system_timestamp=$(date +%s)
interval=$((system_timestamp-created_timestamp))

#Get container version
container_version=$(docker exec ${container_id} [ -f "/etc/redhat-release" ] && grep -oE  "[0-9.]+" /etc/redhat-release|awk -F . ‘{print $1}‘ || echo 0)

#Get ip
docker_gwbridge=$(ifconfig docker_gwbridge|awk ‘NR==2‘|awk ‘{print $2}‘|awk -F . ‘{print $1"."$2"."$3}‘)
if [ ${container_version} -eq 7 ];then
    service_ip=$($(which docker) exec ${container_id} hostname -I | grep -o ${docker_gwbridge}.*|sed s/[[:space:]]//g)
else
    service_ip=$($(which docker) exec ${container_id} ifconfig | grep ${docker_gwbridge}|awk ‘{print $2}‘|awk -F ‘:‘ ‘{print $2}‘)
fi

#Get status_code
status_code=$(curl -I -m 10 -o /dev/null -s -w %{http_code} ${service_ip}:${service_port}/info)

#Return value
if [ ${status_code} -eq 200 ];then
    echo "1"
else
    if [ ${interval} -gt ${interval_time} ];then
        echo "0"
    else
        awk ‘BEGIN{printf "%.2f\n",(1-‘${interval}‘/‘${interval_time}‘)}‘
    fi
fi

说明：脚本接受容器名字作为参数，访问预先商议好对应服务的API。返回200则表示正常，将输出1。若返回0，则判断容器的创建时间，如果是创建时间小于约定的时间，说明容器是新生成的，返回对应代表进度的数值，否则返回0。

2、编写监控JVM的脚本

#!/bin/bash
pid=1

container_id=$($(which docker) ps|grep "$1"|awk ‘{print $1}‘)

function Survivor0 {
    $(which docker) exec ${container_id} jstat -gcutil ${pid} | awk ‘NR==2 {print $1}‘
}
function Survivor1 {
    $(which docker) exec ${container_id} jstat -gcutil ${pid} | awk ‘NR==2 {print $2}‘
}
function Eden {
    $(which docker) exec ${container_id} jstat -gcutil ${pid} | awk ‘NR==2 {print $3}‘
}
function Old {
    $(which docker) exec ${container_id} jstat -gcutil ${pid} | awk ‘NR==2 {print $4}‘
}
function Perm {
    $(which docker) exec ${container_id} jstat -gcutil ${pid} | awk ‘NR==2 {print $5}‘
}
function Metaspace {
    $(which docker) exec ${container_id} jstat -gcutil ${pid} | awk ‘NR==2 {print $5}‘
} 

"$2"

说明：脚本接受2个参数。第一个参数为容器名，第二个参数为监控项目名。

3、编写自动发现脚本

#!/bin/bash
key=cloud
#Get app_name
array=($($(which docker) ps |awk ‘{print $NF}‘|grep ${key}|awk -F ‘[_.]‘ ‘{print $2 "." $3}‘))

printf ‘[‘
for ((i=0;i<${#array[@]};i++))
do
        printf ‘{‘
        num=$(echo $((${#array[@]}-1)))
        if [ "$i" == ${num} ];
        then
                printf "\"{#APP_NAME}\":\"${array[$i]}\"}"
        else
                printf "\"{#APP_NAME}\":\"${array[$i]}\"},"
        fi
done
printf "]"

说明：以特定的关键字过滤需要监控的容器名字，按照规则输出json格式的字串。

4、在Agent的配置文件中加入自定义key

UserParameter=app.status[*],sudo /usr/bin/bash /script/appstatus.sh $1
UserParameter=docker.java[*],sudo /usr/bin/bash /script/dockerjava.sh $1 $2
UserParameter=app.discovery,sudo /usr/bin/bash /script/appdiscovery.sh

5、在Zabbix创建自动发现规则
(1)填写键值

(2)创建监控原型

(3)创建触发器

(4)创建图形原型