python开发prometheus exporter--用于hadoop-yarn监控_Python

首先写python的exporter需要知道prometheus提供4种类型metrics

分别是：counter, gauge, summary和histogram

我们需要的pip模块

代码思路实例

def push_yarn():

    # 监控zk_rm

    yarn_zkrmapproot()





    # 监控yarn任务信息

    yarn_appsinfo()



def run():

    start_http_server(8006)  # 8006端口启动

    while true:

        push_yarn()

        time.sleep(10)



if __name__ == '__main__':

    run()

push_yarn()为监控的数据数据

循环进行监控拿取数据进行监控

我们使用gauge实例

注意⚠️：gauge与counter类似，唯一不同的是gauge数值可以减少，常被用于温度、利用率等指标。

新增gauge实例

yarn_zkrmapproot_code = gauge('yarn_zkrmapproot', 'yarn_zkrmapproot_num', ['instance'])

started_time_gauge = gauge('yarn_started_time', 'started_time', ['application'])

launch_time_gauge = gauge('yarn_launch_time', 'launch_time', ['application'])

finished_time_gauge = gauge('yarn_finished_time', 'finished_time', ['application'])

memory_seconds_gauge = gauge('yarn_memory_seconds', 'memory_seconds', ['application'])

vcore_seconds_gauge = gauge('yarn_vcore_seconds', 'vcore_seconds', ['application'])

实现一下我们要监控的指标

# --------yarn-------- #####

def yarn_zkrmapproot():

    # 命令

    # 命令

    if kerberos_switch:

        command = f'''

            echo 'ls /rmstore/zkrmstateroot/rmapproot' | /opt/dtstack/dtbase/zookeeper/bin/zkcli.sh | grep application_ | awk -f , '{{print nf}}'

            '''

    else:

        command = f'''

                    export client_jvmflags="$client_jvmflags -djava.security.auth.login.config=/opt/dtstack/dtbase/zookeeper/conf/jaas.conf -djava.security.krb5.conf=/opt/dtstack/kerberos/kerberos_pkg/conf/krb5.conf -dzookeeper.server.principal=zookeeper/{hostname}@dtstack.com"

                    echo 'ls /rmstore/zkrmstateroot/rmapproot' | /opt/dtstack/dtbase/zookeeper/bin/zkcli.sh | grep application_ | awk -f , '{{print nf}}'

                    '''

    # 使用subprocess模块执行命令

    result = subprocess.getstatusoutput(command)  # (0, '455')

    if result[0] == 0:

        yarn_zkrmapproot_code.labels('yarn_' + hostname).set(result[1])





    else:

        print(f"failed to execute command: {command}")









def yarn_appsinfo():

    list_apps = []

    command = "yarn rmadmin -getservicestate rm1"

    apps_url = "http://{}/ws/v1/cluster/apps"

    rm_info = subprocess.getstatusoutput(command)

    if rm_info[0] == 0:

        if rm_info[1] == 'active':

            rm_host = yarn_rm1

        else:

            rm_host = yarn_rm2

    response = requests.get(url=apps_url.format(rm_host))

    html = response.text

    data = json.loads(html)

    for i in range(0, len(data['apps']['app'])):

        need_data = data['apps']['app']

        if need_data[i]['memoryseconds'] > 102400:  # 大于10g的任务

            list_apps.append([need_data[i]['id'],

                              need_data[i]['startedtime'],

                              need_data[i]['launchtime'],

                              need_data[i]['finishedtime'],

                              need_data[i]['memoryseconds'], need_data[i]['vcoreseconds']])

    sorted_lst = sorted(list_apps, key=lambda x: (x[4], x[5]))

    for list in sorted_lst:

        application = list[0]

        started_time = list[1]

        launch_time = list[2]

        finished_time = list[3]

        memory_seconds = list[4]

        vcore_seconds = list[5]

        started_time_gauge.labels(application=application).set(started_time)

        launch_time_gauge.labels(application=application).set(launch_time)

        finished_time_gauge.labels(application=application).set(finished_time)

        memory_seconds_gauge.labels(application=application).set(memory_seconds)

        vcore_seconds_gauge.labels(application=application).set(vcore_seconds)

其中yarn_zkrmapproot是检测znode数量的

yarn_appsinfo是检测大于10g的任务的

传到服务器启动这个exporter