Metrics-Python操作Metrics Server API 获取node、pod资源率

一、原理说明

• 其实目前新版的K8S在监控这块的架构已经非常明确了，只不过国内很少有文章解释这一块，其官方架构说明见：
  https://github.com/kubernetes/community/blob/master/contributors/design-proposals/instrumentation/monitoring_architecture.md。

• 其架构分位2个部分：

  • 内置于K8S的核心指标采集，安装K8S就自带了（下图黑色部分）。
  • 第三方的监控采集方案，需要大家自己选型，比如prometheus等（下图蓝色部分）。

• 像kubectl top获取的cpu/mem使用情况，就属于K8S内置的核心指标采集而来，完全不需要第三方的支持
  

二、API 访问 metrics-server

• 那么当前，metrics-server在API SERVER注册的GROUP叫做metrics.k8s.io，VERSION是v1beta1，所以其对应的Restful资源URI就是以：/apis/metrics.k8s.io/v1beta1/为前缀的。

• metrics-server官方已经说明了其资源URI的几种形式：https://github.com/kubernetes/community/blob/master/contributors/design-proposals/instrumentation/resource-metrics-api.md。

  • The list of supported endpoints:

    • /nodes – all node metrics; type []NodeMetrics
    • /nodes/{node} – metrics for a specified node; type NodeMetrics
    • /namespaces/{namespace}/pods – all pod metrics within namespace with support for all-namespaces; type []PodMetrics
    • /namespaces/{namespace}/pods/{pod} – metrics for a specified pod; type PodMetrics

  • The following query parameters are supported:

    • labelSelector – restrict the list of returned objects by labels (list endpoints only)

• 所以，为了获取某个namespace下面的pods的资源利用率，我们可以有2种方式：

  • 按标签筛选：/apis/metrics.k8s.io/v1beta1/namespaces/{namespace}/pods?labelSelector=xxxx
  • 按pod名字获取：/apis/metrics.k8s.io/v1beta1/namespaces/{namespace}/pods/{pod}

• 首先启动一个proxy，它会帮我们解决和API SERVER之间的认证问题，我们只需要关注于接口参数即可：

# kubectl proxy --port=8181 --address=0.0.0.0
# nodes -> curl localhost:8181/apis/metrics.k8s.io/v1beta1/nodes
# pods -> curl localhost:8181/apis/metrics.k8s.io/v1beta1/namespaces/default/pods/redis-master-dasdhjasd
          curl localhost:8181/apis/metrics.k8s.io/v1beta1/namespaces/test/pods

• 通过接口访问节点资源信息路径

查看node资源指标
        # kubectl get --raw "/apis/metrics.k8s.io/v1beta1/nodes" | jq  | less
查看pods资源指标
        # kubectl get --raw "/apis/metrics.k8s.io/v1beta1/pods" | jq  | less

三、requests封装API

# 仅为示例，代码不完整！！！
import requests
import re

def eventer_alarm_cutpods(string):
    string = re.sub(r‘\\n‘, ‘\n‘, string)
    pods = string.split(‘\n‘)[2].split()[1]
    print(pods)
    return pods

def metric_get_pods(host, pods):
    """
    :param hosts -> 127.0.0.1:8181
    """
    if pods:
        url = ‘http://‘ + host  + ‘/apis/metrics.k8s.io/v1beta1/namespaces/test/pods‘
        response = requests.get(url)
        with response:
            print(response.json())

    else:
        url = ‘http://‘ + host + ‘/apis/metrics.k8s.io/v1beta1/pods‘

def metric_get_nodes(host, node):
    """
    :param hosts -> 127.0.0.1:8181
    :return:
    """
    if node:
        pass
    else:
        url = ‘http://‘ + host + ‘/apis/metrics.k8s.io/v1beta1/nodes‘
        response = requests.get(url)
        with response:
            print(response.json())





if __name__ == ‘__main__‘:
    # metric_get_nodes(‘127.0.0.1:8181‘)
    string = ‘EventType: Warning\nEventKind: Pod\nEventObject: test-nginx-7d97ffc85d-nrnfn\nEventReason: Failed\nEventTime: 2021-05-23 17:48:12 +0800 CST\nEventMessage: Error: ErrImagePull‘
    eventer_alarm_cutpods(string)

原文：https://www.cnblogs.com/dai-zhe/p/14805691.html