k8s 弹性伸缩,基于prometheus自定义指标

2023年 7月 16日 65.4k 0

简介

上一批文章写了,基于CPU指标的弹性伸缩,资源指标只包含CPU、内存,一般来说也够了。但如果想根据自定义指标:如请求qps/5xx错误数来实现HPA,就需要使用自定义指标了,目前比较成熟的实现是 Prometheus Custom Metrics。自定义指标由Prometheus来提供,再利用k8s-prometheus-adpater聚合到apiserver,实现和核心指标(metric-server)同样的效果。

图片 1

下面我们就来演示基于prometheus监控自定义指标实现k8s pod基于qps的弹性伸缩

这里已经部署好prometheus环境

准备扩容应用

部署一个应用,且应用需要允许被prometheus采集数据

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: metrics-app
  name: metrics-app
spec:
  replicas: 3
  selector:
    matchLabels:
      app: metrics-app
  template:
    metadata:
      labels:
        app: metrics-app
      annotations:		
        prometheus.io/scrape: "true"			# 设置允许被prometheus采集
        prometheus.io/port: "80"					# prometheus 采集的端口
        prometheus.io/path: "/metrics"			# prometheus 采集的路径
    spec:
      containers:
      - image: metrics-app
        name: metrics-app
        ports:
        - name: web
          containerPort: 80
        resources:
          requests:
            cpu: 200m
            memory: 256Mi
        readinessProbe:
          httpGet:
            path: /
            port: 80
          initialDelaySeconds: 3
          periodSeconds: 5
        livenessProbe:
          httpGet:
            path: /
            port: 80
          initialDelaySeconds: 3
          periodSeconds: 5
---
apiVersion: v1
kind: Service
metadata:
  name: metrics-app
  labels:
    app: metrics-app
spec:
  ports:
  - name: web
    port: 80
    targetPort: 80
  selector:
    app: metrics-app

应用部署之后,通过prometheus web端验证是否已经自动发现

image005

验证指标能否正常采集

image006

 

这样我们部署的扩容应用就准备完成了,接下来的内容就是HPA如何获取其qps进行扩容的配置了

部署 Custom Metrics Adapter

prometheus采集到的metrics并不能直接给k8s用,因为两者数据格式不兼容,还需要另外一个组件(k8s-prometheus-adpater),将prometheus的metrics 数据格式转换成k8s API接口能识别的格式,转换以后,因为是自定义API,所以还需要用Kubernetes aggregator在主APIServer中注册,以便直接通过/apis/来访问。

prometheus-adapter GitHub地址:https://github.com/DirectXMan12/k8s-prometheus-adapter

该 PrometheusAdapter 有一个稳定的Helm Charts,我们直接使用,这里使用helm 3.0版本,使用微软云的镜像

先准备下helm环境(如已有可忽略):

wget https://get.helm.sh/helm-v3.0.0-linux-amd64.tar.gz
tar zxvf helm-v3.0.0-linux-amd64.tar.gz 
mv linux-amd64/helm /usr/bin/
helm repo add stable http://mirror.azure.cn/kubernetes/charts
helm repo update
helm repo list

部署prometheus-adapter,指定prometheus地址:

# helm install prometheus-adapter stable/prometheus-adapter --namespace kube-system --set prometheus.url=http://prometheus.kube-system,prometheus.port=9090
# helm list -n kube-system

验证部署成功

# kubectl get pods -n kube-system
NAME                                  READY   STATUS    RESTARTS   AGE
prometheus-adapter-86574f7ff4-t9px4   1/1     Running   0          65s

确保适配器注册到APIServer:

# kubectl get apiservices |grep custom 
# kubectl get --raw "/apis/custom.metrics.k8s.io/v1beta1"

创建HPA策略

我们配置的扩容规则策略为每秒QPS超过0.8个就进行扩容

[root@k8s-master-02 ~]# cat app-hpa-v2.yml 
apiVersion: autoscaling/v2beta2
kind: HorizontalPodAutoscaler
metadata:
  name: metrics-app-hpa 
  namespace: default
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: metrics-app
  minReplicas: 1
  maxReplicas: 10
  metrics:
  - type: Pods
    pods:
      metric:
        name: http_requests_per_second
      target:
        type: AverageValue
        averageValue: 800m		# 800m 即0.8个/秒,如果是阀值设置为每秒10个,这里的值就应该填写10000m
[root@k8s-master-02 ~]# kubectl apply -f app-hpa-v2.yml 

配置prometheus自定义指标

当创建好HPA还没数据,因为适配器还不知道你要什么指标(http_requests_per_second),HPA也就获取不到Pod提供指标,接下来我们就要解决监控值没有正常获取的问题,即配置自定义指标

[root@k8s-master-02 ~]# kubectl get hpa
NAME              REFERENCE                TARGETS          MINPODS   MAXPODS   REPLICAS   AGE
metrics-app-hpa   Deployment/metrics-app   /800m   1         10        3          36s

ConfigMap在default名称空间中编辑prometheus-adapter ,并seriesQuery在该rules: 部分的顶部添加一个新的配置:(获取该服务的所有pod 2分钟之内的http_requests监控值并计算平均值,然后相加对外提供数据)

[root@k8s-master-02 ~]# kubectl edit cm prometheus-adapter -n kube-system
apiVersion: v1
data:
  config.yaml: |
    rules:
    - seriesQuery: 'http_requests_total{kubernetes_namespace!="",kubernetes_pod_name!=""}'
      resources:
        overrides:
          kubernetes_namespace: {resource: "namespace"}
          kubernetes_pod_name: {resource: "pod"}
      name:
        matches: "^(.*)_total"
        as: "${1}_per_second"
      metricsQuery: 'sum(rate({}[2m])) by ()'
……

配置好之后,因为adapter pod不支持配置动态加载,所以我们修改了配置,需要删除一下pod,让他重新加载一个新的生效配置

[root@k8s-master-02 ~]# kubectl delete pod prometheus-adapter-86574f7ff4-t9px4 -n kube-system

删除pod重新生成配置之后,大约一两分钟在观察hpa的值就正常了

[root@k8s-master-02 ~]# kubectl get hpa
NAME              REFERENCE                TARGETS     MINPODS   MAXPODS   REPLICAS   AGE
metrics-app-hpa   Deployment/metrics-app   416m/800m   1         10        2          17m

验证基于自定义指标的扩容

接下来通过压测验证扩容缩容

[root@k8s-master-02 ~]# kubectl get svc
NAME          TYPE        CLUSTER-IP   EXTERNAL-IP   PORT(S)   AGE
kubernetes    ClusterIP   10.0.0.1             443/TCP   84d
metrics-app   ClusterIP   10.0.0.80            80/TCP    3d2h
[root@k8s-master-02 ~]# ab -n 100000 -c 100  http://10.0.0.80/metrics

压测过程中观察hpa和pod状态发现pod已经自动扩容到了10个pod

[root@k8s-master-02 ~]# kubectl get hpa
NAME              REFERENCE                TARGETS        MINPODS   MAXPODS   REPLICAS   AGE
metrics-app-hpa   Deployment/metrics-app   289950m/800m   1         10        10         38m
[root@k8s-master-02 ~]# kubectl get pods
NAME                                      READY   STATUS    RESTARTS   AGE
metrics-app-7674cfb699-2gznv              1/1     Running   0          40s
metrics-app-7674cfb699-2hk6r              1/1     Running   0          40s
metrics-app-7674cfb699-5926q              1/1     Running   0          40s
metrics-app-7674cfb699-5qgg2              1/1     Running   0          2d2h
metrics-app-7674cfb699-9zkk4              1/1     Running   0          25s
metrics-app-7674cfb699-dx8cj              1/1     Running   0          56s
metrics-app-7674cfb699-fmgpp              1/1     Running   0          56s
metrics-app-7674cfb699-k9thm              1/1     Running   0          25s
metrics-app-7674cfb699-wzxhk              1/1     Running   0          2d2h
metrics-app-7674cfb699-zdbtg              1/1     Running   0          40s

停止压测一段时间之后,pod数量就会根据HPA策略自动缩容成1个,说明我们的配置是成功的

相关文章

对接alertmanager创建钉钉卡片(1)
手把手教你搭建OpenFalcon监控系统
无需任何魔法即可使用 Ansible 的神奇变量“hostvars”
openobseve HA本地单集群模式
基于k8s上loggie/vector/openobserve日志收集
openobseve单节点和查询语法

发布评论