上传文件网站根目录,wordpress媒体库配置,潜江资讯网最新招聘信息,网站建设 模仿阿里云官网本节重点介绍 :
服务发现的应用3种采集的k8s服务发现role 容器基础资源指标 role :nodek8s服务组件指标 role :endpoint部署在pod中业务埋点指标 role :pod
服务发现的应用
所有组件将自身指标暴露在各自的服务端口上#xff0c;prometheus通过pull过来拉取指标但是promet…本节重点介绍 :
服务发现的应用3种采集的k8s服务发现role 容器基础资源指标 role :nodek8s服务组件指标 role :endpoint部署在pod中业务埋点指标 role :pod
服务发现的应用
所有组件将自身指标暴露在各自的服务端口上prometheus通过pull过来拉取指标但是prometheus需要知道各个目标的地址是多少而且需要及时感知他们的变化所以采用服务发现是最好的解决方式
容器基础资源指标
我们可以看到prometheus采用k8s服务发现其中role :node 代表发现所有的node。
- job_name: kubernetes-nodes-cadvisorkubernetes_sd_configs:- role: node其中的原理是通过监听k8s node一旦node加入(扩容)node离开(缩容)prometheus可以及时收到node的信息通过访问节点的cadvisor指标path如node_ip:10250/metrics/cadvisor获取到相关指标通过prometheus的target展示页面(/targets)可以看到cadvisor node发现的结果target结果discovery 结果
k8s服务组件指标
kube-scheduler
采集配置如下
- job_name: kube-schedulerkubernetes_sd_configs:- role: endpointskubeconfig_file: follow_redirects: true采用k8s服务发现其中role :endpoints 代表发现所有的endpointsendpoint 可以理解为service向其发送流量的对象的IP地址在之前我们创建的控制平面暴露的service中kube-scheduler的配置如下
---
apiVersion: v1
kind: Service
metadata:namespace: kube-systemname: kube-schedulerlabels:k8s-app: kube-scheduler
spec:selector:component: kube-schedulerports:- name: http-metricsport: 10259targetPort: 10259protocol: TCP
那么对应的endpoint可以describe到就是下面所示的172.20.70.205:10259
[rootk8s-master01 ~]# kubectl describe svc kube-scheduler -n kube-system
Name: kube-scheduler
Namespace: kube-system
Labels: k8s-appkube-scheduler
Annotations: none
Selector: componentkube-scheduler
Type: ClusterIP
IP Families: none
IP: 10.96.208.114
IPs: 10.96.208.114
Port: http-metrics 10259/TCP
TargetPort: 10259/TCP
Endpoints: 172.20.70.205:10259
Session Affinity: None
Events: none这个和prometheus kube-scheduler target页面是一致的
kube-controller-manager
采集配置如下
- job_name: kube-controller-managerkubernetes_sd_configs:- role: endpointskubeconfig_file: follow_redirects: true采用k8s服务发现其中role :endpoints 代表发现所有的endpoints在之前我们创建的控制平面暴露的service中kube-controller-manager的配置如下
---
apiVersion: v1
kind: Service
metadata:namespace: kube-systemname: kube-controller-managerlabels:k8s-app: kube-controller-manager
spec:selector:component: kube-controller-managerports:- name: http-metricsport: 10257targetPort: 10257protocol: TCP
那么对应的endpoint可以describe到就是下面所示的172.20.70.205:10257
[rootk8s-master01 ~]# kubectl describe svc kube-controller-manager -n kube-system
Name: kube-controller-manager
Namespace: kube-system
Labels: k8s-appkube-controller-manager
Annotations: none
Selector: componentkube-controller-manager
Type: ClusterIP
IP Families: none
IP: 10.96.35.204
IPs: 10.96.35.204
Port: http-metrics 10257/TCP
TargetPort: 10257/TCP
Endpoints: 172.20.70.205:10257
Session Affinity: None
Events: none这个和prometheus kube-controller-manager target页面是一致的
kube-etcd
采集配置如下
- job_name: kube-etcdkubernetes_sd_configs:- role: endpointskubeconfig_file: follow_redirects: true采用k8s服务发现其中role :endpoints 代表发现所有的endpoints在之前我们创建的控制平面暴露的service中kube-etcd的配置如下
---
apiVersion: v1
kind: Service
metadata:namespace: kube-systemname: kube-etcdlabels:k8s-app: kube-etcd
spec:selector:component: etcdtier: control-planeports:- name: http-metricsport: 2379targetPort: 2379protocol: TCP
那么对应的endpoint可以describe到就是下面所示的172.20.70.205:2379
[rootprome-master01 ~]# kubectl describe svc kube-etcd -n kube-system
Name: kube-etcd
Namespace: kube-system
Labels: k8s-appkube-etcd
Annotations: none
Selector: componentetcd,tiercontrol-plane
Type: ClusterIP
IP Family Policy: SingleStack
IP Families: IPv4
IP: 10.96.136.217
IPs: 10.96.136.217
Port: http-metrics 2379/TCP
TargetPort: 2379/TCP
Endpoints: 192.168.3.200:2379
Session Affinity: None
Events: none
这个和prometheus kube-etcd target页面是一致的
部署在pod中业务埋点指标
采集配置如下
- job_name: kubernetes-podskubernetes_sd_configs:- role: podkubeconfig_file: follow_redirects: true采用k8s服务发现其中role :pods 代表发现所有的pods相当于执行kubectl get pod -A
[rootk8s-master01 ~]# kubectl get pod -A
NAMESPACE NAME READY STATUS RESTARTS AGE
calico-system calico-kube-controllers-854b9dcf89-gct84 1/1 Running 5 139d
calico-system calico-node-58m74 1/1 Running 7 139d
calico-system calico-node-8pwz5 1/1 Running 1 42d
calico-system calico-typha-56958ddd97-9zpd2 1/1 Running 2 42d
calico-system calico-typha-56958ddd97-gnt8k 1/1 Running 8 139d
default grafana-d5d85bcd6-f74ch 1/1 Running 0 4d5h
default grafana-d5d85bcd6-l44mx 1/1 Running 0 4d5h
default ink8s-pod-metrics-deployment-85d9795d6-95lsp 1/1 Running 0 20h
ingress-nginx ingress-nginx-controller-6cb6fdd64b-p4s65 1/1 Running 0 4d5h
kube-admin k8s-mon-daemonset-z6sfw 1/1 Running 1 42d
kube-admin k8s-mon-deployment-6d7d58bdc8-rxj42 1/1 Running 0 4d5h
kube-system coredns-68b9d7b887-ckwgh 1/1 Running 2 139d
kube-system coredns-68b9d7b887-vfmft 1/1 Running 2 139d
kube-system etcd-k8s-master01 1/1 Running 7 125d
kube-system kube-apiserver-k8s-master01 1/1 Running 2 74d
kube-system kube-controller-manager-k8s-master01 1/1 Running 66 136d
kube-system kube-proxy-kc258 1/1 Running 1 42d
kube-system kube-proxy-zx87g 1/1 Running 2 139d
kube-system kube-scheduler-k8s-master01 1/1 Running 64 83d
kube-system kube-state-metrics-564668c858-dnmnh 1/1 Running 0 4d3h
kube-system metrics-server-7dbf6c4558-zwp5m 1/1 Running 0 4d5h
kube-system prometheus-0 2/2 Running 0 4d3h
tigera-operator tigera-operator-cf6b69777-mlgk9 1/1 Running 85 139d然后访问的时候pod的ip因为在k8s中是pod之间网络是扁平的所以prometheus的pod可以访问到其他的podtarget结果discovery结果
本节重点总结 :
服务发现的应用3种采集的k8s服务发现role 容器基础资源指标 role :nodek8s服务组件指标 role :endpoint部署在pod中业务埋点指标 role :pod
