kubernetes

Kubernetes vs. Docker: What Does it Really Mean?

Kubernetes 101: Pods, Nodes, Containers, and Clusters

Demystifying Kubernetes Operators with the Operator SDK: Part 1 Kubernetes Components

Kubernetes Cluster vs Master Node

api

kubectl-commands

Dive Deep into Resource Requests and Limits in Kubernetes

Kubernetes中pod的创建流程

10 张图带你搞懂 Kubernetes Pod 的创建流程

kubectl 创建 Pod 背后到底发生了什么？

what is kubernetes?

docker: create containers k8s: 管理(编排)容器:

1. deploy container
2. scaling up and removing containers
3. monitor container

architecture

1. master: manager worker
2. worker: 运行container

1. master node(control plane)

1. etcd: 集群数据
2. kube-scheduler: 分配new pod 到node 上
3. apiserver: 提供对外和对内交流的枢纽
4. controller: a state machine; watch the state of your cluster, move the state to the desired state
   • Node controller: Responsible for noticing and responding when nodes go down.
   • Replication controller: Responsible for maintaining the correct number of pods for every replication controller object in the system.

2. Work node

1. Kubelet

   1. manage pods

2. kube proxy

   1. manage network;

3. pod: k8s编排的对象；

pod

k8s管理的基本单位； a pod is a group of container which share storage and network resoureces;

1. workload

用户部署的应用，可能有多个pod组成；

1. pods
2. deployment and replicaSet: 复制N份pod;; deployment–> replicaSet–> pods replicaSet—> pods
3. statefulset: 有状态应用
4. daemonset: 运行在每个node 上
5. job: 运行N次后退出
6. cornJob:定时运行

2.how to create;

1. kubuctl send command to apiserver;
2. apisever save yaml to etcd;
3. controller watch etcd, 生成pod配置信息, save to etcd;
4. scheduler watch etcd pod配置信息，指定相关node创建pod；
5. 指定node.kubelet创建pod;

2.lifeCycle

1. pending: 等待被创建；
2. running: pod所有容器已经被创建，至少一个容器正在运行或者正在启动
3. failed:pod所有容器都终止，至少一个 是非0退出；
4. unkonwon:无法获取pod状态

yaml

yaml decribe a k8s resouce state, the k8s will work constantly to ensure the state; resource type

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21

 apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx-deployment
spec:
  selector:
    matchLabels:
      app: nginx
  replicas: 2 # tells deployment to run 2 pods matching the template
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx:1.14.2
        ports:
        - containerPort: 80
        request:
        limit:

kind

define a k8s object:

1. workload
2. service
3. storage

metadata

help to uniquely identify the object;
1. name:
2. namespace:
3. lables： are meaningful and relevant to users, but do not directly imply semantics to the core system.

spec

spec: what state you desire for the obejct

1. replicas:
2. selector: select pods
3. spec.template: the actual template for the container

template

resouce

1
2
3
4
5
6
7

    resources:
            limits:
              cpu: '2'
              memory: 1Gi
            requests:
              cpu: 500m
              memory: 128Mi

memory: Mi,Gi cpu: 100m, 0.1; (0.1cpu core)

cpu compressible; ram incompressible;

1. request: used to schedule, if there are no node satisfy the resources the pod won’t be scheduled;

   2 cpu cluster max request 2cpu 2g ram cluster max request 2g

2. limit: moniter the runing, if reach;

   1. cpu: be restricted, continue run;
   2. memroy: oom(out of memory) error, pod restart;

3. the used resouce exceed the node’s resouce;

   1. cpu:
      1. the pod run normaly, but the pod’s load average will increase;
   2. ram:
      1. pod will evicted from current node by a order
         1. request && limit = 0
         2. at least one container in pod have reqeust specified for cpu or memory
         3. every container request == limit

kubectl command

command type:

1. manage resouces
2. debug/inspect

1. manage resouce

delete a deployment

for controller resourxe, you should delete the controller, otherwise the pods will recreate by the controller

1
2

  kubectl delete -f a.yaml
  kubectl delete delopyment  hellogo-deployment 

2. inspect/debug

1. exec a command in a container

kubectl exec (POD | TYPE/NAME) [-c CONTAINER] [flags] – COMMAND [args…]

1

kubect  exec centos-container -it  --- bash

scale

cluster-autoscaler

HorizontalPodAutoscaler

1. monitor cpu, memory, and if need scale

1
2
3

kind:

cpu