Concept

![[images/截屏2026-03-20 14.24.38.png]]

A Pod is the basic execution unit of a Kubernetes application.

The smallest and simplest unit in the Kubernetes object model that can be created or deployed.
Environment for containers organize application “parts” into Pods (server, caching, APIs, database…)
Containers run within Pods and share a Pod’s memory, IP, volumes, and more
Scale horizontally by adding Pod replicas
Pods live and die but never come back to life
If a pod get sick, Kubernetes will monitor that and automatically remove it then create a healthy pod

Network

Pod containers share the same Network namespace (share IP / port)
Pod containers have the same loopback network interface (localhost)
Container processes need to bind to different ports within a Pod
Ports can be reused by containers in seperate Pods

Pods do not span nodes

kubectl

Creating

kubectl run
kubectl create/apply with a yaml file

Run the image container in a Pod
kubectl run [pod_name] --image=[imgae_name:image_version]

Expose Port

Pods and containers are only accessible within the Kubernetes cluster by default

kubectl port-forward: expose a container port externally

kubectl port-forward [pod_name] [external_port:interal_port]

Delete

Running a Pod will cause a Deployment to be created.

Will cause Pod to be recreated

kubectl delete pod [pod_name]

delete Deployment

kubectl delete deployment [deployment_name]

YAML

Composed of maps and lists
Indentation matters
Always use spaces

key: value
map:
	key1: value
	key2:
		subKey: value
list:
	- item1
	- item2
listMap:
	- map1: value
	  map1p: value
	- map2: value
	  map2p: value

Define

apiVersion: v1
kind: Pod
metadata:
	name: test
spec:
	containers:
	- name: test
	  image: nginx:alpine

apiVersion: Kubernetes API version
kind: Type of Kubernetes resource
metadata: Metadata about the Pod
spec: The spec/bludprint for the Pod
- Information about the containers that will run in the Pod

Create

with a YAML file

kubectl create -f [yaml_name.yaml] --dry-run --validate=true

# Will error if Pod already exists
kubectl create -f [yaml_name.yaml]

Update

Create if not exist or apply changes to a Pod from YAML

kubectl apply -f [yaml_name.yaml]

When you want to use kubectl apply in the future:

kubectl create -f [yaml_name.yaml] --save-config

--save-config causes the resource’s configuration settings to be saved in the annotations.

In-place/non-disruptive changes can also be made to Pod using kubectl edit / kubuctl patch

Delete

Delete Pod using YAML file that created it

kubectl delete -f [yaml_name.yaml]

YAML Fundamental

K8s 所有资源（Pod/Deployment/Service 等）的 YAML 配置遵循统一核心结构，分为 4 个基础段：

apiVersion: v1 # API 版本（必选）
kind: Pod	   # 资源类型（必选）
metadata:	   # 资源元数据（必选）
  name: my-nginx
  labels:
    app: nginx
    rel: stable
spec:		   # 资源规格（必选，不同资源类型字段不同）
  containers:
  - name: my-nginx
    image: nginx:alpine
    ports:
    - containerPort: 80
    resources: {}

Commands:

# Basic Commands
kubectl create -f nginx.pod.yml --save-config
kubectl describe pod [pod-name]
kubectl apply -f nginx.pod.yml
kubectl exec [pod-name] -it sh
kubectl edit -f nginx.pod.yml
Kubectl delete -f nginx.pod.yml

# To exec into container use:
kubectl exec -it [pod-name] -- sh

核心通用属性详解

apiVersion

API 版本

取值示例	适用资源类型	说明
v1	Pod/Service/Namespace/ConfigMap/Secret	核心基础资源的稳定版本
apps/v1	Deployment/StatefulSet/DaemonSet	应用工作负载相关资源
batch/v1	Job/CronJob	批处理任务相关资源
networking.k8s.io/v1	Ingress/NetworkPolicy	网络相关资源

kind

资源类型

取值	中文名称	核心作用
Pod	最小部署单元	运行容器的最小载体
Deployment	无状态部署	管理 Pod 副本、自动重启、滚动更新
Service	服务发现	暴露 Pod 网络访问、负载均衡
Namespace	命名空间	资源逻辑隔离
ConfigMap	配置映射	存储非敏感配置数据
Secret	密钥	存储敏感数据（密码/令牌等）
StatefulSet	有状态部署	管理有状态应用（固定网络标识、持久化存储）
DaemonSet	守护进程	集群所有/指定 Node 上运行一个 Pod 副本
Job	一次性任务	执行完成后终止的任务
CronJob	定时任务	按时间规则重复执行 Job
Ingress	入口规则	集群外部访问集群内服务的规则

metadata

资源元数据

metadata:
  name: string               # 资源名称（必选，同一命名空间内唯一）
  namespace: string          # 所属命名空间（可选，默认 default）
  labels:                    # 标签（可选，键值对，用于资源筛选/关联）
    key1: value1
    key2: value2
  annotations:               # 注解（可选，键值对，存储非标识性元数据）
    key1: value1
  uid: string                # 唯一标识（自动生成，无需配置）

labels 核心用途：Service 关联 Pod、Deployment 管理 Pod、资源筛选（kubectl get pods -l key=value）
annotations 典型场景：存储资源描述、运维备注、第三方工具配置

Pod Health

Kubernetes relies on Probes to determine the health of a Pod container.

A Probe is a diagnostic performed periodically by the kubelet on a container.

Liveness probes can be used to determine if a Pod is healthy and running as expected
- When should a container restart
Readiness probes can be used to determine if a Pod should receive requests
- When should a container start receiving traffic

Failed Pod containers are recreated by default (restartPolicy defaults to Always)

Way to Check

ExecAction: Executes an action inside the container
TCPSocketAction: TCP check against the container’s IP address on a specified port
HTTPGetAction: HTTP GET request against container

Probes can have the results:

Success
Failure
Unknown

apiVersion: v1
kind: Pod
metadata:
  name: my-nginx
  labels:
    app: nginx
    rel: stable
spec:
  containers:
  - name: my-nginx
    image: nginx:alpine
    resources:
      requests:
        memory: "64Mi"
        cpu: "250m"
      limits:
        memory: "128Mi"
        cpu: "300m"
    ports:
    - containerPort: 80
    livenessProbe:
      httpGet:
        path: /index.html
        port: 80
      initialDelaySeconds: 15
      timeoutSeconds: 2 # Default is 1
      periodSeconds: 5 # Default is 10
      failureThreshold: 1 # Default is 3
    readinessProbe:
      httpGet:
        path: /index.html
        port: 80
      initialDelaySeconds: 3
      periodSeconds: 5 # Default is 10
      failureThreshold: 1 # Default is 3

httpGet: Check /index.html on port 80
initialDelaySeconds: Wait 15 seconds for init
timeoutSeconds: Timeout after 2 seconds
periodSeconds: Check every 5 seconds
failureThreshold: Allow 1 failure before failing Pod