API World!

我一直在用的mac终端是electerm，不是很有名，我甚至想不起它的名字，不过因为这款软件支持ssh、ftp，虽然有时候速度比较慢，但是我很喜欢。
支持mac、linux和windows

安装很简单

brew install --cask electerm

介绍很简洁

github官网
https://github.com/electerm/electerm
知乎介绍
https://zhuanlan.zhihu.com/p/348324919

几个高级的minikube命令

Now that we have familiarized ourselves with the default minikube start command, let’s dive deeper into Minikube to understand some of its more advanced features.

The minikube start by default selects a driver isolation software, such as a hypervisor or a container runtime, if one (VitualBox) or multiple are installed on the host workstation. In addition it downloads the latest Kubernetes version components. With the selected driver software it provisions a single VM named minikube (with hardware profile of CPUs=2, Memory=6GB, Disk=20GB) or container to host the default single-node all-in-one Kubernetes cluster. Once the node is provisioned, it bootstraps the Kubernetes control plane (with the default kubeadm tool), and it installs the latest version of the default container runtime, Docker, that will serve as a running environment for the containerized applications we will deploy to the Kubernetes cluster. The minikube start command generates a default minikube cluster with the specifications described above and it will store these specs so that we can restart the default cluster whenever desired. The object that stores the specifications of our cluster is called a profile.

As Minikube matures, so do its features and capabilities. With the introduction of profiles, Minikube allows users to create custom reusable clusters that can all be managed from a single command line client.

The minikube profile command allows us to view the status of all our clusters in a table formatted output. Assuming we have created only the default minikube cluster, we could list the properties that define the default profile with:

$ minikube profile list

|———-|————|———|—————-|——|———|———|——-|——–|

Profile	VM Driver	Runtime	IP	Port	Version	Status	Nodes	Active
minikube	virtualbox	docker	192.168.59.100	8443	v1.25.3	Running	1	*
———-	————	———	—————-	——	———	———	——-	——–

This table presents the columns associated with the default properties such as the profile name: minikube, the isolation driver: VirtualBox, the container runtime: Docker, the Kubernetes version: v1.25.3, the status of the cluster – running or stopped. The table also displays the number of nodes: 1 by default, the private IP address of the minikube cluster’s control plane VirtualBox VM, and the secure port that exposes the API Server to cluster control plane components, agents and clients: 8443.

What if we desire to create several reusable clusters instead, with other drivers (Docker or Podman (still experimental on Linux)) for node isolation, or different Kubernetes versions (v1.23.3 or v1.24.4), another runtime (cri-o or containerd), and possibly 2, 3, or more nodes (if permitted by the resources of our host system)? What if we desire to further customize the cluster with a specific networking option or plugin? The minikube start command allows us to create such custom profiles with the –profile or -p flags. Several of the isolation drivers support creation of node VMs or node containers of custom sizes as well, features that we will not explore in this course as not all are very stable at the time of this writing.

Below are a few examples of more complex start commands that allow custom clusters to be created with Minikube. They assume that the desired driver software (Docker and/or Podman) has been installed on the host workstation. There is no need to download the desired CNI (network plugin) or the container runtime, they will be set up and enabled by Minikube on our behalf:

$ minikube start –kubernetes-version=v1.23.3
–driver=podman –profile minipod

$ minikube start –nodes=2 –kubernetes-version=v1.24.4
–driver=docker –profile doubledocker

$ minikube start –driver=virtualbox –nodes=3 –disk-size=10g
–cpus=2 –memory=4g –kubernetes-version=v1.25.1 –cni=calico
–container-runtime=cri-o -p multivbox

$ minikube start –driver=docker –cpus=6 –memory=8g
–kubernetes-version="1.24.4" -p largedock

$ minikube start –driver=virtualbox -n 3 –container-runtime=containerd
–cni=calico -p minibox

Once multiple cluster profiles are available (the default minikube and custom minibox), the profiles table will look like this:

$ minikube profile list

|———-|————|———|—————-|——|———|———|——-|——–|

Profile	VM Driver	Runtime	IP	Port	Version	Status	Nodes	Active
minibox	virtualbox	crio	192.168.59.101	8443	v1.25.3	Running	3
minikube	virtualbox	docker	192.168.59.100	8443	v1.25.3	Running	1	*
———-	————	———	—————-	——	———	———	——-	——–

The active marker indicates the target cluster profile of the minikube command line tool. The target cluster can be set to minibox with the following command:

$ minikube profile minibox

The target cluster can be set to the default minikube with one of the following commands:

$ minikube profile minikube

$ minikube profile default

一个实际的例子

增加一个新的cluster

$ minikube start --driver=docker -p minibox

阿里云的这篇文章，介绍了如何利用亲和度（or反亲和度）特性，来实现pod的调度。
这里有一个例子，用来把pod调度到特定名字的node上。

比如要将Pod调度到nodeName是k8s-worker-2的节点上
apiVersion: v1
kind: Pod
metadata:
  name: webapp
  namespace: demo
  labels:
    app: webapp
spec:
  nodeName: 'k8s-worker-2'
  containers:
    - name: webapp
      image: nginx
      ports:
        - containerPort: 80

下面是chatgpt对这段脚本的详细解释

The YAML configuration you've provided is for a Kubernetes Pod named "webapp" in the "demo" namespace. Here's a breakdown of the configuration:

- `apiVersion: v1` and `kind: Pod`: These fields specify the Kubernetes API version and the resource type, which, in this case, is a Pod.

- `metadata`: This section contains metadata information for the Pod, such as its name, namespace, and labels.

  - `name: webapp`: The name of the Pod is set to "webapp."

  - `namespace: demo`: The Pod is created within the "demo" namespace.

  - `labels`: Labels are key-value pairs that help you organize and categorize Pods. In this case, the label "app" is set to "webapp."

- `spec`: This section defines the specifications of the Pod, including its containers, resources, and other settings.

  - `nodeName: 'k8s-worker-2'`: This field specifies the node (Kubernetes worker node) where the Pod should be scheduled. It pins the Pod to a specific node named "k8s-worker-2." This is known as node affinity and ensures that the Pod runs on a specific node, which can be useful for various scenarios such as hardware requirements or node-specific configurations.

  - `containers`: This is a list of containers running within the Pod. In this case, there is one container defined.

    - `name: webapp`: The container is named "webapp."

    - `image: nginx`: This field specifies the Docker image to use for this container. In this case, it uses the official Nginx image.

    - `ports`: This field defines the ports that are exposed by the container.

      - `containerPort: 80`: The container exposes port 80, which is the default port for HTTP traffic, allowing the Nginx web server within the container to listen for incoming HTTP requests.

This configuration will create a Pod running an Nginx web server on a specific node named "k8s-worker-2" in the "demo" namespace. The web server within the Pod will be accessible on port 80, and it has the label "app=webapp" for categorization and selection purposes.

Kubelet and kubectl are two essential components of the Kubernetes ecosystem, each serving distinct but interconnected roles.

Kubelet:

• Definition:

  • Kubelet is an agent that runs on each node in a Kubernetes cluster. Its primary responsibility is to ensure that the containers running on the node are in the desired state, as defined by the Kubernetes control plane.

• Key Functions:

  • Pod Lifecycle Management: Kubelet is responsible for starting, stopping, and maintaining the lifecycle of pods on its node. It communicates with the container runtime (e.g., Docker, containerd) to create and manage containers within pods.
  • Health Checks: Kubelet regularly performs health checks on the containers it manages. If a container or pod is not healthy, Kubelet can take corrective actions, such as restarting the container.
  • Resource Management: Kubelet ensures that containers have access to the necessary resources (CPU, memory) as specified in pod resource requests and limits.
  • Image Management: Kubelet pulls container images as needed and ensures they are available for pod instantiation.
  • Node Status: Kubelet reports the node’s status and resource usage to the Kubernetes control plane.

kubectl:

• Definition:

  • kubectl (Kube Control) is the command-line tool used to interact with a Kubernetes cluster. It serves as the primary interface for administrators, developers, and operators to manage and control the cluster.

• Key Functions:

  • Cluster Management: kubectl allows you to create, update, and delete Kubernetes resources like pods, services, deployments, and config maps.
  • Resource Inspection: You can use kubectl to inspect the status and details of resources in the cluster, e.g., kubectl get pods, kubectl describe service.
  • Scaling: kubectl enables you to scale deployments or replica sets up or down, e.g., kubectl scale deployment my-app –replicas=3.
  • Resource Creation: You can create resources from YAML or JSON files using kubectl apply -f <file.yaml>.
  • Pod Execution: You can execute commands inside pods using kubectl exec, access logs with kubectl logs, and copy files in and out of pods with kubectl cp.
  • Interactivity: kubectl offers an interactive mode, where you can run a shell in a container using kubectl exec -it.

In summary, Kubelet is an agent running on each node that takes care of pod and container management on that node, while kubectl is the command-line tool used to manage and control the entire Kubernetes cluster from a centralized location. They work together to ensure that containerized workloads are orchestrated and maintained according to the desired configuration.