[reading Notes] 09 talk about the essence of Kubernetes from container to container cloud

An in-depth Analysis of Kubernetes-09 from Container to Container Cloud to talk about the essence of Kubernetes

Kubernetes mainly solves the problems: service orchestration, node management, task scheduling.

Kubernetes node roles:

(1) Master: apiserver (providing API services), kube-scheduler (responsible for scheduling), kube-controller-manager (responsible for container orchestration)

(2) worknode: the core component is kubelet, which is responsible for communicating with apiserver and container runtime (such as docker)

Kubernetes recommends declarative API to choreograph services through API objects.

(1) multiple pod acts as proxies through service. Information such as IP and port of proxy pod is automatically updated to provide a unified entry (IP will not change throughout the lifecycle)

(2) the smallest management granularity is that pod,pod is a collection of container, and a pod can contain multiple container.

(3) configuration is managed by configmap, and directories or files are mounted to the container.

(4) sensitive files, such as certificates, password files, etc., are mounted through secret

(5) if you need to fix the hostname and require the start and stop order of the process, you can deploy it through statefulset.

(6) agent processes, such as log collection, monitoring processes, etc., each host needs a service that can be deployed through daemonset

(7) scheduled execution tasks are executed through cronjob

(8) one-time execution of tasks is completed through job

(9) automatic scaling is realized through HPA