Author: Zen and the Art of Computer Programming
1 Introduction
In September 2017, the Linux Foundation (LF) released the Containerd project. As one of its main container runtime engines, containerd provides a portable, lightweight container runtime interface (CRI), which enables consistent running of containers between different Linux systems. Well-known container platforms such as Docker and RKT already support the CRI interface. Subsequently, the Linux Foundation solicited opinions from the community in the CCv0 version of the CRI specification. The results of this solicitation prove that CRI has clear goals, good scalability and stability, and provides highly abstract interface specifications, which are very important for unified container management and cross-platform collaboration. Therefore, this document is based on the open source CRI interface standard and formulates a container runtime interface specification specifically for the Linux environment.
Before reading this document, it is recommended that readers first understand the container runtime interface specifications such as Docker and RKT, and become familiar with the terminology in the field of Kubernetes container orchestration. It will be easier to understand the content of the specification when you read this document after understanding the basic concepts.
Before continuing to read this document, please note the following points:
- This document is based on the open source cri-api project, and the document structure refers to this project.
- This document is not a formal standard for CRI. It only describes the current functions of CRI and its specific interface protocols.
- This document does not cover the implementation details of individual container runtimes or management tools; these aspects will be described in separate documents.
# 2.基本概念术语说明
2.1.Container
A container is a running environment that encapsulates an application and all its dependent packages, including the kernel, libraries and other files. A container includes container images, resource restrictions, storage information, startup parameters, etc. The container isolates the underlying hardware resources (such as CPU and memory) shared with the host and provides a self-contained running environment that is completely independent from other containers. The life cycle of a container is usually