Installation and Deployment

This chapter describes how to install and deploy the KubeOS tool.

• Installation and Deployment
  • Software and Hardware Requirements
    • Hardware Requirements
    • Software Requirements
    • Environment Preparation
  • KubeOS Installation
  • KubeOS Deployment
    • Building the os-operator and os-proxy Images
    • Building a Container OS Image
    • Deploying CRD, operator, and proxy

Software and Hardware Requirements

Hardware Requirements

• Currently, only the x86 and AArch64 architectures are supported.

Software Requirements

• OS: openEuler 22.03 LTS

Environment Preparation

• Install the openEuler system. For details, see the openEuler Installation Guide.
• Install qemu-img, bc, Parted, tar, Yum, Docker, and dosfstools.

KubeOS Installation

To install KubeOS, perform the following steps:

1. Configure the Yum sources openEuler 22.03 LTS and openEuler 22.03 LTS:EPOL:

   [openEuler22.03 LTS] # openEuler 22.03 LTS official source
   name=openEuler22.03 LTS
   baseurl=http://repo.openeuler.org/openEuler-22.03-LTS/everything/$basearch/ 
   enabled=1
   gpgcheck=1
   gpgkey=http://repo.openeuler.org/openEuler-22.03-LTS/everything/$basearch/RPM-GPG-KEY-openEuler
   

   [Epol] # openEuler 22.03 LTS:EPOL official source
   name=Epol
   baseurl=http://repo.openeuler.org/openEuler-22.03-LTS/EPOL/main/$basearch/
   enabled=1
   gpgcheck=1
   gpgkey=http://repo.openeuler.org/openEuler-22.03-LTS/OS/$basearch/RPM-GPG-KEY-openEuler 
   

2. Install KubeOS as the root user.

   # yum install KubeOS KubeOS-scripts -y
   

NOTE:

KubeOS is installed in the /opt/kubeOS directory, including the os-operator, os-proxy, os-agent binary files, container OS image build tools, and corresponding configuration files.

KubeOS Deployment

After KubeOS is installed, you need to configure and deploy it. This section describes how to configure and deploy KubeOS.

Building the os-operator and os-proxy Images

Environment Preparation

Before using Docker to create a container image, ensure that Docker has been installed and configured.

Procedure

1. Go to the working directory.

   cd /opt/kubeOS
   

2. Specify the image repository, name, and version for os-proxy.

   export IMG_PROXY=your_imageRepository/os-proxy_imageName:version
   

3. Specify the image repository, name, and version for os-operator.

   export IMG_OPERATOR=your_imageRepository/os-operator_imageName:version
   

4. Compile a Dockerfile to build an image. Pay attention to the following points when compiling a Dockerfile:

   • The os-operator and os-proxy images must be built based on the base image. Ensure that the base image is safe.
   • Copy the os-operator and os-proxy binary files to the corresponding images.
   • Ensure that the owner and owner group of the os-proxy binary file in the os-proxy image are root, and the file permission is 500.
   • Ensure that the owner and owner group of the os-operator binary file in the os-operator image are the user who runs the os-operator process in the container, and the file permission is 500.
   • The locations of the os-operator and os-proxy binary files in the image and the commands run during container startup must correspond to the parameters specified in the YAML file used for deployment.

   An example Dockerfile is as follows:

   FROM your_baseimage
   COPY ./bin/proxy /proxy
   ENTRYPOINT ["/proxy"]
   

   FROM your_baseimage
   COPY --chown=6552:6552 ./bin/operator /operator
   ENTRYPOINT ["/operator"]
   

   Alternatively, you can use multi-stage builds in the Dockerfile.

5. Build the images (the os-operator and os-proxy images) to be included in the containers OS image.

   # Specify the Dockerfile path of os-proxy.
   export DOCKERFILE_PROXY=your_dockerfile_proxy
   # Specify the Dockerfile path of os-operator.
   export DOCKERFILE_OPERATOR=your_dockerfile_operator
   # Build images.
   docker build -t ${IMG_OPERATOR} -f ${DOCKERFILE_OPERATOR} .
   docker build -t ${IMG_PROXY} -f ${DOCKERFILE_PROXY} .
   

6. Push the images to the image repository.

   docker push ${IMG_OPERATOR}
   docker push ${IMG_PROXY}
   

Building a Container OS Image

Precautions

• The root permissions are required for creating a container OS image.
• The RPM package source of the container OS image is the Everything and EPOL repositories of a specific version of openEuler. You are advised to configure both the Everything and EPOL repositories as Yum sources in the repo file provided for creating the image.
• By default, the container OS image built using the default RPM list is stored in the same path as the build tool. This partition must have at least 25 GiB free space.
• When creating a container OS image, you cannot customize the file system to be mounted.

Procedure

The command for creating a container OS is as follows:

generate.sh REPO_PATH VERSION AGENT_PATH ENCRYPTED_PASSWD

The parameters are described as follows:

• REPO_PATH: repo file path

• AGENT_PATH: os-agent binary file path

• VERSION: version of the created container OS image

• ENCRYPTED_PASSWD: password of the root user of the container OS image. The password must be encrypted and salted. You can generate a password using OpenSSL or KIWI commands.

To create a container OS, perform the following steps:

1. Go to the working directory.

   cd /opt/kubeOS/scripts
   

2. Run generate.sh to create the container OS. The following is a command example:

   bash generate.sh xx.repo v1 ../bin/os-agent '''$1$xyz$RdLyKTL32WEvK3lg8CXID0'''
   

   In the command, xx.repo indicates the actual Yum source file used for creating the image. You are advised to configure both the Everything and EPOL repositories as Yum sources.

   After the container OS image is created, the following files are generated in the /opt/kubeOS/scripts directory:

   • system.img: system image in raw format. The default size is 20 GB. The size of the root file system partition is less than 2,020 MiB, and the size of the Persist partition is less than 16 GiB.
   • system.qcow2: system image in QCOW2 format.
   • update.img: partition image of the root file system that is used for upgrade.

   The created container OS image can be used only in a VM of the x86 or AArch64 architecture.

Deploying CRD, operator, and proxy

Precautions

• The Kubernetes cluster must be deployed first. For details, see the openEuler 22.03 LTS Kubernetes Cluster Deployment Guide.

• The OS of the worker nodes to be upgraded in the cluster must be the container OS built using the method described in the previous section. If it is not, use system.qcow2 to deploy the VM again. For details about how to deploy a VM, see the openEuler 22.03 LTS Virtualization User Guide. Currently, KubeOS does not support the master nodes. Use openEuler 22.03 LTS to deploy the upgrade on the master nodes.

• The YAML files for deploying CustomResourceDefinition (CRD), os-operator, os-proxy, and role-based access control (RBAC) of the OS need to be compiled.

• The os-operator and os-proxy components are deployed in the Kubernetes cluster. os-operator must be deployed as a Deployment, and os-proxy as a DaemonSet.

• Kubernetes security mechanisms, such as the RBAC, pod service account, and security policies, must be deployed.

Procedure

1. Prepare YAML files used for deploying CRD, RBAC, os-operator, and os-proxy of the OS. For details, see YAML examples. The following uses crd.yaml, rbac.yaml, and manager.yaml as examples.

2. Deploy CRD, RBAC, os-operator, and os-proxy. Assume that the crd.yaml, rbac.yaml, and manager.yaml files are stored in the config/crd, config/rbac, and config/manager directories, respectively. Run the following commands:

   kubectl apply -f config/crd
   kubectl apply -f config/rbac 
   kubectl apply -f config/manager
   

3. After the deployment is complete, run the following command to check whether each component is started properly. If STATUS of all components is Running, the components are started properly.

   kubectl get pods -A