Cluster Scale-up ​

This section describes how to scale up a cluster. The following uses openEuler 22.03 on the Arm64 architecture as an example. The scale-up procedure is the same for other supported operating systems and architectures.

Configuration Requirements ​

Node Resource Requirements ​

Table 1 Recommended resource configuration of the worker node

NOTE
When simultaneous multithreading (SMT) or hyper-threading is disabled, a vCPU is equivalent to a physical core. After SMT or hyper-threading is enabled, the following formula is used for calculating the number of vCPUs: Number of vCPUs = (Number of threads per core x Number of cores) x Number of sockets.

Table 2 Recommended resource configuration of the master node

NOTE

• The required resource of the master node increases with the number of worker nodes.
• When SMT or hyper-threading is disabled, a vCPU is equivalent to a physical core. After SMT or hyper-threading is enabled, the following formula is used for calculating the number of vCPUs: Number of vCPUs = (Number of threads per core x Number of cores) x Number of sockets.

Node Configuration Requirements ​

• Select the installation host as the scale-up host.
• The scale-up host must be connected to the cluster network.
• Root user login is available on all hosts to be scaled up.

Environment Preparation ​

Prepare two Linux computers running openEuler 22.03 in an HA cluster or a non-HA cluster.

Table 3 Environment information

Scale-up Configuration File Example ​

The following are examples of configuration files for scaling up HA and non-HA clusters. In the following scale-up steps, the cluster planning configuration file is modified to create a cluster scale-up configuration file.

1. The following is an example YAML file for scaling up an HA cluster.

   yaml

   apiVersion: openfuyao.io/v1beta1
   kind: cluster
   metadata:
       name: example-cluster
   spec:
       nodes:
           masters:
               - hostname: fuyao-master
                 user: root
                 ip: 192.168.0.6
                 port: 22
                 privateKeyPath: /root/.ssh/id_rsa
                 removeTaint: true
                 labels:
                   - name: label-name
                     value: label-value
           workers:
               - hostname: fuyao-worker
                 user: root
                 ip: o.x.y.z
                 port: 22
                 privateKeyPath: /root/.ssh/id_rsa
                 labels:
                   - name: label-name
                     value: label-value
       components:
           - name: kube-vip
             param:
               - name: vip
                 value: 192.168.100.129
                 type: string
                 describe: VIP
               - name: interface
                 value: eth0
                 type: string
                 describe: NIC

2. The following is an example YAML file for scaling up a non-HA cluster.

   yaml

   apiVersion: openfuyao.io/v1beta1
   kind: cluster
   metadata:
       name: example-cluster
   spec:
       nodes:
           masters:
               - hostname: fuyao-master
                 user: root
                 ip: 192.168.0.6
                 port: 22
                 privateKeyPath: /root/.ssh/id_rsa
                 removeTaint: true
                 labels:
                   - name: label-name
                     value: label-value
           workers:
               - hostname: fuyao-worker
                 user: root
                 ip: o.x.y.z
                 port: 22
                 privateKeyPath: /root/.ssh/id_rsa
                 labels:
                   - name: label-name
                     value: label-value

Scale-up Procedure ​

1. Log in to the installation host as the root user.

2. Download the installation tool and installation script. Skip this step if the installation package has already been downloaded during the cluster installation.

   curl -sfL https://openfuyao.obs.cn-north-4.myhuaweicloud.com/openFuyao/installer/releases/download/v25.06/download.sh | bash -

   By default, the installation package of the latest version is downloaded by running the preceding command. To download the installation package of another version, run the following command with VERSION set to the desired version:

   NOTE
   For details about how to verify the integrity of the installation package, see Installation Package Integrity Verification.

   export VERSION=x.y.z
   
   curl -sfL https://openfuyao.obs.cn-north-4.myhuaweicloud.com/openFuyao/installer/releases/download/v25.06/download.sh | bash -

3. Generate a cluster planning configuration file.

   • Go to the installation package directory.

     NOTE
     Here uses openEuler 22.03 on the Arm64 architecture as an example. Adjust the path according to the actual situation.

     shell

     export INSTALLER_PACKAGE_PATH=fuyao-v25.06-openEuler-22.03-arm64
     
     cd $INSTALLER_PACKAGE_PATH

   • Run the following command to generate a configuration file for an HA cluster:

     shell

     ./fuyao create cluster-config --ha

   • The following shows the content of the configuration file.

     shell

     apiVersion: openfuyao.io/v1beta1
     kind: cluster
     metadata:
         name: example-cluster
     spec:
         nodes:
             masters:
                 - hostname: fuyao-master
                   user: root
                   ip: 192.168.0.6
                   port: 22
                   privateKeyPath: /root/.ssh/id_rsa
                   removeTaint: true
                   labels:
                     - name: label-name
                       value: label-value
             workers:
                 - hostname: fuyao-worker
                   user: root
                   ip: o.x.y.z
                   port: 22
                   privateKeyPath: /root/.ssh/id_rsa
                   labels:
                     - name: label-name
                       value: label-value
         components:
             - name: kube-vip
               param:
                 - name: vip
                   value: 192.168.100.129
                   type: string
                   describe: VIP
                 - name: interface
                   value: eth0
                   type: string
                   describe: NIC

4. Modify the cluster planning configuration file.

   • Edit the cluster planning configuration file.

     shell

     export HA_ADD_NODE_CONFIG_FILE=example-cluster.yaml
     
     vi $HA_ADD_NODE_CONFIG_FILE

     • Enter the node list information. The following is an example.

       • Enter information about at least one master node and one worker node. The master node information specified in the node list must correspond to an existing master node in the cluster.
       • hostname: name of a node in the cluster, which must be the same as the actual host name.
       • user: username for logging in to the node during deployment. Currently, only the root user is supported.
       • ip: IP address of the node.
       • port: port number used by SSH. The default value is 22.

   • Enter i to enter the edit mode for modifications. After the modifications are complete, enter :wq to save the modifications and exit. The following shows the modified configuration file.

     yaml

     apiVersion: openfuyao.io/v1beta1
     kind: cluster
     metadata:
       name: example-cluster
     spec:
       # Specify the nodes.
       nodes:
         # Master node configuration
         masters:
           # The hostname in the cluster can be different from the hostname of the host.
           - hostname: node5   # Configuration item
             user: root        # Configuration item
             ip: 192.168.0.6   # Configuration item
             port: 22          # Configuration item
             privateKeyPath: /root/.ssh/id_rsa  # Configuration item (optional)
           - hostname: node12  # Configuration item
             user: root        # Configuration item
             ip: 192.168.0.13  # Configuration item
             port: 22          # Configuration item
             privateKeyPath: /root/.ssh/id_rsa  # Configuration item (optional)
             labels:
               - name: node12       # Configuration item (optional)
                 value: master-04   # Configuration item (optional)
         workers:
           - hostname: node13  # Configuration item
             user: root        # Configuration item
             ip: 192.168.0.14  # Configuration item
             port: 22          # Configuration item
             privateKeyPath: /root/.ssh/id_rsa  # Configuration item (optional)
             labels:
               - name: node13      # Configuration item (optional)
                 value: worker-02  # Configuration item (optional)
       components:
         - name: kube-vip  # or non-HA cluster scale-up, the kube-vip component is not required.
           param:
             - name: vip
               value: 192.168.0.50  # Configuration item
               type: string
               describe: VIP
             - name: interface
               value: eth0          # Configuration item
               type: string
               describe: NIC

   • If SSH key–based login is configured for nodes in the cluster during configuration and installation, make sure to configure SSH key–based authentication on the installation host in advance.

5. Perform scale-up in the installation package directory.

   ./fuyao add nodes -f $HA_ADD_NODE_CONFIG_FILE

   If password-based login is configured for nodes in the cluster, make sure to enter the corresponding node passwords as prompted during the installation.

6. Check the scale-up result.

   After the automatic scale-up is successful, run the following command on the master node to check the scale-up result. If all pods are in running state, the environment is ready:

   kubectl get pod -A

   NOTE

   • The scale-up is complete when all pods are in running state. The actual time may vary depending on actual network conditions.
   • The openFuyao authentication and authorization mechanism uses self-signed keys. All authentication and authorization keys and configuration files are stored under /etc/kubernetes/webhook on all master nodes. After the installation is complete, you may choose to remove the files in this path. However, if the files are removed and the API server is restarted, authentication will fail.
   • The components deployed for scale-up using the installation and deployment tool are configured with default settings and are intended only for testing openFuyao platform functions. For a production environment, adjust the security configurations of Kubernetes, container runtime, and network plug-in components.

Precautions:

• When performing multiple scale-up operations, for example, if you first add worker1, you need to include the information about worker1 in the YAML file. When scaling up the cluster later by adding worker2, you only need to add the information about worker2 to the YAML file, without repeating the information about worker1.

Uninstallation ​

1. Log in to the installation host as the root user.

2. Remove the asset/install-component-manifest.yaml file.

   rm asset/install-component-manifest.yaml

3. Run the following command to enter the installation package directory:

   shell

   cd $INSTALLER_PACKAGE_PATH

   Edit the configuration file. The modifications to the configuration file should be consistent with those made during installation. Enter information about hostname, user, passwd, ip, and port for the master and worker nodes in the cluster.

   shell

   vi $CLUSTER_HA_CONFIG_FILE

   CAUTION
   You need to enter the information about the added nodes in the configuration file.

4. Perform the uninstallation in the installation package directory.

   NOTICE
   If multiple applications or extensions are installed in the environment through the application market, you are advised to uninstall them through the application management or extension management function. Otherwise, the uninstallation may be slow or fail.

   shell

   ./fuyao destroy cluster -f <cluster-yam-path, eg:./example-cluster.yaml>

5. When Successfully completed cluster destroy is displayed, the cluster is uninstalled.

   During the uninstallation, the logs generated during the installation and uninstallation are not removed. The logs are stored under /var/log/fuyao-installer/. To remove these logs, run the following command:

   shell

   rm -rf /var/log/fuyao-installer/