Skip to main content
Version: v25.09

FAQ

1. Why does the installation of openFuyao fail on other OSs?​

Currently, openFuyao supports only Ubuntu and openEuler. If you install openFuyao on another operating system (OS), an error is reported and the installation stops.

2. Why does the installation of openFuyao fail when I use runc 1.1.3?​

Currently, openFuyao supports only runc 1.1.12. If another runc version is detected during installation, an error is reported and the installation stops.

3. Why is the newly added package not displayed correctly on the package management page?​

This issue may be caused by network fluctuations or synchronization failures. You can click the Synchronize button manually to resolve it.

4. Why is the error message "unable to find index.yaml, please provide correct ChartMuseum project url" displayed when I add a remote repository?​

This error occurs when the remote repository URL is incorrect. Ensure that the URL specifies the exact project name in the Harbor repository, such as openFuyao's official repository URL: https://harbor.openfuyao.com/chartrepo/openfuyao-catalog, which specifies the project "openfuyao-catalog".

If a user is already assigned the platform admin role, any subsequent cluster-level role assignments will not override the platform admin permissions. The permissions of the platform admin role have the highest priority.

6. Why cannot a cluster admin user invite other users to join the cluster?​

This issue occurs because the user does not have the platform admin role. The platform admin role must be assigned first in order to invite other users.

7. Why the error message "open/root/.ssh/known_hosts: no such file or directory" is displayed during installation?​

The possible cause is that passwordless login is not configured or you do not re-log in after configuring passwordless login. For details, see Installation and Deployment Document.

8. Why does the deployment fail after I uninstall an application and try to deploy it again?​

The uninstallation process may take some time, and deploying the application again during this period may result in failure due to incomplete uninstallation. You are advised to wait a few minutes and then try deployment again.

9. Why does the image pull fail during online cluster deployment, node addition, or offline artifact export?​

To unify the source of images, all required images are obtained from openFuyao's official Harbor repository (https://harbor.openfuyao.com) except for offline installation. Currently, openFuyao is not fully open-source. Therefore, an access trustlist is configured for the official Harbor repository of openFuyao. That is, only IP addresses in the trustlist can access the official Harbor repository. If the network is normal, but the image fails to be pulled during online cluster deployment, node addition, or offline artifact export, contact the maintenance personnel of the openFuyao community.

10. Why does the binary file acquisition fail during online cluster deployment, node addition, or offline artifact export?​

To standardize the image acquisition process, all required binary files are obtained from openFuyao's official OBS (https://openfuyao.obs.cn-north-4.myhuaweicloud.com) except for offline installation. The binary files are obtained from the corresponding official channels. If a binary file fails to be obtained during the installation, contact the maintenance personnel of the openFuyao community.

11. Why does the openFuyao platform's log module display incorrectly when I access it through Firefox?​

If the openFuyao platform's log module fails to load or displays abnormally in the Firefox browser, you can manually clear the browser cache to resolve the issue.

12. Why might monitoring data be affected by time differences between the browser and server?​

A time difference between the browser and the server may cause data delays or misalignment. To prevent continuous impacts, you are advised to check the time settings of the local device and ensure that the time is synchronized with the server time. Adjustments to the browser time may take a while to take effect.

13. How do I change the port used to expose openFuyao's management plane?​

By default, openFuyao's management plane uses port 31616. To change the port in openFuyao v25.06, follow the following steps:

Input image descriptionNOTE
The following uses the fuyao-v25.06-openEuler-22.03-amd64 installation package as an example. The procedure for other installation packages is similar.

  • Procedure

    1. After the openFuyao installation package is downloaded, decompress the compressed shell script file in the asset directory.
      cd asset
      tar -zxvf fuyao-openEuler-22.03-amd64.tar.gz
    1. Open the file under fuyao-openEuler-22.03-amd64/ingress/ingress-nginx/resource/ingress-nginx.yaml, find nodePort: 31616 in line 372, and change 31616 to another port number.
    367 - appProtocol: https
    368   name: https
    369   port: 443
    370   protocol: TCP
    371   targetPort: https
    372   nodePort:  31616 # Configuration item
    1. After the modification, remove the original fuyao-openEuler-22.03-amd64.tar.gz and compress fuyao-openEuler-22.03-amd64 into fuyao-openEuler-22.03-amd64.tar.gz.
    rm -f fuyao-openEuler-22.03-amd64.tar.gz
    tar -czvf fuyao-openEuler-22.03-amd64.tar.gz ./fuyao-openEuler-22.03-amd64/*
    1. The port number has been changed. You can continue to install the cluster according to the installation guide. After the cluster is installed, use the new port number to access the openFuyao management plane.

14. Why does a pod fail to be pulled with the error message "failed to create fsnotify watcher: too many open files" recorded in the logs printed by the kubectl logs command?​

This issue is usually caused by the system's default value of fs.inotify.max_user_instances being too low. Resetting this value on the node where the pod is running resolves the problem. For multi-node cluster deployments, update this parameter on all nodes to prevent errors.

Solution 1: Run the following command to temporarily set the value:

sudo sysctl fs.inotify.max_user_instances=8192

Input image descriptionNOTE The setting must be reapplied each time you re-enter the console.

Solution 2: Run the following command to permanently set the value:

echo fs.inotify.max_user_instances=8192| tee -a /etc/sysctl.conf && sudo sysctl -p

15. Why are the yum and apt commands not working?​

The system files may have been modified during BKE installation and deployment, which can cause yum and apt commands to fail during or after the installation and deployment. In this case, manually revert the system files.

cd /etc/yum.repos.d/bak
cp ./openEuler.repo ..

16. Why does cluster deletion fail on the Cluster Lifecycle Management page?​

Clusters with unreachable IP addresses cannot be removed from the frontend. If deletion is required, perform it manually from the backend.

kubectl get bc -n <namespace>
kubectl delete bc <bc-name> -n <namespace>

17. Why does it take a long time to automatically refresh the cluster status on the management plane of the bootstrap node?​

  • Procedure

  1. Run the following command to check the cluster connectivity and latency:

    1.1 Find the IP address of the cluster in the ENDPOINT column shown in Figure 1.

     kubectl get bc -A

    Figure 1 Finding the IP address

    Finding the IP addresses

    1.2 Run the ping command, as shown in Figure 2.

     ping  <ENDPOINT IP address>

    Figure 2 Running the ping command

    Finding the IP addresses

  2. For BC clusters with connectivity issues, first determine whether they are still in use.

    • If a BC cluster is no longer in use, remove it with the following command:

      kubectl delete bc -n <bc namespace> <bc name>

    • If a BC cluster is still in use, restore its connectivity first and verify it with the ping command.