Hi NocoBase team,
I’m trying to deploy NocoBase on OpenShift, but the pod fails immediately with a MODULE_NOT_FOUND error targeting the entrypoint script.
The issue is that OpenShift runs containers with a random UID for security. Since the /app folder and its subdirectories are owned by root, the entrypoint script can’t write or initialize anything, and the Node process chokes on the .sh file. It works fine on Docker because it runs as root by default, but it breaks on any hardened Kubernetes environment.
To fix this properly, could you update the Dockerfile to align with OpenShift best practices? Specifically, the app directories should belong to group 0 with group-write permissions:
RUN chgrp -R 0 /app && chmod -R g=u /app
This would make the image truly “cloud-native” and allow it to run on any K8s cluster without needing to grant anyuid privileges.
Thanks!