Forum

Hello,

We have checked the status of this server (Tencent Tokyo, 43.130.246.201):

Reason: The ingress-controller component is stuck in the STARTING state (it never successfully started, with the timestamp stuck at 1970-01-01), which is why services on this server are inaccessible. Other system components (DNS, log-api, fluent-bit, node-exporter, etc.) are running normally, and resources are sufficient (CPU 10%, Memory 50%), so this is not a resource shortage issue.

The reason why restarting and multiple "Reinstall Zeabur System Components" attempts failed is that the ingress-controller is blocked by the same issue every time it starts. Based on previous similar cases, the most common cause is that another program on the host server is occupying port 80 or 443 (e.g., a system-default nginx starting automatically after a reboot), preventing the ingress-controller from binding to the ports and causing it to crash repeatedly.

Zeabur's k3s requires exclusive access to the following ports:

Please assist with the following troubleshooting:

1. SSH into the server and run:

sudo ss -tlnp | grep -E ':(80|443)\b'
# or
sudo lsof -i :80
sudo lsof -i :443

2. If you find that nginx, apache, or other programs are occupying 80/443, please stop and disable them (using nginx as an example):

sudo systemctl stop nginx && sudo systemctl disable nginx

3. After stopping, wait 1–2 minutes; the ingress-controller will automatically rebind and recover, and all services will return to normal.

If 80/443 are not occupied, please paste the output of the following commands so we can further investigate the specific errors at the pod level:

sudo k3s kubectl -n zeabur get pods -o wide | grep ingress
sudo k3s kubectl -n zeabur describe pod <ingress-pod-name>
sudo k3s kubectl -n zeabur logs <ingress-pod-name> --previous