Hello, I logged into your server and checked the issue. Here are my findings:
1. Why the ingress-controller won't start
The ingress-controller is in a CrashLoopBackOff state and has restarted 264 times. The container logs are clear:
HTTP server error: listen tcp :80: bind: address already in use
This means ports 80/443 are occupied by another process. Using ss -tlnp, I found that the manually installed nginx (systemd nginx.service, PID 1369) is binding these ports, with the configuration file /etc/nginx/sites-enabled/health-api.conf serving your tools.inews119.com (using a manually requested Let's Encrypt certificate).
Side effect: Because the ingress-controller keeps retrying, your server's IP has hit the Let's Encrypt rate limit (429 too many new registrations from this IP in the last 3h). Once the port conflict is resolved and the crash loop stops, the rate limit will naturally expire.
2. What does the ingress-controller do? Can I ignore it?
No. The ingress-controller is the "external traffic gateway" that Zeabur runs on every dedicated server:
- It routes external HTTP/HTTPS requests to the corresponding Zeabur service on your server based on the host.
- It automatically requests and renews Let's Encrypt certificates.
Looking at the ingress logs, your OpenClaw is actually running as a Zeabur project (the fusclaw.zeabur.app rule is there), but this external URL is currently unreachable. Additionally, Zeabur system URLs like logs.<server>.servers.onzeabur.com, cadvisor.<server>..., and node-exporter.<server>... are also broken. So, while you intended for nginx and OpenClaw to coexist, the Zeabur external services are currently non-functional.
3. Fixing it via SSH
The most direct approach is to stop and disable the system nginx to free up ports 80/443 for the ingress-controller:
sudo systemctl stop nginx
sudo systemctl disable nginx
The ingress-controller will detect the ports and start within a minute or two, and your Zeabur service URLs (including fusclaw.zeabur.app) will be restored.
I recommend deploying the tools.inews119.com site as a Zeabur service on the same server, then binding tools.inews119.com to that service in the Zeabur Dashboard. Zeabur will automatically handle SSL—no need to manually install nginx or certbot.
If you must keep the host-level nginx, you need to configure it to listen on different ports (e.g., 8080/8443) instead of 80/443; otherwise, it will always conflict with the Zeabur ingress.
A quick reminder
This machine has 4GB of RAM (+2GB swap), and currently, 2.5 GiB of RAM + 879 MB of swap are in use. The OpenClaw image itself is 1GB. BestEffort pods (including the ingress-controller) are the first to be killed by the kernel when memory is tight. If you plan to add more services, I recommend upgrading your specifications.
