Wiki· Updated 18 Oct 2026
When the install fails
Common RunStack automation install errors and what to do about them.
Start with /tmp/rs.log: it has everything shown on screen and every command run. It contains no keys or tokens — you can attach it to a feedback ticket.
| Message | Cause and fix |
|---|---|
| "Download failed… token" | The token in the command is valid for 12 hours. Reload the service page and copy the command again. |
| "DNS: domain → …, but this server's public IP is …" | The domain's A record points elsewhere. Fix it and wait a few minutes. Behind the Cloudflare proxy, turn the orange cloud off for now. |
| "Port 80 is busy" / "Port 443 is busy" | nginx, Apache or another service already runs on the server. Choose local mode and proxy the port with your nginx — or free 80/443. |
| "Docker Hub is unreachable", 429 errors | Docker Hub is blocked or rate-limited. Add a mirror: RS_REGISTRY_MIRROR=https://mirror.gcr.io. |
| "Cannot pull image" | Wrong version in the answer or the registry is unreachable — Docker's error is printed above. |
| "Root is required" | Run with sudo: … | sudo bash. |
| HTTPS does not respond after install | Check that 80/443 are open from outside (provider firewall, cloud security groups). Proxy logs: docker logs rs-edge. |
| "… is already installed in …, moving is not supported" | The install stays in its directory; to move it, back up and install again into the new one. |
If you deleted the service directory by hand while the containers still run, rs <service> down finds and removes them.