Troubleshooting
Common problems, why they happen, and what to do. Each section reads: symptom, cause, fix.
The panel does not open after install
Symptom: the browser cannot reach the panel's address.
Cause: the wrong address or port, a closed port at your provider, or DNS not pointing to the server yet.
Fix:
- Without a domain the panel is at
http://<server-ip>:3000/admin. With a domain, port 3000 is closed: usehttps://panel.example.com/admin. - Open ports 80 and 443 (and 3000 without a domain) in your provider's firewall or security group.
- Check that the domain's A record points to the server, then run the installer again with the same options.
A domain does not answer
Symptom: the app's domain times out, or answers "404 page not found".
Cause: DNS does not point to this server, the service is not running, or the domain leads to the wrong container port.
Fix:
- Check the DNS record (A, or a wildcard for generated domains) points to the server.
- Open the Domains page: the State column says Live, Not running (start or deploy the service), or Waiting for deploy (a Compose service's domain: deploy it).
- Check the domain's Container port is the port the app listens on inside its container (80 by default in the dialog).
- A domain change applies when saved, but only while the service runs.
HTTPS: the certificate is not issued
Symptom: the browser warns about the certificate, or HTTPS fails while HTTP worked.
Cause: Let's Encrypt could not check the name. Over HTTP it must reach the server on port 80 under that name. A wildcard (*.example.com) needs a DNS resolver. A name behind Cloudflare is checked at Cloudflare, not here.
Fix:
- Make sure DNS points here and port 80 is open, then wait a minute: the first certificate takes a few seconds to arrive.
- For a wildcard, add a resolver over DNS in Settings → Certificates (with a Cloudflare, DigitalOcean, Hetzner, Vultr or Linode token), or upload a wildcard certificate.
- Through Cloudflare Tunnel, Cloudflare gives visitors HTTPS: turn HTTPS off on the domain (the dialog offers Turn HTTPS off). Proxied by Cloudflare to this server, use a DNS resolver or an uploaded certificate.
- Read Traefik's live logs in Settings → Traefik: certificate errors are there.
A deployment failed
Symptom: a deployment ends in Failed, or the service keeps restarting.
Cause: the image could not be pulled (a typo, a private registry), the build failed, the app crashed at start, or its health check failed.
Fix:
- Open the Deployments tab and read that deployment's log: every step (pull, build, start) is there.
- The Overview shows Recent errors: failed tasks with their exit code.
- For a private image, set Registry user and Registry password or token on the Source tab.
- If an update failed, Swarm has already gone back to the previous version. To deploy an older configuration on purpose, use Roll back to this on a deployment that worked.
If Deploy answers that a deployment is already under way: one deployment runs at a time per service. Wait for it to end.
"Is already published" or "is reserved" on a port
Symptom: saving a published port is refused.
Cause: another service already publishes that port, or it is one the panel keeps for itself and Traefik (80, 443, 3000, 8000, 8080, 8443 by default).
Fix: choose another published port. For a website, prefer a domain: it needs no published port at all.
"Not enough memory" before a deploy
Symptom: creating or deploying a service shows Not enough memory for this, Memory will be low, or This service cannot run.
Cause: the panel estimates what the service needs (a build takes about 1 GB while it runs) and compares it with what the server has free.
Fix:
- Memory will be low: it can still be deployed; keep an eye on it.
- Not enough memory for this: services may be stopped for lack of memory. Free memory (stop unused services) or confirm with I understand, deploy anyway.
- This service cannot run: its memory reservations exceed the server; Swarm would leave it pending. Lower its Memory reservation (MB) on the Advanced tab, or add memory.
On a Mac, the memory is the Docker VM's: raise it in OrbStack's or Docker Desktop's settings.
Nothing answers after a reboot
Symptom: after the server restarts, the panel and the apps are down, but SSH works.
Cause: with the Firewall on, Docker starts only once the firewall's rules are loaded. If the saved rules do not load, a fallback (SSH and the web only) loads instead and the Firewall page says so; if neither loads, Docker does not start, so nothing is exposed unfiltered.
Fix: sign in over SSH and check systemctl status docker. To recover by hand when the firewall's files are gone:
rm /etc/systemd/system/docker.service.d/rdockpanel-firewall.conf && systemctl daemon-reload
systemctl start dockerOn a Mac, the Mac must be awake and Docker running; Swarm starts everything again with Docker.
Locked out by a firewall change
Symptom: after Apply on the Firewall page, you cannot reach the panel or SSH.
Fix: wait. Each change is undone by the server itself after 120 seconds unless you press Keep these rules. Before pressing it, test from a new connection (a private window, a new SSH session).
An import is blocked or refused
Symptom: the import's plan marks a container as blocked, or the start is refused.
Cause: the plan lists what blocks: for example a privileged container, the host network, a folder the panel never gives a service (/, the Docker socket, /proc…), a volume shared by two containers, or a password on the command line it cannot move. A start is refused when the data and a margin (10%, at least 2 GiB) do not fit on the disk.
Fix: read the reason next to each item, change the container or leave it out, free disk space, then read it again. While an import runs, its services refuse deployments and backups: finish or undo it first. See Import existing containers.
Still stuck?
Search or open an issue on GitHub, with the panel's version (Settings → Version) and the deployment's or Traefik's log, without secrets.