Troubleshooting

The installer refuses port 3000

Something else is listening. The daemon publishes 3000 until the instance has a domain, and the installer refuses rather than fighting whatever is there. Free the port, or install with another.

A name has no certificate

Certificates under Platform says why, per name:

• not deployed — the name was added after the app's last deploy. Redeploy.
• pending — the proxy asked and Let's Encrypt has not answered yet, or refused; the proxy's own complaint is quoted. A record written late is retried every half hour on its own. A rate limit on sslip.io waits, or takes a domain of your own.
• no domain — set one in Settings.

A deploy failed

Open the deployment on the app's Overview tab. A build carries its own output, tail kept; a container that never came up says so. The previous container is still serving — a failed deploy is a no-op.

An app cannot reach another app

Use the internal address, not the public name: from inside the box the public name leaves for a record pointing back at the machine, which most hosts do not route. An app that has not deployed since the instance gained internal addresses does not answer to its own yet.

docker push is refused

• The username with the key must be your current username; a renamed account logs in again.
• The registry follows the instance's domain; with no domain there is nowhere to push.
• A member's key can push; a worker's or the builder's cannot.

A database is failed

error on it says what the engine printed. Start retries. The commonest cause is a login the engine refuses — root on MySQL, a user other than default on Redis — which is refused earlier now.

Redis has no Backups tab

On purpose. See Engines.

A worker says pending for ever

The daemon on it never called home. Check the box reaches the control plane's address on 443, and that the token was the one minted for that row — a token is shown once, and a lost one means removing and adding the server again.

A worker is ready and not on the mesh

The three mesh ports are blocked between the machines — at the provider's firewall, which Cubeship cannot see — or the machine is behind NAT. See The mesh.

The dashboard shows an error after an upgrade

The daemon is restarting underneath it; every write answers 503 until the new one is up. Reload in a minute.

Where the logs are

docker logs cubeship-daemon
docker logs cubeship-traefik      # ACME refusals are here and nowhere else
docker logs cubeship-postgres