Troubleshooting

Start with Docker Manager’s own log. It explains most problems:

docker logs docker-manager

Signing in

I can’t find the first password. It’s printed once, in the log of the very first start, under INITIAL ADMIN ACCOUNT CREATED. Run docker logs docker-manager and scroll up. If you removed and recreated the container but kept the data volume, the admin user already exists and no new password is printed.

“That username and password don’t match.” Check for spaces you copied with the password. After several wrong attempts, more are refused for a while; the wait doubles with each further failure, up to 15 minutes.

Someone forgot their password. An administrator can open Users, choose Reset password in that person’s menu, and give them the temporary password it shows.

The only administrator forgot their password. Reset it from the server:

1. Add DM_RESET_PASSWORD=admin (use the user name) to the container’s environment, for example under environment: in the compose file.
2. Recreate the container: docker compose up -d.
3. Find the temporary password in the log: docker logs docker-manager, after PASSWORD RESET (DM_RESET_PASSWORD).
4. Sign in with it and choose a new password.
5. Remove DM_RESET_PASSWORD and run docker compose up -d again.

The user is signed out everywhere, and enabled again if they were disabled. If you forget step 5, nothing is reset again once they’ve chosen their new password; the log just reminds you to remove the variable. If the user is managed by the configuration file with a password:, sign in with that password instead.

Everyone shares one sign-in limit, or the activity log shows the proxy’s address for every user. Docker Manager is behind a reverse proxy that it doesn’t trust. See Behind a reverse proxy.

Docker access

No hosts appear after installing. The local host is only added when the Docker socket is mounted. Check that the compose file has - /var/run/docker.sock:/var/run/docker.sock under volumes, then recreate the container with docker compose up -d.

A host shows “Not reachable”.

• For an SSH host: the user must be in the docker group on the server, its public key must be in ~/.ssh/authorized_keys, and the server needs AllowTcpForwarding yes in /etc/ssh/sshd_config. The key must not have a passphrase.
• For a TLS host: check that port 2376 is reachable from Docker Manager’s server, and that the CA certificate, client certificate and client key are the ones made for that daemon.
• For any host: check that the server is on and Docker is running on it.

Stacks

A stack’s container can’t find its files, or a relative bind mount (./data) is empty. Docker resolves relative paths on the host. Mount the stacks and git folders at the same path on the host and in the container, as in the install example.

A stack is “Not managed here”. It was started outside Docker Manager. Use Find the compose file on its page to import it. See Stacks started elsewhere.

I can’t deploy or stop Docker Manager’s own stack. That’s on purpose: it would go offline halfway. Update it on the server with docker compose pull && docker compose up -d.

I can deploy a stack but not edit its file. Only administrators edit compose files, because they can give containers access to the host.

Containers

A container I removed came back. It belongs to a stack, which recreates it on its next deploy. Take the stack down, or remove the service from its compose file.

The container page says it ran out of memory. Its memory limit is too low for what it does. Raise Memory limit (MiB) under Settings, Advanced, or set it to 0 for no limit.

| or && don’t work in the Console. The Console runs one command at a time without a shell. To use shell features, run the shell yourself, for example sh -c "ps aux | grep node", if the image has one.

Uploads and proxies

Uploading a file or a database backup fails with “413” or a similar error. Your reverse proxy limits request size. Raise it, for example client_max_body_size 2g; in nginx. Docker Manager itself accepts volume uploads of up to 100 MB per file.

After upgrading from 1.x

Registry passwords, git tokens or the email password stopped working. If you set your own ENCRYPTION_KEY in 1.x, put the same value back in the compose file. See Upgrade from 1.x.

DNS

Port 53 is already in use. On many Linux servers, systemd-resolved holds port 53. The DNS guide explains how to free it or publish DNS on one address only.

Still stuck?

Search the GitHub issues, or open a new one with the relevant part of docker logs docker-manager (remove any passwords from it first).