dry

Dry is a terminal application for Docker and Docker Compose.

It lets you browse containers, images, networks, and volumes, and manage Compose projects — bring them up, take them down, see what has drifted from the compose file — without leaving the terminal. It can be used with both local or remote Docker daemons.

Besides showing information, it can be used to manage Docker. Most of the commands that the official Docker CLI provides, are available in dry with the same behaviour. A list of available commands and their keybindings can be found in dry's help screen or in this README.

It can also be used as a monitoring tool for Docker containers, and, when the daemon is running a Swarm cluster, to manage Nodes, Services, and Stacks — see Docker Swarm below.

Dry is installed as a single binary and does not require external libraries.

The demo below shows a dry session.

dry --workspace enables the Phase 1 workspace shell. This keeps the current list view visible together with a passive context pane and a bottom activity pane.

• Tab and Shift+Tab switch focus between Navigator, Context, and Activity in the full workspace layout.
• When Context is focused, you can scroll it with the usual navigation keys.
• p / P pins or unpins the current preview.
• : opens the command palette with global and context-aware actions.
• Space opens Quick Peek, a temporary side panel with recent logs or inspect/details for the current selection.
• f toggles follow mode in the activity pane.
• Existing primary actions stay on their original keys. For example, Enter still opens the container command menu and still inspects resources in views where Enter already meant inspect/drill-down.
• Embedded activity logs start from a recent tail instead of replaying the full log history.
• On narrow or short terminals, workspace mode falls back to a compact single-pane layout and lets Tab switch between navigator and activity.

The easiest way to install the latest binaries for Linux and Mac is to run this in a shell:

curl -sSf https://moncho.github.io/dry/dryup.sh | sudo sh
sudo chmod 755 /usr/local/bin/dry

If you dont like to curl | sh, binaries are provided.

• darwin amd64 / arm64
• freebsd 386 / amd64
• linux 386 / amd64 / arm64 / armv6 / armv7
• windows amd64

If you're on OS X and want to use homebrew:

brew tap moncho/dry
brew install dry

docker run --rm -it -v /var/run/docker.sock:/var/run/docker.sock -e DOCKER_HOST=$DOCKER_HOST moncho/dry

yay -S dry-bin

Open a console, type dry. It will try to connect to:

• A Docker host given as a parameter (-H).
• if none given, a Docker host defined in the $DOCKER_HOST environment variable.
• if not defined, to unix:///var/run/docker.sock.

If no connection with a Docker host succeeds, dry will exit.

dry can talk to a remote Docker daemon through SSH, the same way the docker CLI does:

DOCKER_HOST=ssh://user@host dry

Host, port, and user are resolved like the ssh command resolves them: Hostname, Port, User, and IdentityFile directives from ~/.ssh/config are honored, the port defaults to 22, the user defaults to the current OS user, and the remote Docker socket defaults to /var/run/docker.sock (a different socket path can be given in the URL, e.g. ssh://user@host:2222/run/user/1000/docker.sock).

Authentication tries, in order: the identity files configured for the host in ~/.ssh/config (or, when none is configured, the ~/.ssh/id_* keys), keys held by a running SSH agent (SSH_AUTH_SOCK), and a password given in the URL. Passphrase protected key files are used through the agent.

The remote host key is verified against ~/.ssh/known_hosts and /etc/ssh/ssh_known_hosts, and the connection fails if the host is unknown or its key does not match. Connect once with ssh host to record the key. Verification can be disabled with DRY_SSH_INSECURE_SKIP_HOST_KEY_CHECK=1, which leaves the connection open to interception and should only be used against hosts you fully control.

dry -T light launches dry with the light color theme. Available themes: dark (default), light.

dry --workspace launches the experimental Phase 1 workspace layout.

dry -p launches dry with pprof package active.

Compose projects show up in their own view (key 8); pressing Enter on a project opens its services in the Compose Services view.

u and c work in both the Compose Projects and Compose Services views; d works in Compose Projects only.

The SYNC column reports whether a service's running containers match its compose file:

Projects are discovered two ways: from container labels — including stopped containers — and from a compose file in the directory dry was started in (that directory only; it is not a recursive search). A project known only from a file is listed with no running containers, and pressing u brings it up.

All of the above — the u/d/c keys and the SYNC column — require the docker compose CLI plugin. Without it, the Compose views still list projects discovered from container labels, and the keys respond with a status message naming the missing plugin instead of acting.

dry also works with Docker Swarm. When the connected daemon reports an active swarm, three extra views become available: Nodes (5), Services (6), and Stacks (7). Without an active swarm these views, their keybindings, and their command-palette entries are hidden.

All contributions are welcome.

• Fork the project.
• Make changes on a topic branch.
• Pull request.

Code released under the MIT license. See LICENSE for the full license text.

Built on top of:

• Bubbletea
• Bubbles
• Lipgloss
• Docker
• Docker CLI

See Awesome Docker list for similar tools to work with Docker.