Terminal UI

Run mm with no arguments in an interactive terminal, from anywhere inside a project:

mm
mm --environment staging

The terminal UI opens on the default environment, or on the one you name. It refreshes on its own: every 3 seconds for a local environment and every 10 seconds for a remote one. Press ? in any view for its keys and q to quit.

Use explicit subcommands in scripts; the terminal UI needs an interactive terminal.

Find your way around#

The header shows the project, the environment, and where that environment runs: local, or remote and the target's name. A remote environment gets a solid label so it is hard to miss.

Views are listed in the sidebar and numbered. Press the number to open one.

Providers, Jobs, and Guided have no number: reach them with tab, or with [ and ], which jump between sidebar groups.

See and switch environments#

Press 2.

Each row is one environment from the project configuration: where it runs, its runtime, how many services are running, the deployed commit, and when it was last checked. The current environment is live. The others are checked when you open the view; press r to check them again. Checking a remote environment opens one SSH connection to its host, so the view does not poll them.

Select an environment and press enter to switch to it. Every other view then shows that environment. If the host cannot be reached, the terminal UI says why and stays where it was. esc cancels a switch in progress.

An environment marked not set up names a target this machine does not know. Register the target, or use the repository's target definition, then press r.

Act on a service or a container#

Press 3. Services with more than one container list them underneath; space folds and unfolds the selected service.

Select a service or one of its containers and press enter for its actions, or use the action's key directly.

Follow logs#

l opens the last 200 lines and follows new ones.

Lines from standard error have a bar in the margin. Lines that mention an error or a warning are tinted. Configured secret values are redacted, as in mm service logs. The view keeps the newest 5,000 lines.

Run a command#

x asks for a command and runs it in one container.

The arguments are passed exactly as typed: no shell runs and there is no TTY. Quotes group words. For pipes, redirects, or variables, start a shell yourself:

sh -c 'ls /app | wc -l'

The up and down arrows recall earlier commands. A command may run for 30 seconds, the same limit as mm service exec; esc stops it sooner. When it ends, enter asks for another and r runs the same one again. Commands that need input, such as a shell or a database console, need mm service shell instead.

Restart#

r asks first, then restarts the containers. The image, configuration, and volumes stay as deployed; to pick up a changed variable or image, press u to deploy instead. On Docker Swarm, restart replaces the service's tasks in the order its update policy sets, and a single container cannot be restarted.

Change the configuration#

Environments, jobs, schedules, configuration values, and the team can be changed without leaving the terminal UI. The keys are the same in every view that supports them:

Each change follows the same three steps as mm config:

1. Fill in a short form. tab moves between fields and the arrow keys change a choice. An edit keeps every key the form does not show.
2. Review the exact diff of the configuration file. Nothing is written before this point; b goes back to the form.
3. enter writes the file. If your review is out of date because the file changed meanwhile, mm refuses and asks you to review again.

A change is a change to files in your checkout. On a remote environment it takes effect when you commit, push, and deploy; the terminal UI lists those steps after it saves.

Change a variable#

Press 5, or v on a service to see only its variables. / searches and p shows only the variables with a problem.

After a local change, press u to start the stack again with the new value.

Add a teammate#

Access to remote environments is declared in the repository's users: key: who each person is, their public keys, and their level per environment. See manage remote access for the levels.

Press 9 for the Access view, then a.

1. Ask your teammate to run mm key create and then mm key show, and to send you the line that starts with ssh-ed25519. It is a public key; the private half never leaves their machine.
2. Enter their name and paste the key.
3. Choose a level for each remote environment with the arrow keys. none leaves them out of that environment.
4. Press enter and review the diff.
   
5. Press enter again to write the file. The terminal UI then lists what is left to do.
   

The change gives nobody access by itself. It takes effect when it is merged and each environment is deployed, or when an admin runs mm team sync there. Who may merge it is up to your repository: move the users into their own file with mm config split users and list .config/mm/users.yaml in CODEOWNERS to give access its own reviewers.

To change someone's level, select them and press e. To remove them, press D; they keep access to an environment until it is deployed again.

Request access for yourself#

On a new machine, press 9 and then j. The terminal UI creates this device's key if it has none, proposes your entry with the levels you ask for, and shows the same review. Commit the change on a branch and open a pull request; a code owner grants the access by merging it.

Work on a remote environment#

On a remote environment the terminal UI talks to the host's remote daemon with your device key, and the host decides what you may do. The Services view shows the target, who the host says you are, your access, and the deployed commit.

Variables are shown from your checkout. When your checkout differs from the deployed commit, the view says so instead of showing values the host does not use.