Features19 Background & settings

Quiet in the tray, loud when it matters.

Docker Client MX keeps running in the system tray and tells you when a container crashes, becomes unhealthy, loses its network, runs into its memory limit or sets off one of your own rules, on this computer and, if you like, on your phone. A rule can even restart or stop the container for you. It cleans up after itself and keeps itself up to date.

Window closed. Now what?Read the comic: a chef in the back room, quiet until it matters →Away from your desk?Read the comic: the chef sends word, wherever you are →
Fig. — Settings, Notifications: what you hear about, rules of your own, and channels elsewhere.

What you can do

Lives in the tray

Start and stop projects and containers from the tray menu, and search from there.

Starts at sign-in

Straight to the tray, so notifications arrive even when you don't open the app.

Notifications that mean something

A crash, an unhealthy container, a lost network or port, a container at its memory limit, a WSL that hangs after sleep, or in a swarm a node that goes down or a service below its count. Never for a container you stopped yourself, and switched off per container when you want quiet.

A warning before it breaks

Memory heading for its limit, restarts piling up, a CPU far above the days before, a disk filling: a notification of its own kind, at most once a day per container, with its own switch under Notifications.

Notifications elsewhere

ntfy, Gotify, Discord, Slack, Pushover or a webhook with your own JSON template. Per channel the kinds you want, a test button, and at most 20 an hour. A server behind a certificate of your own, such as a home certificate authority, works once Windows trusts it.

Picks a container up again

Turn it on per container with a health check: when it turns unhealthy, the app restarts it, waits longer each time, and stops after three times in an hour.

Your own rules

A notification when a container stays above a CPU, memory, network or disk threshold for some minutes, when its log shows an error, a text, an expression or a field in a JSON log, or when it stops, crashes, runs out of memory or keeps restarting. Or when an image is pulled or removed, or a volume or network is created or removed, by name or any. A rule can be for another machine, checked also while you are elsewhere if you follow it, log rules included.

Rules that act

A rule on a measure or an event can restart or stop the container: checked right before, at most three times an hour unless you allow up to ten, and with your yes when you set it up. The notification says what the app did.

Cleans up once a week

Untagged images and build cache, never volumes or tagged images. A notification tells you what it freed. If you like, also the containers of old swarm tasks, an hour after they stopped.

Everything it does by itself

A Schedules page with everything the app does on its own: backups, updates at a fixed time, the checkup, cleanup, and the tasks of the app itself. Where and when each one runs, how the last time went and when the next one comes. Run, pause, resume or adjust one from its menu, or set a new one up with New task…, in the same two steps as on Backups.

Updates itself

The Windows installer, with a build of its own for ARM processors, and the Linux AppImage check for new versions and download them in advance. The x64 version on an ARM processor points you to the ARM build.

Five languages

English, Nederlands, Deutsch, Français and Español. Technical terms such as container and volume stay in English.

Light and dark

Toggle the theme in the header.

Remembers your filters

Filters on the list pages stay. Search fields deliberately don't, so no container seems to have vanished.

A menu under the right mouse button

Right-click a row in any table, or a project or service on the Compose page, for the same menu as its three-line button, at the mouse. With text selected, on a link or in a field, the usual menu stays, for copying and pasting.

A sidebar of your own

Right-click the sidebar, or use the card in Settings, to move pages up or down within their group and untick the ones you never open. It counts at once, and Back to default undoes it all. Settings always stays, and a hidden page is still one Ctrl+K away.

In the tray

A crash reaches you, even with the window closed.

The app sends a notification when a container stops unexpectedly or becomes unhealthy. Stopping one yourself, from the app, the terminal or compose down, never triggers one.

Notifications elsewhere

On your phone, through a service you pick.

Add a channel under Settings, choose which notifications go there, and send a test before you save. Tokens and webhook addresses are stored encrypted on this computer, and the app never shows them again; change the server of a channel and you type them again. The title and text of a notification go along, with container and machine names, and the dialog says so.

Fig. — A channel for ntfy: the topic stays hidden, and a new channel gets the problems, not the reports.

Your own rules

You decide what counts as trouble.

A rule watches one container or a whole compose project: CPU or memory above a threshold for some minutes, something in the log, or an event such as crashed or started three times in ten minutes. The log line that set it off stays in the app, with passwords and tokens covered up; a channel elsewhere only hears that the rule went off. A rule on a measure or an event can also restart or stop the container: right before, the app checks whether that still makes sense, it does it at most three times an hour, and it never acts on what a log says.

Fig. — Rules on measures, the log and events; "then restart" and "then stop" do something about it.

All features

Try it on your own daemon.
Free and open source. Windows installer, Linux AppImage, or npx on any platform.
Download for Windows
$ npx docker-client-mx