The whole swarm, at a glance.
Nodes that are ready, managers, services below their count, failing tasks, and the load of the swarm, live and over the last 24 hours. Docker has no swarm-wide statistics, so the app measures every node itself.
What you can do
Counts that open their page
Nodes that are ready, managers (with a warning for an even number), services below their desired count, failing tasks, and secrets and configs. Each count opens its page.
Load across the swarm
CPU, memory and network, live and over the last 24 hours, for the whole swarm, a node or a service. Docker has no swarm-wide statistics, so the app measures every node through its Docker context.
Measured in the background
The app keeps measuring once a minute in the background, so the 24-hour history also fills while the window is closed. You can switch it off under Settings › General.
The managers
Every manager with its address, which one is the leader, and whether it can be reached.
Counts you can trust
The daemon keeps reporting tasks on a node that is down as running. The counts of the app leave them out, so a service that is short of replicas looks short.
Always in the status bar
The status bar shows the engine, the swarm role, the host and the Docker version, and switches between contexts. At the far right it shows the version of the app.
Per node
docker stats sees one daemon. The app asks every node.
The CLI only sees the daemon you are connected to. The app matches each node to a Docker context on this computer by its node ID, and measures it there. That is how the overview shows the load of the whole swarm, and per node its CPU, memory, network and tasks side by side.
All features