Topology (platform admin)
The whole estate at a glance — every Docker context, the managed engines they reach into, the metal underneath, and what depends on what. It lives under Admin → Topology and requires the platform-admin role.
Read from reconciled state, not from the daemons. Unlike a single context's own diagram, which is a live read of one Engine API, this one is seven database queries against what the platform already believes — cheap enough to render inline rather than behind a frame, and the reason it can sit on one page instead of costing a round trip per context. Open a context for the live picture of it.
Four rows, always in this order, because the order is the argument:
| Row | What's in it |
|---|---|
| Internet | Everything outside the estate — the one thing every public context is reachable from. |
| Contexts | The clusters the platform deploys onto. Each box names what it's carrying — containers, revisions — and a dashed border marks one an organization owns rather than the platform pool. |
| Data services | Managed databases and object stores, and which contexts reach into them. |
| Hypervisors | What provisioned the machines underneath a context, where one exists. |
Boxes within a row are ordered to minimise crossed lines, not alphabetically — contexts that share a dependency are pulled toward each other. Some crossings survive regardless (two contexts reaching two engines in opposite orders cross no matter how you arrange them), and the count is reported next to the dependency total rather than hidden.
Nothing registered yet says so instead of drawing an empty diagram — there's nothing wrong with a fresh platform having nothing to show here.