VIGIL MESH

Documentation

Real-time topology

The console displays a living map of your mesh: the nodes that are online and the paths actually elected between them. It is the view that answers the everyday questions — who is reachable, where the traffic goes through, where a friction point sits. This page explains what the graph shows, how often it refreshes and how to read it for diagnosis.

The real-time graph

The topology represents the nodes of your networks and the links that actually connect them at the present moment. A node stands for an enrolled machine; a link stands for the path elected between two nodes — the one traffic really takes, not a theoretical list of possible connections.

VIGIL-MESH first establishes an immediate connection through a blind relay, then attempts a direct peer-to-peer path in the background and switches to it without interruption when it succeeds. The topology reflects this outcome: it shows the currently elected path — direct when it was obtained, relayed otherwise — not the intent. The full selection and migration mechanism is described in /en/docs/chemins-relais.

Nodes

Each online machine appears with its internal name. A node absent from the graph is a node that is not participating right now — offline, suspended or unreachable.

Elected paths

Links show where traffic between peers really goes: a direct path when obtained, or a relay fallback. It is state, not configuration.

Relays

The vigies that provide the immediate connection and the fallback appear as transit points when a direct path could not be established.

Freshness: on the order of thirty seconds

The view is not frozen: it refreshes continuously. Two dynamics combine — a periodic refresh, on the order of thirty seconds, which keeps the whole picture current, and an immediate update on every path change, as soon as a migration happens or a node appears or disappears.

  • Expect a display latency on the order of thirty seconds for the overall state: a node that has just connected or left appears or disappears within that window.
  • A path change — a switch from a relay fallback to the direct path, for example — is reflected as soon as it happens, without waiting for the next periodic cycle.
  • A passing gap between the graph and reality during that short interval is normal: the view converges, it does not claim to be perfectly instantaneous.

Reading the topology to diagnose

The topology is the first diagnostic reflex: it locates a problem before you even open the logs. A few simple readings quickly orient the search.

  • A node that is expected but absent from the graph: the machine is not participating — check its enrollment, its status and its client-side connectivity.
  • A peer that stays on a relayed path when a direct one was expected: a NAT or a firewall is probably blocking direct establishment — the lead is on the network side.
  • Several peers all falling back to the same relay: look at that relay's availability and how the vigie fleet is distributed.
  • A node repeatedly appearing and disappearing: instability of its network access rather than a policy problem.

The topology tells you where to look; it does not replace the tools that tell you why. To dig deeper — health endpoint, logs, connectivity checks and NAT leads — see /en/docs/depannage-diagnostics.

Is the graph truly instantaneous?
No, and it does not claim to be. It refreshes periodically on the order of thirty seconds and immediately on every path change; a slight lag during that interval is normal.
Why does a peer show a relayed path rather than a direct one?
Because the direct path could not be established at that moment, often due to a NAT or a restrictive firewall. The connection remains handled by the blind relay; the switch to the direct path happens as soon as it becomes possible.
Can I change a path from the topology?
No. The topology is an observation view. Path choices are set through path profiles and policy, not from the graph.
Read nextThe audit log