VIGIL MESH

Documentation

Troubleshooting connection problems

A VIGIL-MESH node opens no inbound port: it always establishes its connections outward, joins the network immediately through a relay, then switches to a direct path as soon as it finds one. Most blockages therefore come down to one of two causes: either the node did not complete its enrolment and has no address, or it is properly connected but stays on the relay for lack of a direct path. This page tells them apart and gives, for each, a course of action.

Quick checklist

Before any deeper diagnosis, run through these checks in order. They isolate the layer at fault in a few minutes: the first one that fails tells you where to look.

  1. 1
    Device status in the consoleDoes the device show as active? If it is pending approval, suspended, revoked or expired, it will not connect until that status is resolved.
  2. 2
    Assigned addressDoes the node carry a mesh address (100.64.0.0/10 range) on its interface? Without an address, enrolment did not complete — see the dedicated section below.
  3. 3
    Network egressCan the machine reach the outside? The node needs outbound access to the vigie. If UDP is filtered, the tcp-only profile takes over; the detail is in /en/docs/depannage-nat.
  4. 4
    Peer reachabilitySend a ping to the address of another active device on the same network. A reply confirms the path is established, even if it still goes through the relay.
  5. 5
    Topology in the consoleThe console shows the inventory and the links, refreshed roughly every 30 seconds. Let one cycle pass before concluding a device is missing.

The node gets no address

If the VIGIL-MESH interface carries no 100.64.x address, the node did not complete its enrolment. Until the device is admitted into the network and approved, no address is assigned to it and it cannot exchange with anyone.

SymptomProbable causeFix
No 100.64.x address on the interfaceEnrolment not finalisedResume the enrolment procedure through to the end and check that the device shows up in the console.
Device shown as “pending” in the consoleMissing approvalA workspace administrator must approve the device. As long as it is pending, it remains without an address.
Address gone after a whileDevice suspended, revoked or expiredCheck the status in the console; reactivate or re-enrol the device as appropriate.
The device appears nowhereEnrolment never reached the vigieCheck network egress first (next section and /en/docs/depannage-nat), then restart enrolment.

No direct path: the node stays on the relay

A node with a valid address that reaches its peers fine may nonetheless route all its traffic through the vigie rather than over a direct path. Connectivity works — the relay guarantees an exchange always gets through — but the direct path could not be established. This is acceptable and sometimes unavoidable; it is not a failure strictly speaking, but it adds a detour through the vigie.

The mechanism is described in detail in /en/docs/chemins-relais: the node keeps trying to open a direct peer-to-peer path, and switches to it as soon as it succeeds. Common reasons this switchover fails:

  • UDP filtered on both sides: without UDP, the direct path cannot form and traffic stays on the tcp-only profile via the vigie. See /en/docs/depannage-nat.
  • Strict (symmetric) NAT on either peer: direct traversal fails and the relay takes over.
  • An access policy that does not allow this link: the direct link will not be attempted if policy forbids it — check the access rules in the console.
  • Switchover simply not happened yet: the very first traffic goes through the relay by design. Give it a few seconds and watch whether the direct path gets established.

Frequently asked questions

My node is reachable but slow: is that a problem?
Often not. If traffic still goes through the relay, it takes a detour through the vigie. Check in /en/docs/chemins-relais whether a direct path could be established; otherwise, make sure UDP is not blocked (/en/docs/depannage-nat).
Do I need to open a port on the node for it to connect?
No. A node opens no inbound port: it always establishes its connections outward. There is no inbound firewall rule to create on the node side. What matters is egress.
The device does not appear in the console, what should I do?
First wait one refresh cycle (~30 s). If it is still missing, enrolment probably never reached the vigie: check network egress and restart enrolment.
After switching Wi-Fi, were my connections cut?
No, the session migrates. Traffic may briefly go back through the relay while a direct path is found again, but ongoing connections are not reset.
Read nextNAT and firewalls: what to check