Access Control and Audit Log¶
The Hub decides who may reach each registered site, and it records what its users and sites do. Access is given to roles, as three grants that each name tags: Console opens the site's console through the Hub, Jobs lets the person run jobs there, and MCP lets an AI assistant work there through the Hub. Sign-ins, grants, tag and site changes, jobs and certificate steps are all written to the Hub audit log.
Per-Site Access Control¶
Anyone who can sign in to the Hub sees the Sites Dashboard, but it lists only the sites where their role holds Console. Hub administrators see and manage every site, but hold Console, Jobs and MCP only through their role, like anyone else. Only Hub administrators manage tags and role grants.
- Site Access Grants describes the three grants, how a role's grants reach sites through tags, the Site Access section of a role, the read-only Access dialog, and what happens when a Hub user or role changes.
- Tags describes the Tags page, the built-in All sites tag, and how a new site is given its tags when it enrolls.
Audit Log¶
The Hub writes each audit event to the hub_audit_entry table, which the Audit Log page searches, and to the hub-audit.log file for shipping off the host.
- Hub Audit Log describes both destinations, the shape of an entry, the Audit Log page, and shipping the file to a SIEM.
- Hub Audit Event Types lists every event the Hub records and the details each one carries.
Tunnel Takeover Detection¶
TUNNEL_TAKEOVER_DETECTED events deserve special attention. They are written when:
- A site already has an open tunnel.
- A new tunnel arrives presenting the same client certificate (the same subject CN).
- The Hub accepts the new tunnel and force-closes the old one: the newer connection always wins.
The legitimate cause is a dirty disconnect: a TCP-level interruption that did not close the WebSocket cleanly, followed by a quick reconnect before the heartbeat-loss timer expired. This happens routinely and is not by itself an alarm signal.
What is an alarm signal is a storm of repeated takeovers, for example two takeovers within 60 seconds, repeating every few minutes. That pattern indicates two QIE installations presenting the same client certificate and taking the tunnel from each other in turn. The most likely explanation is a copied certificate: an attacker has the site's client certificate and is connecting from a different host than the legitimate site.
The recommended response is to revoke the certificate at once with the reason SUSPECTED_COMPROMISE. The legitimate site then fails to reconnect and its administrator reports the outage, and you can investigate the source of the leak before re-enrolling the site.
This signal is reliable because the Hub always knows the source address of every incoming tunnel. The takeover addresses are recorded in source_ip on every event, and a SIEM rule on the pattern is the recommended detection.