Passive scanner (WIP)
Watches a network feed — a SPAN port, a mirror or a tap — and infers hosts from the traffic that crosses it. It transmits nothing, which makes it the one discovery source you can point at a segment where an active scan would be unacceptable.
| Type name | passive_scanner |
| Creates | Endpoints |
| Runs on | A sensor attached to a mirrored feed |
| Presence flag | passive_scanner_exits |
| Field prefix | passive_scanner_ |
The type is defined and its fields are declared, but nothing behind it runs today. The sensor image no longer ships a capture engine and does not start capture for this type, the sensor has no collector that runs its scheduled tasks, and the console's identification stage for it is still empty. A run produces no asset change entries. Treat this page as a description of the intended integration, not of a working inventory path.
Why passive discovery earns its place
An active scan is authoritative but intrusive, and there are always segments where it is not allowed: OT and process networks, medical devices, anything with a vendor support contract that forbids unsolicited traffic. Those segments are exactly where unmanaged assets accumulate.
A passive sensor sees them anyway, because a device that communicates is visible whether or not it is willing to be probed:
| Passive scanner | Active scanner | |
|---|---|---|
| Traffic generated | None | Probes every address in scope |
| Finds | Anything that talks on the monitored feed | Anything reachable from the sensor |
| Misses | Anything silent, or on a segment with no feed | Anything firewalled off, or that ignores probes |
| Evidence | Inferred from observed traffic | Established by direct response |
| Freshness | Continuous | Per scan window |
| Safe on fragile networks | Yes | Not without agreement |
They are complements, not alternatives. Passive tells you something is there; active tells you what it is.
How it is meant to work
The design runs Zeek on the sensor against its capture interface, and turns the logs Zeek produces into assets:
Capture was tied to whether an enabled passive scanner instance exists for the sensor: start Zeek when one is enabled, stop it when none is. That switch is disabled in the current sensor, and the sensor image does not include Zeek, so enabling an instance today turns nothing on.
The protocols that carry identity are the ones that matter here — DHCP requests and offers name a host and its vendor class, ARP and Ethernet headers give addresses, and ordinary IP conversation reveals which addresses are live.
Document the exact Zeek log set collected, the upload cadence, and the mapping from each log field to an asset field, once capture is back in the sensor image and the identification stage is implemented.
Requirements
The feed
| Need | Why |
|---|---|
| A SPAN, mirror or tap delivering the traffic you care about | The sensor can only infer from what its interface receives |
| A capture interface on the sensor host, in promiscuous mode | Mirrored frames are addressed to other hosts |
| Host networking on the sensor container | The capture engine reads the host's interface directly |
| Sustained throughput headroom | Capture cost scales with the feed, not with the estate |
A passive sensor sees exactly what its feed carries and nothing else. Mirroring an internet edge tells you about north–south traffic and nothing about two hosts talking inside a VLAN. Decide which segments you need visibility into, then arrange a feed per segment — one sensor at the edge is not a substitute.
In this deployment
| Need | Why |
|---|---|
| A registered, healthy sensor on the mirrored segment | See Sensors |
| An enabled passive scanner instance bound to that sensor | Intended to switch capture on; it does not today, see the warning at the top of this page |
Configuration
| Field | Required | Notes |
|---|---|---|
| Name | Yes | Label for this instance |
| Sensor | Yes | The sensor attached to the feed |
The template today presents the same fields as the active scanner — subnets, discovery ports, ICMP, OS detection, HTTP probe, scan rates. None of them describes anything a passive listener does. Fill in the name and description; treat the rest as placeholders until the form is trimmed to what this type actually consumes.
Replace the inherited template with passive-specific settings — the capture interface, the address space to consider internal, and the retention of observed records — and document them here.
Exported fields
Four fields are declared. They are deliberately minimal: passive observation establishes presence and addressing, not platform detail.
| Field | Type | Holds |
|---|---|---|
passive_scanner_exits | Boolean | Presence flag marking this asset as seen by passive capture |
passive_scanner_address | IP | Address the host was observed using |
passive_scanner_hostnames | Text | Name observed on the wire — DHCP client hostname and similar |
passive_scanner_last_online | Date/time | When the host was last observed |
Extend the field set once identification is implemented — vendor class, MAC and observed services are all present in the capture and are the obvious next additions.
See also
- Active scanner — the authoritative counterpart
- Sensors — placement, enrolment and health
- Integrations overview — the shared configuration model