When the print dialog opens: choose "Save as PDF", then uncheck "Headers and footers" under More settings.
Scout Probe — Deployment Requirements and Installation Guide
What the Scout Probe is
The Scout Probe is a lightweight Linux agent you install inside your network. It sends results to Orizon RECON over a single outbound HTTPS connection.
- Discovers every device on the networks you point it at
- Identifies operating systems and services
- Checks for vulnerabilities
- Can assess Active Directory
Before you start
You will need four things:
- One Linux virtual machine for every network you want to scan
- Root access to those machines (sudo)
- Outbound HTTPS access from them to the internet
- An Orizon RECON account with internal scanning enabled
How many probes do I need
If you have five networks, you need five VMs, each running its own Scout Probe, each with an interface in the network it scans. One probe cannot cover another network properly, and it is not designed to.
Why one per network
A probe identifies devices on its own network at a level it simply cannot reach across a router:
- It uses ARP to see every live device and read its hardware MAC address. ARP does not cross routers. On its own network the probe finds devices that answer nothing else — printers, cameras, embedded controllers, appliances.
- A device in another network can only be reached through a router, so the probe cannot verify its MAC address. Those devices are recorded as unverified.
- Where a probe is pointed at a network it cannot reach at all, RECON marks that network unreachable rather than reporting an empty result. A missing probe shows up as an honest gap, never as a clean bill of health.
What a probe can and cannot see
The same probe, three situations. Only the first produces a complete result.
Work it out
Count your networks. Not your sites, not your buildings — your network segments, meaning each VLAN or subnet with its own address range.
You need one VM and one probe for each of them.
Examples
| One office, one flat network | 1 VM, 1 probe |
| One office with three VLANs — servers, clients, guest | 3 VMs, 3 probes |
| Three branches, each with two VLANs | 6 VMs, 6 probes |
| Headquarters with five VLANs, plus a separate isolated OT network | 6 VMs, 6 probes |
If you are not sure how many networks you have, ask whoever manages your firewall or switching for the list of VLANs and subnets in use. That list is your probe count.
They still work as one system
Deploying several probes does not mean managing several disconnected scans, and it should not mean paying several times. Create a probe group first and install every probe into it: the group draws on one shared licence pool, and a single group scan covers all its members at once, combining their assets, vulnerabilities and network topology into one result, while still letting you view any individual probe or site on its own. If you have more than one network, the group is the cheaper and simpler arrangement, and it is far easier to set up at the start than to reorganise later.
One probe, as RECON shows it
Each probe reports its own host, target network, discovered devices and live heartbeat.
Host requirements
Each probe needs its own machine. The requirements below apply to every one of them.
Virtual machine sizing
| vCPU | RAM | Storage | |
|---|---|---|---|
| Minimum | 1 | 2 GB | 8 GB |
| Recommended | 2 | 4 GB | 16 GB |
Measured in production
These are measured from probes running in production, not estimates:
Operating system
The probe requires Linux with systemd, on x86-64 or ARM64. There is no distribution allowlist. The installer does not install packages and does not branch on distribution. If a system has systemd and one of those two CPU architectures, the probe runs on it. The list below is guidance on what is known-good, not a boundary of what is permitted.
Known-good distributions
| Distribution | Versions |
|---|---|
| Ubuntu | 20.04 LTS, 22.04 LTS, 24.04 LTS, 25.04, 25.10 |
| Debian | 11 (Bullseye), 12 (Bookworm), 13 (Trixie) |
| Red Hat Enterprise Linux | 8, 9, 10 |
| Rocky Linux | 8, 9, 10 |
| AlmaLinux | 8, 9, 10 |
| CentOS Stream | 9, 10 |
| Oracle Linux | 8, 9 |
| Amazon Linux | 2023 |
| Fedora | 41, 42, 43 |
| openSUSE Leap | 15.5, 15.6 |
| SUSE Linux Enterprise Server | 15 (SP4 and later) |
| Arch Linux | rolling |
| Raspberry Pi OS | 12 (Bookworm), 64-bit only |
Not supported, and why
| Alpine Linux | uses OpenRC, not systemd |
| Devuan, Void, Gentoo with OpenRC | no systemd |
| Windows Server, any version | Linux only |
| macOS | Linux only |
| Docker or Podman containers | no systemd, and no direct access to the host network |
| 32-bit systems (i386, armv7) | x86-64 and ARM64 only |
| Raspberry Pi OS 32-bit | use the 64-bit image instead |
Before you install
Does this machine qualify?
If you are unsure whether a machine qualifies, run these two commands on it. If both succeed, the probe will install:
systemctl --version
uname -mThe first must print a version. The second must print either x86_64 or aarch64.
What must already be present
The installer needs curl, systemctl, ip, grep, awk, sed and useradd. All of these ship by default on every distribution listed above, including minimal server installs. The installer adds no packages and modifies no repositories.
Privileges
Installation and operation require root, via sudo. The installer stops immediately if the host is not Linux, is not running systemd, or is not run as root.
Where to place it
Firewall whitelisting
The probe needs to reach recon.orizon.one on TCP port 443, and nothing else.
If your firewall supports FQDN or domain-based rules — most modern firewalls do, including Fortinet, Palo Alto, Check Point, Sophos, WatchGuard and pfSense — create the rule against the hostname:
recon.orizon.oneThis is the correct approach and the only one that stays accurate over time.
If your firewall cannot do hostname rules, allow the following IPv4 ranges outbound on TCP 443. These are the published ranges of the content delivery network serving the platform:
| 173.245.48.0/20 | 103.21.244.0/22 | 103.22.200.0/22 |
| 103.31.4.0/22 | 141.101.64.0/18 | 108.162.192.0/18 |
| 190.93.240.0/20 | 188.114.96.0/20 | 197.234.240.0/22 |
| 198.41.128.0/17 | 162.158.0.0/15 | 104.16.0.0/13 |
| 104.24.0.0/14 | 172.64.0.0/13 | 131.0.72.0/22 |
If your network uses IPv6, also allow:
| 2400:cb00::/32 | 2606:4700::/32 | 2803:f800::/32 |
| 2405:b500::/32 | 2405:8100::/32 | 2a06:98c0::/29 |
| 2c0f:f248::/32 |
These ranges are maintained publicly and can be re-checked at any time at https://www.cloudflare.com/ips/. If your firewall supports automatic external address lists, point it at that source so the rule updates itself.
If you use an outbound proxy, the probe honours the standard HTTPS_PROXY and NO_PROXY environment variables. Set them in the service environment before starting the probe.
Verify the rule works from the probe host before installing:
curl -I https://recon.orizon.oneA response beginning HTTP/2 200 confirms the path is open. Anything else — a timeout, a refused connection, a certificate error — means the firewall rule is not yet correct, and the installer will fail for the same reason.
Inbound, none
Traffic on your internal network
To do its job, each probe generates ordinary scanning traffic on its own network:
- ARP and ICMP, to establish which addresses are live
- TCP connections to common service ports, for discovery and fingerprinting
- mDNS and NetBIOS queries, to resolve device names
- SMB, SSH and HTTP probes, to identify operating systems and applications
Installation
Repeat every step below once per network. Each network gets its own VM, its own probe, and its own API key.
If you are scanning more than one network, create the probe group first. Go to Probes, choose Add Probe Group, and name it for the organisation or site — for example Acme HQ or Milan Office. Every probe you then install into that group shares one licence pool covering the whole estate. If you have only one network, a single probe is enough and no group is needed.
Then create the probe itself. Go to Probes and choose Add Probe — from inside the group if you created one, so the probe joins it automatically. Give it a name that identifies its location, for example HQ-Servers-VLAN10 or Milan-Clients. Copy the installation command shown; it contains credentials unique to this probe. Treat the API key like a password. Each probe has its own — do not reuse one across several machines.
Provision a Linux VM meeting the requirements above, on the network it will scan. Confirm it can reach the platform:
curl -I https://recon.orizon.onePaste the command you copied in Step 1, as root. It looks like this, with your own probe ID and tokens filled in:
curl -fsSL https://recon.orizon.one/scout/install.sh | sudo bash -s -- \
--probe-id <probe-id> \
--api-key <api-key> \
--install-token <install-token> \
--endpoint https://recon.orizon.one \
--binary-version <version>Copy the command from RECON rather than typing it. It carries credentials unique to this probe, and the install token is single-use. The installer checks the host, downloads the correct binary for your CPU architecture, installs it, registers a systemd service, and enrols the probe with your account. It takes about a minute. It is available in English, Italian and Spanish.
It installs to these locations
Step 6 in RECON
Confirm the target networks and start. Progress is live.
The probe appears in RECON as Online within moments of installation. On the host itself, you can open the built-in status panel:
sudo scout dashboardThe service starts automatically on boot and restarts on failure. Probes update themselves when a new version is released. No action is required, and updates only apply while a probe is idle, never during a scan.
In RECON, open the probe and set its target networks in CIDR form, for example 192.168.1.0/24. The probe suggests the networks it can see, and you can add more.
Only list the network this probe sits on. Pointing a probe at a network it cannot reach directly produces unverified results or an unreachable report — deploy a probe there instead.
Select the probe, choose Scan, confirm the networks, and start it. RECON shows live progress and, when it finishes, the full asset inventory, vulnerabilities, network topology and compliance view.
Repeat Steps 1 to 6 for every remaining network, adding each probe to the same probe group you created in Step 1. Once they are all online, a single group scan covers the whole estate and reports as one result, while still letting you view any individual probe or site on its own.
One probe per network, one estate in RECON
Each probe holds its own segment and its own key. The group is what makes them one result.
Optional: Active Directory assessment
If you want the identity assessment as well as the network scan, the probe additionally needs:
The probe additionally needs
- Network reachability to a domain controller
- LDAP on port 389, or LDAPS on port 636 where encryption is enforced
- A standard domain user account, read-only
Add the credentials in RECON on the probe's Active Directory page, test the connection, then enable Active Directory when you start a scan.
You only need this on a probe that can reach a domain controller, not on every probe.
Troubleshooting
| The probe shows as Offline | Check the service with sudo systemctl status scout. Then confirm outbound HTTPS to recon.orizon.one is permitted by your firewall, using the curl test above. |
| The installer fails to download | Confirm the host has working outbound HTTPS to recon.orizon.one. See the firewall section — a rule pinned to specific IP addresses is the most common cause. |
| A network reports as unreachable | That network has no probe on it, or the probe cannot reach it. Deploy a VM with a Scout Probe on that network. This is the expected fix, not an error to work around. |
| Fewer devices found than expected | Either the network has no probe of its own and is being scanned across a router, or an IDS, IPS or NAC is blocking scan traffic. Check which, and deploy a local probe or allowlist the probe's IP address. |
| The installer refuses to run | It requires Linux, systemd and root. Confirm all three using the two commands given earlier. |
| Active Directory returns no results | Verify the probe can reach a domain controller on port 389 or 636, and that the account credentials are correct. |
Removing a probe
Run the installer with the uninstall flag, as root:
curl -fsSL https://recon.orizon.one/scout/install.sh | sudo bash -s -- --uninstallThis removes the binary, the configuration and the service.
Quick reference
| Probes needed | One VM with a Scout Probe on every network segment you want to scan |
| Supported operating system | Linux with systemd, x86-64 or ARM64 |
| Minimum VM | 1 vCPU, 2 GB RAM, 8 GB storage |
| Recommended VM | 2 vCPU, 4 GB RAM, 16 GB storage |
| Actual probe footprint | About 230 MB memory idle, peaking near 1.25 GB during scans, 26 MB on disk |
| Required outbound | HTTPS port 443 to recon.orizon.one — allow by hostname where possible, not by fixed IP |
| Required inbound | None |
| Privileges | Root |
| API key | One unique key per probe, never shared between machines |
| Active Directory | Optional, standard read-only domain user, LDAP 389 or LDAPS 636 |