# Devices and SNMP profiles

Add network devices over SNMP v2c or v3, learn how Tracix matches each one to a built-in device profile, and read the device page laid out for its kind.

Tracix monitors network devices over SNMP through a collector on your network. This page shows how to add devices, how Tracix decides what to poll on each one, and what you find on a device's page.

## Before you begin

* A collector installed on a host that can reach your devices on UDP port 161. See [Install the collector](https://tracix.io/docs/collect/collector/install).
* SNMP enabled on the devices, with read-only credentials the collector can use: a v2c community, or an SNMPv3 user.
* Permission to manage integrations, which admins have by default, because devices are added through an integration on a collector policy.

> **One collector per policy polls**
>
> SNMP polling runs on one collector of each policy at a time. If that collector stops reporting, another collector of the same policy takes over. Every collector of the policy listens for traps. When collectors sit in networks that can't reach each other, give each network its own policy.

## Add devices

**1. Open the SNMP integration**

In your workspace, open **Network › Devices & Inventory** and select **Run discovery**. The **SNMP** integration opens on its **Configure** step.

**2. Choose what to scan**

Under **CIDRs**, enter the ranges to sweep, such as `10.20.0.0/24`. To add one known device, enter its address as a `/32`, such as `10.20.0.5/32`. Enter a **Site**, such as `dc-east`: it fills the site column on **Devices & Inventory** and groups devices on the topology map.

**3. Enter the SNMP credentials**

Choose the **SNMP version**. For v2c, enter the **Community (v2c)**. For v3, open **Advanced** and enter the **User (v3)** with its auth and privacy protocols and passwords; the security level follows the passwords you enter. [SNMP polling and traps](https://tracix.io/docs/collect/network-data/snmp) covers every field, including the trap listener.

**4. Pick the policy**

Under **Where to add this integration**, choose the policy whose collectors can reach the devices. Select **Review**, check the configuration, then select **Connect**. Every collector on the policy picks up the change with the next configuration push.

**5. Confirm what discovery finds**

Each discovered device waits in **Network › Discovery Candidates** until you confirm it. Turn on **Auto-accept discovered devices** in the integration to skip that step. See [Discovery and topology](https://tracix.io/docs/network/discovery-topology).

Confirmed devices appear under **Network › Devices & Inventory**, with a strip at the top that counts them by type. The integration polls each device every 5 minutes by default (**Collection interval**) and sweeps the ranges again every 15 minutes (**Rescan interval**, an advanced field).

## How Tracix picks a device profile

A device profile says which SNMP objects to poll on a kind of device and how to map them to metrics. Tracix ships 174 built-in device profiles, plus 21 building blocks, such as IF-MIB, LLDP-MIB and HOST-RESOURCES-MIB, that profiles extend.

During discovery the collector reads each device's `sysObjectID`, `sysName` and `sysDescr`. Tracix then picks the most specific profile, in this order:

1. **Exact match**: a profile lists the device's `sysObjectID`.
2. **Longest prefix**: the most specific wildcard the `sysObjectID` starts with, so a model profile wins over a vendor-wide one.
3. **sysDescr pattern**: a profile's regular expression matches the device's `sysDescr`.
4. **No match**: the device is flagged as needing a profile (see the next section).

Whatever matches, every device is polled for standard interface metrics (IF-MIB), system information (SNMPv2-MIB) and link aggregation membership. The **Device profiles** field on the integration is empty by default, which means automatic matching; select profiles there only to narrow a mixed environment.

The built-in profile definitions are derived from the Kentik SNMP profile library, licensed under the Apache License 2.0.

## Devices no profile matches

When **Generate profiles for unknown devices with Brain** is on (the default), the collector walks one device of each unknown make and model once, and Brain writes a profile from that walk. Every device of the same kind then uses it. This needs an AI model connected to your workspace; see [Model providers](https://tracix.io/docs/ai/model-providers). Without one, the device keeps the standard interface and system metrics, plus any building blocks its walk shows it supports.

To start generation for candidates that are waiting, open **Network › Discovery Candidates** and select **Parse with Brain**. Follow progress under **AI & Automation › Built-in agents › SNMP profile generator**.

## The device page

Select a device on **Devices & Inventory** to open its page. The layout follows the device's type:

| Device type                                    | What the page shows                                                                                           |
| ---------------------------------------------- | ------------------------------------------------------------------------------------------------------------- |
| Routers and switches                           | Ports and optics, traffic, BGP, OSPF, routing table, ARP table, VLANs, MAC address table and link aggregation |
| Firewalls                                      | Sessions and session rate, VPN tunnels, HA state, policies and virtual systems                                |
| Access points and wireless controllers         | Clients, channel utilization, radios, SSIDs, access points and licenses                                       |
| Load balancers                                 | Virtual servers, pools and members, connections, requests and expiring certificates                           |
| Servers and storage                            | Load, file systems, disks, volumes, IOPS, latency and replication                                             |
| UPS units and PDUs                             | Battery, runtime, load, voltage, phases and outlets                                                           |
| Environment sensors                            | Temperature, humidity, dew point, contacts and alarms                                                         |
| Printers                                       | Status, supplies, trays, covers and pages                                                                     |
| Anything else, such as DNS and IPAM appliances | Interfaces, traffic, uptime and hardware                                                                      |

Every page also has **Traps**, **Details** and **Alerts** tabs. A value the device doesn't report shows as `-`, never as 0. A device Tracix hasn't heard from in 30 minutes is marked stale on the inventory and on the topology map.

### Correct a device's type

Tracix infers each device's type from its profile, its hardware identity and its telemetry. When it gets one wrong, open the device page, choose the right value under **Device type** and select **Pin type**. The pin sets the inventory type, the page layout and the device's role on the topology map. Choose **Automatic** and select **Clear pin** to let Tracix decide again. Pinning needs write access to network devices.

## Alert on device problems

Tracix ships ready-made monitors for network devices, such as interfaces going down, interface errors and discards, CPU, memory, power supplies, fans and temperature, and traps such as link down or a BGP peer going down. See [Monitors](https://tracix.io/docs/alerting/monitors) to turn them on and route their alerts.

## Next steps

- [Discovery and topology](https://tracix.io/docs/network/discovery-topology): Confirm candidates and map LLDP links

- [SNMP polling and traps](https://tracix.io/docs/collect/network-data/snmp): Credentials, trap listeners and ports

- [Back up configurations](https://tracix.io/docs/network/ncm/backups): Add devices to configuration management