SNMP device

The most universal type: almost every piece of network hardware speaks SNMP. Switches, routers, NAS, UPS and printers all qualify. SNMP v1, v2c and v3 are supported.

Network ports to open

From the DumbMonit server (or its relay agent) to the device:

Port	Protocol	Purpose
161	UDP	SNMP (default)

What it watches

The collection profile is detected automatically from the device’s sysObjectID (and, for UPS, its sysDescr). Ten profiles ship with the product; a device can match several, and the profiles include each other.

Profile	Applies to	Metrics (prefix dumbmonit_)
System (SNMPv2-MIB)	Every device	system_uptime_seconds, system_info (name, description, location, contact as labels)
Network interfaces (IF-MIB)	Every device, as a fallback	if_octets_in/out, if_packets_in/out, if_errors_in/out, if_discards_in/out, if_admin_status, if_oper_status, if_speed_bps, per interface
Host resources (HOST-RESOURCES-MIB)	Net-SNMP (Linux, BSD, macOS), the Windows SNMP agent, Synology, QNAP, Solaris	host_uptime_seconds, host_processes, host_users, cpu_load_percent, storage_bytes_total/used per filesystem, memory_bytes_total/used, load_average (1, 5 and 15 minutes, Net-SNMP only)
UPS (UPS-MIB)	APC, Eaton, MGE, CyberPower, Riello, Socomec, Tripp Lite, Vertiv, Salicru, and any agent announcing UPS-MIB	ups_battery_status, ups_battery_seconds_on, ups_battery_minutes_remaining, ups_battery_charge_percent, battery volts/amperes/temperature, input volts/hertz/line bads, ups_output_source, output volts/hertz/watts, ups_output_load_percent, ups_alarms_present
Printer (PRINTER-MIB)	HP, Kyocera, Epson, Canon, Brother, Lexmark, Ricoh, Xerox, Samsung, OKI, Konica Minolta	printer_supply_level/max per supply, printer_input_level/max per tray, printer_pages_printed, printer_status
Ethernet switch (EtherLike-MIB, LLDP)	Chosen by hand for any switch; included by the vendor switch profiles	Everything from IF-MIB, plus per port ether_fcs_errors, ether_alignment_errors, ether_symbol_errors, ether_late_collisions, ether_excessive_collisions, ether_mac_transmit_errors, ether_mac_receive_errors, ether_duplex_status (1 unknown, 2 half, 3 full), if_broadcast_packets_in, if_multicast_packets_in (64-bit), and lldp_neighbor_info (who is plugged where: remote_sysname, remote_port, remote_port_desc; the local port number is the second part of the index label). For the switch panel of the device page: switch_port_status (every physical Ethernet port, disabled and never-used ones included; admin label 1 enabled, 2 disabled), port_pvid (native VLAN, Q-BRIDGE-MIB), poe_port_status (POWER-ETHERNET-MIB: 1 disabled, 2 searching, 3 delivering, 4 fault, 6 other fault; class label), poe_power_budget_watts, poe_power_consumed_watts and poe_main_status
Zyxel switch	Every Zyxel sysObjectID (1.3.6.1.4.1.890)	Everything from Ethernet switch, plus cpu_load_percent (one-minute average), zyxel_cpu_5min_percent, memory_usage_percent, zyxel_firmware_info (firmware and model as labels), and on switches that publish the hardware monitor, zyxel_temperature_celsius and zyxel_temperature_high_threshold_celsius per sensor, zyxel_fan_rpm and zyxel_fan_low_threshold_rpm per fan
WatchGuard Firebox	Every WatchGuard sysObjectID (1.3.6.1.4.1.3097)	Everything from IF-MIB, plus cpu_load_percent (one-minute average), watchguard_cpu_5min_percent, watchguard_cpu_15min_percent, watchguard_active_connections, watchguard_bytes_in/out, watchguard_packets_in/out, watchguard_stream_requests_dropped, watchguard_ipsec_tunnels, watchguard_ipsec_tunnel_kbytes_in/out per peer address, and on a cluster watchguard_ha_status, watchguard_ha_peer_status, watchguard_ha_error, watchguard_ha_peer_error
Stormshield SNS	Every Stormshield sysObjectID (1.3.6.1.4.1.11256)	Everything from IF-MIB, plus stormshield_ha_nodes, stormshield_ha_dead_nodes, stormshield_ha_active_nodes, stormshield_ha_links, stormshield_ha_faulty_links, stormshield_ha_sync_status, stormshield_ha_node_quality per node serial, stormshield_cpu_temperature_celsius per processor and stormshield_fan_rpm per fan
HPE / Aruba switch	1.3.6.1.4.1.11.2.3.7 (HPE ProCurve family) and 1.3.6.1.4.1.47196.4 (Aruba CX)	Everything from Ethernet switch, nothing proprietary

Counters (interface octets, pages printed) are stored raw: use rate() in a query to get a throughput. A device reboot shows as a counter reset, never as a fake spike.

The built-in rules that apply to SNMP devices: Device unreachable, High CPU, Disk almost full, Filesystem almost full (forecast), UPS on battery, UPS battery low, Port accumulating errors, Port flapping, Unusual CPU (baseline).

Zyxel switches

The standard part of the Zyxel profile (interfaces, Ethernet errors, LLDP) is common to every Zyxel switch. The CPU, memory and firmware figures come from the private branch that the current managed switches share under 1.3.6.1.4.1.890.1.15 (the XS1930 and GS1900 among them); older models and Zyxel firewalls lay theirs out differently, and those series simply stay empty there. The profile was checked against a real XS1930-10; its temperature and fan series follow Zyxel’s published hardware-monitor MIB but have not yet been seen on real hardware.

dot3StatsFrameTooLongs is deliberately not read: on the XS1930 it counts billions of perfectly healthy jumbo frames, and a series that always climbs on a healthy network only teaches people to ignore it.

WatchGuard Firebox

The profile reads the three MIBs WatchGuard publishes under its enterprise number: system statistics (CPU, connections, traffic), IPsec tunnels and high availability. The CPU figures are reported in hundredths of a percent and are scaled to a percentage. The OIDs come from WatchGuard’s MIBs and documentation; they have not yet been seen on a real Firebox, so a missing series means the model or Fireware version does not publish it. The high-availability values are the MIB’s raw codes. SNMP has to be enabled in the Firebox’s configuration; the 64-bit counters need SNMPv2c or v3.

watchguard_ipsec_tunnels counts the security associations established at this moment, site-to-site tunnels and mobile users together: it is shown, but no built-in rule alerts on it. The generic rules (device unreachable, high CPU, port errors, port flapping) apply. No REST API is used: this integration is SNMP only.

Stormshield SNS

The profile reads STORMSHIELD-HA-MIB (cluster nodes, dead nodes, faulty HA links, node quality) and STORMSHIELD-SYSTEM-MONITOR-MIB (processor temperature, fan speed) under 1.3.6.1.4.1.11256. A firewall outside a cluster reports a single node; a virtual appliance has no fan or sensor, and those series stay empty. The OIDs come from Stormshield’s MIBs; they have not yet been seen on a real appliance. The health columns of STORMSHIELD-HEALTH-MONITOR-MIB are text whose values the MIB does not list, so they are not read.

Two built-in rules apply on top of the generic ones: a dead HA node and a faulty HA link. Enable the SNMP agent (Configuration → Notifications → SNMP agent) and add a filter rule that lets only the DumbMonit server query it, as Stormshield recommends. No web API is used: this integration is SNMP only.

HPE / Aruba switches and Aruba Instant On

Aruba Instant On is a cloud-managed range. HPE publishes no public API for its portal that this integration could rely on, and no SNMP agent is documented for its access points, so DumbMonit cannot read the access points nor the Instant On portal. The Instant On 1930 switches do offer SNMP; when they announce an HPE sysObjectID of the ProCurve family (1.3.6.1.4.1.11.2.3.7), the HPE / Aruba switch profile applies the standard switch series (interfaces, Ethernet errors, LLDP neighbours) without any manual choice. If a switch announces another identifier, choose the Ethernet switch profile by hand: the result is the same. For Aruba access points, gateways and switches managed from the cloud, use Aruba Central.

Server management controllers (BMC)

A BMC’s own SNMP agent describes the BMC, not the server: its small ARM Linux, its memory, its network links. A Supermicro BMC, for instance, answers as Net-SNMP and gets the Host resources profile, which reads everything that agent publishes about itself. The server’s fans, temperatures, power supplies and drives are read over Redfish instead.

What to prepare on the device

1. Open the device’s web interface and find its SNMP settings, often under Network, Services or Administration.

2. Enable SNMP v2c and note the read-only community (public on many devices).

3. If the device filters by address, allow this server’s address.

4. In DumbMonit, enter the address and the community: the rest is detected automatically.

Warning

A community is not encrypted on the network: on a shared network, prefer SNMP v3, which authenticates and encrypts.

Credentials

Credential	Fields
SNMP v1 / v2c (community)	The read-only community.
SNMP v3	User name; optional authentication (MD5, SHA-1, SHA-224, SHA-256, SHA-384 or SHA-512 and a passphrase); optional privacy (DES, AES-128, AES-192 or AES-256 and a passphrase); optional context name.

Address: an IP or host name, for example 192.168.1.10. The default port is 161; write host:port to use another one.

Options

This type reads no options. Free tags are still available and are copied on every series as tag_<key>.

Common errors

Symptom	Likely cause
Unreachable right after adding	Wrong community, SNMP not enabled, or the device only accepts requests from allowed addresses. The device never answers a wrong community: it looks exactly like a device that is off.
Only system_* and if_* series	The sysObjectID matched no specific profile. IF-MIB and System apply to everything; HOST-RESOURCES, UPS, PRINTER and Zyxel need a known vendor OID. On a switch of another brand, choose the Ethernet switch profile by hand.
No interface traffic	The device does not expose the 64-bit ifXTable counters.
Empty scan	The scan tries one community on every address; devices with another community or SNMP v3 only do not answer.

Use Probe now on the monitor page: it reports the number of samples and the series names produced by one probe.