banshee/network-collection/network/mikrotik/snmp/health

Hardware health of a MikroTik device, by SNMP

Returns, for each sensor of a MikroTik router or switch, its measure (temperature, fan speed, voltage, current, power) or its state (a power supply ok or fail). Five watches are offered, each one set on the probe.

Versions

RouterOS 6 and 7, SwitchOS. The collector is generic across them: it reads whichever form the device serves. Which objects a model serves depends on its sensors.

How it works

The collector reads the health branch of MIKROTIK-MIB (mtxrHealth), in one of its two forms:

mtxrGaugeTable, served by RouterOS 7 and the late versions of RouterOS 6: one row per sensor, named by the device as /system health shows it (cpu-temperature, fan1-speed, psu1-state), with its unit. Temperatures are in degrees, voltages, currents and power in tenths, which the collector divides. A state reads ok for 0 and fail for 1, as RouterOS shows them; any other value reads state and the number.
The fixed objects of mtxrHealth, read when the table is empty: an older RouterOS, and SwitchOS. Each object present is one sensor, named after the object (cpu-temperature, board-temperature, psu-state, fan1-speed); temperatures, voltages and power are in tenths, the current in mA, and psu-state and backup-psu-state are ok when the agent says true.

A device that serves neither, such as a virtual router (CHR), stops the collector as misconfigured: never a green light without a measure.

The collector only reads: SNMP sends nothing but reads.

Prerequisites

SNMP enabled on the device (/snmp set enabled=yes on RouterOS, the SNMP tab on SwitchOS), reachable from the poller (UDP 161), its access restricted to the poller.
Read-only access, SNMPv2c or SNMPv3, set on the host or one of its folders (protocol:snmp:* variables).

Outputs

Name	Label	Unit	Per instance
`temperature`	Temperature	°C	yes, per sensor
`fan-speed`	Fan speed	rpm	yes, per sensor
`voltage`	Voltage	V	yes, per sensor
`current`	Current	A	yes, per sensor
`power`	Power	W	yes, per sensor
`status`	State	—	yes, per sensor

Watches

Name	Title	Kind	Defaults	Checked by default
`hot`	High temperature	above a threshold, on `temperature`	warning above 70 °C, critical above 85 °C, for 3 runs	yes
`failed`	Power supply failed	state to avoid, on `status`	critical on `fail`	yes
`fan-stopped`	Fan stopped	below a threshold, on `fan-speed`	critical below 1 rpm, for 2 runs	no: many MikroTik devices stop their fans while they are cool
`voltage-low`	Low voltage	below a threshold, on `voltage`	no threshold: the input voltage depends on the power source, set it on the probe	no
`missing`	Sensor missing	expected instances, on `status`	no sensor named: it raises nothing until names are set on the probe	no

Limits

The temperature thresholds, 70 °C and 85 °C, are generic defaults, taken from no maker's documentation: adapt them to the model on the probe.
The meaning of a state gauge, 0 for ok and 1 for fail, is not written in MIKROTIK-MIB; it is the one RouterOS shows and the common monitoring tools read. It is not verified here without a device.
The names of the sensors of mtxrGaugeTable are those of the device: they differ from one model to the next. A device upgraded from the fixed objects to the table renames its sensors, and the former ones go missing.
The version of RouterOS 6 from which mtxrGaugeTable is served is not verified here: the collector reads the fixed objects whenever the table is empty.