What is the APC temperature SNMP OID?

1.3.6.1.4.1.318.1.1.10.2.3.2.1.4 is iemStatusProbeCurrentTemp from PowerNet-MIB, the temperature of an environmental probe attached to a Network Management Card. There is no scaling applied, but the unit is whatever the card is configured for, so you must also read iemStatusProbeTempUnits at 1.3.6.1.4.1.318.1.1.10.2.3.2.1.5, where 1 means Celsius and 2 means Fahrenheit.

Why does my APC temperature alert never fire?

Almost always because the card is reporting Fahrenheit. An NMC set to Fahrenheit returns 78 for a perfectly normal 25.5 C room, so a threshold of 32 C is never crossed — the number is always far above it and the comparison never triggers a warning in the direction you expected. Read iemStatusProbeTempUnits alongside the value, or set the card to Celsius and document it.

Can I read humidity from an APC probe over SNMP?

Yes, if the probe supports it. iemStatusProbeCurrentHumid at 1.3.6.1.4.1.318.1.1.10.2.3.2.1.6 returns relative humidity as a percentage. Humidity matters in a server room for two opposite reasons: too low and you get static discharge risk, too high and you get condensation on cold surfaces.

My APC rack PDU returns nothing on those OIDs. What now?

Rack PDUs on newer firmware expose their sensors under the rPDU2Sensor tree instead of the integrated environmental monitor tree. Walk 1.3.6.1.4.1.318.1.1.26 and look for the temperature and humidity entries there. The integrated environmental monitor OIDs apply to Network Management Cards with a probe attached, such as an AP9335T.

Which APC hardware exposes environmental sensors?

Network Management Cards in the AP9630, AP9631, AP9640 and AP9641 families with an environmental probe such as the AP9335T attached, and rack PDUs with sensor ports. The card is what speaks SNMP; the probe is what measures. Without a probe attached, the temperature OIDs return nothing.

Vendor guide

APC temperature and humidity over SNMP

An APC Network Management Card with a probe attached is the most common way a small server room gets a real room temperature, rather than a server's opinion of its own inlet. The OIDs are simple. The unit handling is where people lose a night's sleep.

The trap: the card reports whatever unit its web interface is set to.

An NMC configured for Fahrenheit returns 78 for a perfectly ordinary 25.5 °C room. If your alert says “warn above 32”, that reading is already 78 — so either it fires permanently and you mute it, or your comparison is written the other way round and it never fires at all.

This is the worst class of monitoring bug, because nothing looks broken. The graph has numbers on it. The card is online. The alert is configured. And the room can climb to 40 °C without a single notification.

What you need

Two separate things, which is a common source of confusion:

A Network Management Card — the AP9630, AP9631, AP9640 or AP9641 families. This is what speaks SNMP.
An environmental probe such as the AP9335T, plugged into the card. This is what actually measures.

Without a probe attached, the temperature OIDs below return nothing at all. If you are getting empty walks on a card you know is online, check the physical probe before you go any further.

The OIDs

From PowerNet-MIB, the integrated environmental monitor table:

Object	OID	What it gives you
`iemStatusProbeCurrentTemp`	`1.3.6.1.4.1.318.1.1.10.2.3.2.1.4`	Probe temperature. No scaling — but see the unit below.
`iemStatusProbeTempUnits`	`1.3.6.1.4.1.318.1.1.10.2.3.2.1.5`	1 = celsius, 2 = fahrenheit. Poll this.
`iemStatusProbeCurrentHumid`	`1.3.6.1.4.1.318.1.1.10.2.3.2.1.6`	Relative humidity, percent.
`iemStatusProbeName`	`1.3.6.1.4.1.318.1.1.10.2.3.2.1.2`	Probe name as configured on the card.

# temperature and its unit, together — never one without the other
snmpget -v2c -c public 10.0.0.30 \
  1.3.6.1.4.1.318.1.1.10.2.3.2.1.4.1 \
  1.3.6.1.4.1.318.1.1.10.2.3.2.1.5.1

# INTEGER: 25      <- the reading
# INTEGER: 1       <- 1 = celsius, so 25 °C. If this were 2, it would be 25 °F.

Two ways to handle it, and both are defensible:

Poll the unit alongside the value and convert in your monitoring. Correct whatever anyone does to the card later.
Set the card to Celsius and write it down somewhere a future colleague will find. Simpler, until someone changes it during an unrelated bit of maintenance.

Rack PDUs answer somewhere else

If you are polling an APC rack PDU with sensor ports rather than an NMC with a probe, the OIDs above may return nothing. Newer PDU firmware exposes sensors under the rPDU2Sensor tree instead:

snmpwalk -v2c -c public 10.0.0.31 1.3.6.1.4.1.318.1.1.26

Walk it and read the descriptions to find the temperature and humidity entries for your model. The tree is large — it covers outlets, phases and banks as well as sensors — so start from the names rather than guessing indexes.

Humidity is worth watching too

Temperature gets the attention, but humidity fails in both directions. Too low and you get static discharge risk when someone works in the rack; too high and you get condensation on anything running cold. A probe that already reports humidity costs you nothing extra to graph, and it is the kind of number that explains an otherwise inexplicable hardware failure six months later.

What to alert on

A room probe is the closest thing you have to “how hot is it actually in here”, so it deserves your primary threshold. As a working starting point:

Probe temperature	Status	Action
up to 27 °C	Nominal	None.
27–32 °C	Warning	Investigate airflow and cooling.
32 °C and above	Critical	Cooling is likely failing. Act now.

Add a hold window of a few minutes so one odd sample does not page anyone, and — given everything above — sanity-check the first reading against the card's own web interface before you trust the threshold. That one comparison is what catches the Fahrenheit problem on day one instead of during an outage.

One probe is not a rack

A single room probe tells you the room. It does not tell you that the top of rack B is running eight degrees hotter than the bottom because a fan died three weeks ago. If you have servers and switches that already report their own inlet over SNMP, poll those too and put them together.

Rack thermal map combining room probes and per-device inlet readings, each device drawn in its U position — Room probes and per-device inlets on the same picture — which is how you notice the hot U, not just the hot room.

RackMon ships the APC profile above, unit handling included, alongside the server and switch profiles. Self-hosted in one container, free for unlimited devices.

See it with live data Install it

Other hardware

Cisco, Juniper and the vendor-neutral standard are all in the SNMP temperature OID reference. For servers there are step-by-step guides for Dell iDRAC and HPE iLO, and if you are still deciding where to put sensors at all, start with the rack temperature monitoring guide.

← All vendor OIDs Monitoring guide