Strange temperature readings from SmartUPS 1400xl

[Anonymous user]

I'm getting what I would consider to be odd temperature readings from one of a pair of 1400xl units powering a couple of servers.

Basic setup:

2 SmartUPS 1400XL each with 2 external battery packs (SU24R2XLBP) w/ APC 9606 SNMP cards being monitored via Network UPS tools.

Each UPS powers one of the redundant power supplies from the servers causing about a 30% load on each.

I've recently started graphing the outputs of several stats (load, voltage in, runtime, and Temperature) and noticed some odd "ramping" in the temperature between 80* and 100* on one of the units while the other one remains fairly constant at 100* degrees.  The load on the UPS remains fairly constant throughout the day (also graphed), I do not have the battery voltages monitored because those values are not reported to NUT though I can see them via telnet.

Is there any reason for this? And should I be concerned?  See the graphs below for more info.

The batteries in each unit are about 3.5 years old, the last long power outage we suffered was >3 hours.  As far as I can tell from the logs both UPS ran for the estimated times then shutdown.

[BillP]

Can you clarify - are you talking about the internal temperature of the UPS or the ambient temperature of the location? I ask because I thought it was the UPS internal temperature then one of your graphs refers to ambient. AP9606 does not have the capability itself to measure ambient temperature so that is why I am asking and a little confused.

[Anonymous user]

The ambient reading is coming from another device in the same closet, I was also mistaken, the cards are 9605 SNMP/Telnet (no web).  I was including the Ambient temp for reference.  The other two graphs are from each UPS's internal temperature. 

[BillP]

Ok, I understand better now. Can't say I've seen a specific issue like this before regarding the reading for the internal temperature. Since the batteries are older, I'd definitely just give them a visual check for piece of mind and also if possible, see if they are warm to the touch. Also, I know this is an AP9605 so it does not have an event log but I was wondering if the temperature fluctuations lined up with any other types of events i.e. on battery, self test, etc.

Also, when it jumps around, do you think it could be anything like server exhaust or something blowing on this particular UPS to heat everything up?

[Anonymous user]

The 2 UPS units are stacked on top of each other at the bottom of the rack along with the external batteries, servers are on top so the heat from their exhaust should not be a factor.  I was planning on some downtime next weekend anyway, I'll plan on taking a look at the batteries while I'm at it.  These are all slated to be replaced next summer anyway, I don't want my first indication that a battery is failing to be when the UPS cuts out during a real power outage. 

At this point I'm guessing thinking that either one of the batteries is on its way out or the temperature sensor is going out.  I'm assuming the sensor is a simple thermoresistor that changes its resistance as the temperature changes, they can be susceptible to noise in the system and give off misleading info.

I've setup nagios to warn me when the temperature gets above 110 degrees, is this a fair warning threshold?  The 1400 documents state it can work in ambient temperatures up to 104 degrees but I can't find anything official about the internal temperature. 

[BillP · Accepted solution]

It sounds like these units may be 10+ years old based on the fact you have an AP9605 so we'd guess that it is the temperature sensor at fault in the one unit since it is fluctuating and not constantly reading high (versus the batteries being the root problem). 110 should be a good threshold. Beyond that I'd be concerned about thermal runaway (assuming the temp sensors are accurate).