A picture says a thousand words.



These two routers are new, in an air-conditioned data-centre, with minimal CPU load (3%).

The temperatures come from SNMP, as reported by the MIB iso.3.6.1.4.1.14988.1.1.3.10.0

Does anybody else see this behaviour? Is it incorrect temperature data, or is it really getting that hot?!

Off hand, I would say that is incorrect data. I could not determine if that router has a fan, but a fan that is failing to start once in a while could cause it to get hotter than normal, but unlikely that hot, that fast. If it does not have a fan (and presumed not to need one), I can't see if getting that hot and then cooling off that fast.

RB3011UiAS-RM does not have a built-in fan.

I've seen similar behavior on my 3011 unit as well. I believe it's just erroneous data returned by routeros.
It spikes for a second or two and then returns to normal values.

On RB850Gx2 is even worse, when temperature goes over 60 degrees it stops reporting the temperature altogether until it goes bellow 60 degrees again.

I could not determine if that router has a fan

No fan, completely silent operation. Just very strange spikes! :)

I've seen similar behavior on my 3011 unit as well. I believe it's just erroneous data returned by routeros.
It spikes for a second or two and then returns to normal values.

Ok, at least it doesn't sound like I have a faulty unit - though the three units I do have all spike by different amounts, but it's probably ok.

On RB850Gx2 is even worse, when temperature goes over 60 degrees it stops reporting the temperature altogether until it goes bellow 60 degrees again.

Is the RB850Gx2 actually getting that warm, though? Because I'm fairly sure nothing's getting hot in the RB3011 - if only because I have not encountered the "hot electronics smell"!

Spikes look like a bug and also "not reporting temp after 60 degrees" is a bug. Did you report the problem to support?

No I haven't reported it. I just put a fan keeping the temperature bellow 60 degrees

RB850Gx2 gets quite warm without a fan. RB3011 not that much, it doesn't need a fan so far.

We have three 3011s and we monitor all of them via Zabbix (SNMP). We get these crazy spikes too. Changed monitor times to anywhere from 15 seconds to 2 minutes and regardless of the monitoring interval, the problem shows up. They happen on all three and they are all in air-conditioned environments. They all have similar behaviors regardless of application or load. In view of what was said in this forum (above) I'm now more than ever starting to believe that this is indeed a bug but I'd love for someone at Mikrotik to confirm this.

Hello, this is probably not exactly related to your problem, but I'm adding it here just in case it might be useful to you, or if someone searches for topics on overheating units at some point in the future.

About 6 months ago I had a 2011 reach 80+ degrees celcius in a forced air ventilated rack (no air conditioning). This was witnessed in winbox, under system/health. The other 2011 in the same rack was at around 40 degrees, and a couple of 750UPs that were in the same rack were at aroung 35-40 degrees. CPU load on the overheating unit was not very high, if I recall correctly it was between 15% and 30%. In summary, the one 2011 was the only unit with a high temp, and it was the first time that I had seen such a high temp. I placed my hand on it just to make sure it wasn't an erroneous temp reading, since I was in the building, and I can attest to the fact that it was REALLY hot. Another symptom was that port 6 stopped working (no link, no light, no nothing)... It was very strange.

At any rate, panic set in, so I did not document my exact steps, but can clearly remember that:

1) I shut down the unit, let it cool down,
2) changed the power supply from an 18V generic to a 24 V original,
3) Pulled an export
4) upgrated Router OS and firmware to the most current.
5) Reset the unit and ran the export, and added a few things manually that the export doesn't script out.

The unit has not overheated again, and port 6 came back to life immediately. Since then I check the temp every 30 minutes and log it if it is >=50 degrees, and I haven't had a single event logged. If I had to guess, I'd say the generic power supply caused it, in conjunction with some other noise in the line, a rare singlular event. Then again, for all I know, it was happenning on and off again for a long time, and I only stumbled upon it that time by chance, which if true, would be similar to what you are witnessing. Nevertheless, the 18v generic switching power supply that I pulled from the 2011 has been used on a non-essential 750 ever since, with no problems whatsoever. Extremely strange, but true.

At any rate, I hope that you or someone searching these forums one day finds this info usefull. Kind regards.

I frequently (especially when booting) battle with 2011 devices when they become warm. Generally starts when they go 60C and above.

Definitely have overheating heat issues.

I have found a bug that causes excessive cpu load when nonexisting oid is requested by snmp. Reported to mikrotik but it looks they do not admit to be a bug for repair, advising not to call such oids. Will take some time to convince them that ros must handle whatever request. Check if it is not your case too.