UPS warnings "Power failure"-"Power is back"

About once a week I get a warnings like this on the “APC-UPS status” page:

2023-04-23 15:26:29 +0200  Power is back. UPS running on mains.
2023-04-23 15:26:26 +0200  Power failure.

In that moment it can be heard that the UPS powers up for two or three seconds and then gets in to standby mode again. When not using an UPS but plugging directly into the wall, the complete appartment loses power in this moment - the RCCB triggers, in case of this type of event appears.

This appeared years ago in my old appartment, too, and the solution was then to buy the UPS. Meanwhile, a lot of the hardware, including the ipFire device, is exchanged but these regular occurences still appear.

I’ve looked into the logs of my server and the ipFire but can’t find events which seem to be related to these situations on a software side.

I wonder why these regular events in the logs do appear causing this situation.

Any hints?

When running the equipment without the UPS the RCCB triggering indicates a leakage current flowing to Earth or another circuit above the trigger level of 30mA. This will be due to a physical fault of some type in the equipment connected to that socket.

When the UPS is connected between your equipment and that socket then when that leakage flow occurs, instead of tripping the RCCB it will cause the Mains voltage coming out of the UPS to be depressed and this can feed back to the Mains Voltage monitor in the UPS which then effectively sees a mains voltage loss and turns onto battery. The Power failure message is then sent by the UPS to APCUPSD.

A leakage current from the Live Mains side of your computer equipment and Earth is not going to be caused by the IPFire software.

Their is a physical fault that is not permanently present, that periodically is causing a >30mA current to flow on the mains side of one of your equipment power supplies.


for the items plugged into the UPS (like the ipfire device), it there something that uses the earth clips. Maybe a printer?

What all is plugged into the UPS?

That’s a good point. Most computer related stuff is double insulated and doesn’t even have an earth.

However the rccd measures the current flow in the live and the neutral lines and if the difference between those two is greater than 30 mA then it will trip as that means current us flowing somewhere it shouldn’t. Usually that is earth on tge socket but it could also be to literal earth or to another live or neutral line connection.

The UPS has attached

  • the ipFire (new)
  • a server (old, but new PSU)
  • Ubiquiti 24 port switch (new)
  • Ubiquiti 8 port PoE switch (old) with 2 PoE APs hanging on it

The term “new” means exchanged without solving the troubles with those power interruption messages.

I’ll pick on this one for a moment. Please send a picture of the plug. Is it a two pin/wire power plug? Or a three pin/wire power plug?

How old is the server box?

I ask because many years ago you could not plug a three wire computer into a RCCB (GFIC for me). It would trip the fault circuit. But not instantly. So all computer three wire devices had to be plugged into a non-RCCB (GFIC) outlet.

The server box is about 8 years old. The plugs are “Type F” on one side:

and C13/C14 on the other side:

I wonder if perhaps not the server box but the PoE switch could cause these regular incidents? I could remove that switch for a while from power and look if these incidents would disappear.

1 Like

Hi @firewire,

Do you have a multimeter?
If yes set it on OHM and measure the ground to the 0 and the ground to the L, on your first picture is the middle thing and first the left ore the right and then visa-versa, on the second picture is the middle longer out and then to the left pole and then to the right pole,
if one of those device’s with a ground give a reading that’s your bad guy.

If your multimeter have a high scale and read something like 1Mohm the device is ok but everting around 1Kohm will trip your earth protection circuit RCCB. i think even 47Kohm will still trip it.

I hope you figure it out witch is giving you the problem, I use to be a electrician and that is one of the way to find that kind of problem but can still be tricky as some device may only give you a bad reading after certian condition.

Best regards


1 Like

Is the PoE switch a three pin/wire device? If not, then it is unlikely the issue.

Make sure the device is unplugged and wait a minute or two before measuring.

There is a good chance the server power supply has small noise capacitors between earth (ground) and the live (hot). So you may not see the “leak” with a multimeter. The only way to know for sure is the find the schematics for the power supply.

1 Like

@jon hit it on the head!

I really hope you find the problem, as i also said, and @jon pin point it to it can be difficult with just a multimeter but it’s worth the try.

Best regards

Thank you all for your hints and tips. I will, step by step, further analyze these incidents and provide report in case of managing to find the cause.