I have two powerbox pro powering 3 ubiquiti rocket ac prism connected to each one.
Both of them turn off power to one port several times a day. In log appears OVERLOAD DETECTED.
I power both powerbox via ethernet and the 24v 2.5a original power switching ...

The rocket datasheet says that the max power consumption of the unit is 0.4a.
Powerbox pro can power max 1a per port and 2a max sum all ports.

So afaik there is no overload ....

What can I do to prevent ports from being shut down?

i replaced 2.5a original power switching with a 10a ...

"overload detection" problem still happens, but not several times a day... but one or twice a week

i replaced 2.5a original power switching with a 10a ...

"overload detection" problem still happens, but not several times a day... but one or twice a week

Maybe cable or connector problem. What if you try to put new RJ45 plugs on booth end of the cable, or try with other cable (with thicker full copper wires).
How long the cabel betweeen Ubi and Miki device?
Maybe the new PSU give higher voltage then factory one, and the radio needs less amper, that cause less frequent overload. Maybe, if the PSU has voltage adjustment, you can add a little more voltage, but can be dangerous, if the radio doesnt support higher voltage then 24V

Cable is 30 meters long
Cable is 100% copper
PSU has voltage adjustment and I adjust it to let the powerbox receive 24v exactly

I don't think it is RJ45 because it happens to more than one powerbox... with different cable...

I will try to power the powerbox with dc-jack

I found a hacky fix to this problem, but it works and can be done remotely.

The only catch is you have to have one spare PoE port.

Set the spare port to have PoE output forced on, and set the PoE priority lower than the other ports.

When overload is detected, it will disable PoE on the spare port and recheck the load before turning other ports off - the overload seems to be instantaneous so it never turns off the other ports (in my experience).

nice workaround. i will try it

I found a hacky fix to this problem, but it works and can be done remotely.

The only catch is you have to have one spare PoE port.

Set the spare port to have PoE output forced on, and set the PoE priority lower than the other ports.

When overload is detected, it will disable PoE on the spare port and recheck the load before turning other ports off - the overload seems to be instantaneous so it never turns off the other ports (in my experience).

BEAUTIFUL !!! - Thanks

I wish it could just be fixed with ether1-long-poe-in like in the old days : )

I wish it could just be fixed with ether1-long-poe-in like in the old days : )

This only work on PowerBox and not PowerBox Pro model...

I found a hacky fix to this problem, but it works and can be done remotely.

Interesting.

P.S.
All my PowerBox, sEX POE (same RB960PGS) have been unstable in power or power supply ports (all in the trash).
The only stable solution that has never let me down is Omnitik POE (old/new).
The price difference is not large. I better buy an Omni than a PowerBox Pro.

An interesting fact can also be observed after the data sheet:
OmniTIK 5 PoE ac - MTBF Approximately 200'000 hours at 25C
PowerBox Pro - MTBF Approximately 100'000 hours at 25C

I am using a 12V battery powered RB260GSB to provide power to a device which draws more than 1 amp for about 200 microseconds about 30 seconds into the devices's boot sequence. When I supply >13VDC to the RG260GSB, that 200 microsecond draw is not detected as a short by the RG260GSB, but if the supply voltage is less than 13VDC the switch detects a 'short' and cycles the POE power on that port. Changing the port priorities and 'calibration' don't fix the problem.

This behavior is unfortunate since no 12V batteries supply 13V and adding a 12V=> 24V boost convertor ahead of the RGB will introduce RFI and another source of failure in the system. Can anyone suggest a work around?

If you can live without having observability and control of the remote device, then you could try using either RBPOE or RBGPOE passive injectors instead of using RB960GSP PoE out. RBGPOE is rated at 2A, I'm not sure about short circuit protection ...

I don't need observability, but I absolutely need control of the POE.
This is for a radio receive site which has been off air for the last year because I couldn't remotely power cycle the receivers.
I can solve the power problem by supplying 24V to the RB, but that would require that I add a 12V->24V boost converter which would add radio interference and add an additional single point of failure.

And where do I find instructions for 'calibration' of the POE? After setting and applying 'calibr' the RB says it is 'waiting for load'. A Lightbeam plugged in doesn't get power, so what 'load' does it want?

but that would require that I add a 12V->24V boost converter

Is adding another 12V battery in series an option?

Also, the original question was about Powerbox Pro, but since in your case it is RB260GSP you have an option to disable/limit that overcurrent protection by enabling the "Port1 PoE In Long Cable" option on the "System" tab. Have you tried that?