Background
Max BR2 Pro HW 4 FW 8.5.4
Powered by a 24 PoE 2.5G Switch Rugged
Mounted on a pole with an Antenna MAX Duo (love that one)
WANs: TMO, VZW (both 5G), Starlink
LANs: A bunch
Clients: 45
SF connections: 2
Issue
When doing staggered (by 15 minutes) hourly connection tests on all three WANs, the router reboots frequently, seemingly at random.
When turning off connection tests the router is stable.
Booting to 8.5.3 build 6030 does not make a difference
Level of Concern
Low.
This is not a critical issue for us, once the correlation with the connection test frequency was established.
But it was concerning, as we worried about more serious causes being possible, and with the router being in the mountains and up on a tall pole, access for further hardware investigation would be irksom.
My first thought was power delivery, not that hitting a single connection hard should draw enough to be an issue but combined with the connection test probably hitting the CPU a bit harder, it could be just enough extra above anything you do with normal traffic conditions to sometimes be a problem.
A marginal crimp on a rj45 could also cause some issues.
The Br2 Pro can pull 30w plus the USB power for the 3rd modem, POE+ ports put out 30W but actually only guarantee 25.5w at the device by spec (loss in connectors, loss in the cable, even the outdoor temp can effect this). A higher gauge network cable may help.
There is a passive POE kit for injecting 36w, that would get you close to 30w at the device. A POE++ adapter might also help but I haven’t tested that, it might not since the device might have to negotiate a higher power budget. It’s been too long since I read the engineering specs of 802.3 to remember.
There is a warning on the device compare page
“1x 802.3at PoE port provide a power supply of 25W. If you require 30W of power budget (eg. using dual cellular WAN and USB WAN together), please empower the MAX BR2 Pro via the bundled power adapter (12V3A PSU, ACW-663), or the optional Passive PoE Injector Kit (ACW-111 + ACW-622).”
It’s the kind of issue that probably would never show up without that many WANs and something hitting the CPU enough to just push over the limit.
EDIT:
I got to thinking about this but don’t have one on hand, I think the ACW-622 kit is 56v at 2.1A, so ridiculously above what that device could ever pull. Can someone with one around chime in? Also if you go that route, both the 622 and 111 are needed.
Though it seems a reasonable thing to look at, the set-up should not cause this. Checking the switch the power draw does not exceed 20W, and there is no USB involved (just two cellular, one ethernet WAN and one LAN Ethernet port).
I am dropping the frequency of tests, to see if that does anything to anything
