RB600 Bridge bug?

I hope someone might be able to shed some light on a problem I’m having with bridging 2 radios and one ethernet interface on an RB600A.

RB600A

  • ROS v3.30
  • Bootloader 2.20
  • 2 Ubiquiti SR4 radios
  • both configured in ap-bridge mode
  • both using wpa2+aes
  • 1 SR4 (wlan1) uses channel 4950 (antenna horizontal polarization, 1.1 mile link)
  • 1 SR4 (wlan2) uses channel 4970 (antenna vertical polarization, 4 mile link)
  • remainder of settings are default
  • Created a bridge, bridge1, and added ether2 (poe port), wlan1 and wlan2 to this bridge.
  • tried bridge with NO STP, STP and RSTP…same results

Client radios are both Teletronics TT4900 units.


I’m not sure if simply creating a bridge like this is the best way, but that’s how I have it. The specific problem I’m having is that layer 2/3 traffic to one of the Teletronics radios stops after a few minutes…very consistently.

(SR4:wlan1)|—(4950-Horiz)—TT1 (172.16.1.:sunglasses:
|
poe—ether2----RB600–|
|
(SR4:wlan2)|—(4970-Vert)----TT2 (172.16.1.5)

ether2, wlan1 and wlan2 are ports in bridge1. bridge1 is assigned IP address 172.16.1.10.


And here is the problem I’m seeing. Once both Teletronics units are connected, I start a MAC Ping to 172.16.1.8 and after several minutes (somewhat variable but roughly 3 minutes), the MAC Ping starts a continuous timeout. Strangely though, I can still PING 172.16.1.8. This certainly looks like a bug to me, since I should NOT be able to PING (layer 3) if I cannot MAC Ping (layer 2). Eventually, the regular ping fails as well.

However, pings FROM TT1 to the RB600 and through it work just fine. It’s as if the RB600 simply stops layer 2 communication with TT1. Just to note, mac pings and regular pings to TT2 remain fine.

To correct the problem, I have to either reboot the RB600 or TT1 or at least disable and enable wlan1.

And to complicate matters further, somehow the problem has stayed with that link. I tried swapping antenna cables and reconfiguring radios in the RB600 so that opposite radios connected to each of the Teletronics radios. Either way, traffic to TT1 drops. I’ve tried swapping all hardware at the TT1 site with same results (tried different tt4900, poe, bypassing switch).

Both wireless links have good signal strength (~ -58 dBm), but both do have low CCQ values around 60%.

I’ve enabled logging for both wireless and debug, but when the problem happens, nothing is logged.

I’ve tried putting wlan1 in a separate bridge (bridge1) with ether2; wlan2 and ether3 in bridge2; both bridges using RSTP, and ether2 and ether3 connected to switch…same problem.

RB600 resources look normal: plenty of free memory and disk space; cpu usage is minimal.


Anyway, I’m hoping someone can shed some light on this problem. If my configuration is wrong, I’d be glad to hear from someone who can suggest how to do it differently. However, I’m highly suspicious of a bug.

BTW, I’m not using WDS for bridging since the Teletronics radios don’t support the necessary options i.e. station-wds or wds-slave, although I haven’t given this a try at all.

Regards,

Mike

Anybody?

Could there be some ‘bad packets’ coming in over the bad link causing something in the bridging code to die.?

I’m completely at a loss as to why this might be happening and only happening one link.

Regards,

Mike