RB922UAGS-5HPacD, ether1 broken?

Hi,

I decided to upgrade a existing office accesspoint with a RB912 board to AC performance, so I got a RB922-board.

After initial configuration I measured absolutley horrible performance, about 1-2Mbps with wireless. I tried all kind of settings and even at basic 11a mode it was terrible. Then I saw that the Ethernet-port had several linkdowns. I took a look at the switch (RB260GS) and there was thousands of FCS Errors, Pause and packet drops.

I created two VLAN’s and tried some simple routing, and the performance is even worse(!) iperf sometime reports 0kbps as the link just dies.

So far i have tried:

  • Different powersupplies
  • Many different cables, cat5e and cat6
  • Factory reset
  • Two different switches
  • With and without traffic-flow
  • 100Mbps/FDX - and then I can reach about ~80Mbps VLAN-routing, still with errors/drops

I’m not using PoE.

The RB912 had no problems with VLAN routing, peaking at 600-700Mbps.

Is there anything else I could try or is it a case of RMA?

Try to change interface queue type for ethernet.

Tested ‘ethernet-default’ and ‘default’ - Same problem.

I did some tests against the RB912 with the bandwidth test. Inbound performance seems fine. Averaging on 500-600Mbps. But outbound performance test generates interface resets.

Graph from RB912 board:

Interface resets on RB922:

How did you do the factory reset? From the quick guide (http://i.mt.lv/routerboard/files/rb9xx-ac-qg.pdf):

RouterOS reset jumper hole (Reset2 in above image) – resets RouterOS software to defaults. Must short circuit the
metallic sides of the hole (with a screwdriver, for example) and boot the device. Hold screwdriver in place until
RouterOS configuration is cleared.

But there is no “above image”, and I cannot find the jumper hole when I study the PCB.

I never used the jumper/pads, did it in CLI (/system reset-configuration) But for what I understand, you simply should short the connection between the pads som something conductive.

Anyways.. My problem got solved. The antenna was to close to the Ethernet-chip, causing it to reset. When a moved the antennas, the problem disappeared. I guess it’s the same symptom as GSM-noise in FM-radios/ampilifiers :slight_smile: