Well, it’s hard to believe this thing, but it’s real.
We gave to a customer a RB2011 (RB2011UiAS-2HnD-IN) as a router.
Internet connection is through via a WiFi link (a SM PMP450) or, as backup, ADSL. On the ADSL there is a bare DLINK modem/router with pppoe-passthrough enabled.
Ethernet port for WAN are ether6(main, wifi) and ether7(adsl). Ether2 is for LAN.
From about a week, pppoe client doesn’t start on the main link, the wifi.
Today, after many tries, the only working workaround was to change ethernet port, moving main link from ether6 to ether4 and setting pppoe-client interface on ether4.
One could think “well, fried ethernet port”, but in a test lab, the RB2011 was connected to two RB1200, both of them linked to a CCR, as a test PPPoE server.
RB1200s had ether1 and ether2 bridged together, with ether1s linked to different port of the CCR.
I did this setup to test my failover scripts and test with semi-random packet loss
Well, the RB2011 worked without any issue.
So, if it is a physical problem, it is really strange: it arises with some equipment and not with others.
Auto-negotiation is always enabled.
Any idea?