MT's PPPoE performance has always been somewhat sketchy IMHO.

All wireless our clients run OpenWRT and Madwifi, or other RouterOS boxes, all Atheros radios on the network.

I've always had problems with certain clients failing in patterns, seemingly based on what gear they're using (ie - what build of OpenWRT I flashed), and my RouterOS boxes seem the least reliable units on the network but that may be because I only use them in 'difficult' places.

My latest "fun with pppoe"...

AP side: RB532, SR9 in 5 mhz w/ nstreme 'best fit' on a medium yagi (11 dBi) for limited distribution, EM9-AB in 11a mode (5.8 ghz) for the "backhaul".

Client side: RB112, SR9 in 5 mhz mode w/ nstreme 'best fit' on a larger yagi (13 dBi), CM9 in 11g (ch 11) mode for the client's laptop to connect.

-81 signal, -100 to -102 noise reported. The radio link seems very consistant since I can always get it to ping 4 times after resetting the PPPoE.

Heres what happens when the PPPoE comes up today:

72 bytes from 12.197.129.62: icmp_seq=14 ttl=61 time=7.319 ms
72 bytes from 12.197.129.62: icmp_seq=15 ttl=61 time=20.684 ms
72 bytes from 12.197.129.62: icmp_seq=16 ttl=61 time=17.846 ms
72 bytes from 12.197.129.62: icmp_seq=17 ttl=61 time=7.785 ms
72 bytes from 12.197.129.62: icmp_seq=18 ttl=61 time=99024.233 ms
72 bytes from 12.197.129.62: icmp_seq=19 ttl=61 time=98015.339 ms
72 bytes from 12.197.129.62: icmp_seq=20 ttl=61 time=97006.149 ms
72 bytes from 12.197.129.62: icmp_seq=21 ttl=61 time=198666.012 ms
72 bytes from 12.197.129.62: icmp_seq=22 ttl=61 time=197657.471 ms
72 bytes from 12.197.129.62: icmp_seq=23 ttl=61 time=196648.271 ms
72 bytes from 12.197.129.62: icmp_seq=24 ttl=61 time=298999.268 ms
72 bytes from 12.197.129.62: icmp_seq=25 ttl=61 time=297989.907 ms
72 bytes from 12.197.129.62: icmp_seq=26 ttl=61 time=296980.548 ms
72 bytes from 12.197.129.62: icmp_seq=27 ttl=61 time=399002.923 ms
72 bytes from 12.197.129.62: icmp_seq=28 ttl=61 time=397993.596 ms
72 bytes from 12.197.129.62: icmp_seq=29 ttl=61 time=499991.537 ms
72 bytes from 12.197.129.62: icmp_seq=30 ttl=61 time=498982.668 ms
72 bytes from 12.197.129.62: icmp_seq=31 ttl=61 time=497973.285 ms
72 bytes from 12.197.129.62: icmp_seq=32 ttl=61 time=599973.723 ms
72 bytes from 12.197.129.62: icmp_seq=33 ttl=61 time=598964.049 ms
72 bytes from 12.197.129.62: icmp_seq=34 ttl=61 time=597979.407 ms
^C
--- 12.197.129.62 ping statistics ---
639 packets transmitted, 21 packets received, 96% packet loss
round-trip min/avg/max/stddev = 7.319/279614.382/599973.723/208328.790 ms

If I reset the PPPoE the exact same scenario occurs again.

It worked *awesome* all day yesterday. Now its doing this. The 900 radio's stats are decent. I'm going out to the customer side to see if theres any reason for this otherwise, but I seriously doubt it due to the consistency of the error.

2.9.27 on both ends of the link, I'll be upgrading these to 2.9.28 immediately.