On 2011-12-29, percy piper <piper.pe...@gmail.com> wrote: > On 28 December 2011 21:36, percy piper <piper.pe...@gmail.com> wrote: > >> pppoe(4) did not work. We need to authenticate with chap first, then >> pap. pppoe(4) forces the exclusive use of either one or the other. > > Actually this is not true. Both use chap. There is no pap. > > The first chap challenge succeeds (to the BT AC) the second chap > challenge does not. No authentication failure response is received, we > just get another chap challenge and this repeats until a timeout > occurs. > > If we deliberately use an incorrect secret the first chap challenge > fails and we do get an auth fail response from BT's AC, so it seems > that BT's AC accepts our chap auth response but our ISP's AC does not. > Very odd especially as userland ppp does not exhibit this odd > behaviour at all. > > It may be that our ISP's AC is dropping the pppoe(4) auth response > packet as invalid (why?) or it never receives it which would explain > the lack of a response. I will need to comb through the packet > captures from both userland and kernel pppoe to see if I can find any > relevant difference. > > Has anyone else seen anything like this? This is with both the Dec > 20th i386 snap and 4.9 i386 release.
I haven't seen this with pppoe(4) and any of: zen fttc, demon adsl (ipstream), aaisp adsl (ipstream or 21cn), bogons adsl (ipstream). Does your ISP have reachable technical people that might be able to give a bit of insight into what they're seeing?
