[c-nsp] Randomly Dead Sessions with MLPPP(oE)

[Gary T. Giesen]

I've got 2x 7206 VXR/NPE-G1 running 12.2(33) SRE6 acting as LNSs being
passed PPPoE sessions from an incumbent provider via L2TP. A couple
customers (including my personal Internet connection) using Multilink
PPPoE (over ADSL) are randomly experiencing dead sessions several
times per day. All 3 sessions will stay up (2x link + bundle session)
but will not pass any traffic. I've been unable to isolate the cause,
as all the DSL links work fine on their own. I'm running SGBP between
the two LNSes (with 1 as master), with Cisco routers as CPE (one is a
2821 running 12.4(25)g and the other a 3745 running 12.4(15)T14) both
experiencing the same problem. It looks like we have to clear the
sessions to restore connectivity (although I'm not 100% certain about
that, it may come back on its own). We've tried several different
models/brands of DSL modem, with no change in behaviour. Anyone know
what the cause/solution for this is, or at least know how to
troubleshoot it?

It's a little hard to do packet captures to isolate as my LNSes are
production, I have no SPAN capability upstream of them, and can't
replicate this problem in the lab.

Cheers,

GTG