I think I understand what''s going on, thanks to a small mistake ;-) I needed to add a 5th nic to the gateway box - this new nic was an identical mate for another isa in there, so I modified the module options accordingly. So, after 10 minutes of trying to figure out why the DSL device was unwilling to talk to, well, anything - it occurred to me that the new card was the next address up, hence the modem is *now* on eth4... I modified the wshaper and re-ran it. Next few hours are spent wondering why the path between the half-dead server and the replacement is at, like, DSL speeds :-/ Today, I''ve dug deeper, and found that eth3 was *still* being shaped (even though it''s now part of a 3-if bridge), and indeed, some bands had seen traffic!! However, eth4 was also now shaped, with the same script (makes sense after reading wshaper again), but the bands are *not* seeing traffic, even after limiting to half the DSL line speed and upping a 6mb file to a remote site via scp... I''m now almost totally convinced that traffic is eluding the filters because it''s *actually* traffic out ppp0 (I''m currently using the Debian userland pppoe thingy), and I''m tc''ing on eth4 (to which the DSL device is attached). However earlier in the game, when I tried to shape ppp0, the kernel would reliably oops after about a minute. What am I doing wrong? -- Cheers, Mattt. icq : 117539757 aboveNetworks www : www.above.nq4u.net mattt@above.nq4u.net jabber: mattt@jabber.above.nq4u.net What''s got four legs and an arm? A happy Pit Bull... _______________________________________________ LARTC mailing list / LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/