Hi!

I've got pf 1.59 w/ ALTQ running and now get into the situation that outgoing packets are queued correctly but I don't have the chance to configure the opposite side (DSL access concentrator, Cisco router). I'm just hacking up pfctl and pf to recognize a rule like this:

scrub out on $ext_if proto icmp from $internal to any tos 0x10 keep state or scrub out on $ext_if proto udp from $internal to any port 4000 tos 0x10 keep
state

This will set the ip_tos bits to 0x10 (if not set otherwise). I still have to conduct some tests (prolly using nemesis-icmp) to see whether my upstream routers will return the answer with the same tos bits.

In case a) the destination host and b) the upstream router both make their queuing decisions based on the tos bits we might manage to influence our upstream routers behaviour and finally play Diablo while doing large <insert random traffic hogs> downloads.

Since weather is nice and I have other things to hack on this might take a few days until it proofs stable.

Just FYI, Hendrik

drag me, drop me - treat me like an object