On Mon, Jul 28, 2003 at 01:04:01AM -0400, Greg Folkert wrote: > On Mon, 2003-07-28 at 00:43, Ron Johnson wrote: > <---SNIP---> > > # cat /proc/sys/net/ipv4/tcp_ecn > > 1 > > > > When /proc/sys/net/ipv4/tcp_ecn had the value "1", I couldn't get > > to thatpetplace either. However, I could, after I did this, and > > then restarted Mozilla: > > # echo "0" > /proc/sys/net/ipv4/tcp_ecn > > # cat /proc/sys/net/ipv4/tcp_ecn > > 0 > > > > Make sure to reenable tcp_ecn when you're finished! > > > > # echo "1" > /proc/sys/net/ipv4/tcp_ecn > > # cat /proc/sys/net/ipv4/tcp_ecn > > 1 > > Ron, as of this writing, 12:55AM EDT, I will have to disagree with you > about turning tcp_ecn back on. For about the next 2 years at least. [ snip 'windows doesn't do ECN' ] > Very little luck with website admins whom have "drunk the Microsoft > Kool-Aid" (I know drank is right but drunk get's the point across > better) stating they are using "Industry Standards" and so on... Er, RFCs are the standards. > Well, overall ECN is a great way to make the Internet "self-regulate" > and of course the biggest obstacle is M$ products. But for quite a while > yet, defaulting it to OFF is a good thing. I disagree. A better idea is to leave ECN on, and use iptables to mangle packets to sites that reject packets with ECN set. AFAIK there's support to do just this with built-in targets since 2.4.20. -- Nathan Norman - Incanus Networking mailto:firstname.lastname@example.org Warning: dates in calendar are closer than they appear.
Description: PGP signature