Alex Bennee wrote: > > alex wrote: > > > >> On Mon, 2001-10-01 at 10:16, Hamish Marson wrote: > > > > I've just done a check with ethereal on both my laptop (NAT'ed), and > > directly from the firewall running the pppoa2/modem_run (i.e. No NAT). > > Looks like the sequence numbers are being screwed... Up until it > > stalls, the Next Seq - Seq for the incoming packets match fine. Then > > they go out of wack... I'm trying to find an RFC that describes the > > NextSeq/Seq flow (RFC 761 doesn't describe the NextSeq field in the tcp > > header from a quick look). I'm trying to determine whether that's the > > actual problem, or just an artifact... > > If the sequence numbers are wrong but the TCP checksums are ok I doubt how > pppoa could be breaking things. It could well be an artifact of the > NAT/ipchains code? What kernel are you running? >
I agree, and don't see how pppoa could screw the SEQ, but keep the TCP checksum OK... But it's either got to be pppoa OR something seriously screwed upstream, but limited to somewhere close to my connection on BTOpenWorld. If it's not pppoa I'd need to goto BTOpenWorld, and convince them to take a look at something upstream... And I can see how far I'd get with that... About as far as I got when my last ISP went under... (i.e. not bloody very) The Kernel is 2.4.9 (And affects 2.4.4 and 2.4.7 as well)... But like I said... This happens when NOT running NAT/iptables... ie.. direct from the server that has the USB speedtouch on it... Might be time to try a (yuk) windoze installation on it & see if that can download the files... That will eliminate anything upstream from myself... > > Alex. > > Liste de diffusion modem ALCATEL SpeedTouch USB > Pour se d�sinscrire : mailto:[EMAIL PROTECTED]?subject=unsubscribe > > -- I don't suffer from Insanity... I enjoy every minute of it... Liste de diffusion modem ALCATEL SpeedTouch USB Pour se d�sinscrire : mailto:[EMAIL PROTECTED]?subject=unsubscribe
