> > on your LAN machines, eth0 can be 1500 MTU. the catch is:
> > route del default ...
> > route add default gw <router> mss 576 ...
> >
> > add the "mss 576" tag. -that- is what fixed it.
>
> I have no control on my SLIP server. Since setting SLIP MTU
> to 1500 works for me, I don't want to change it back to 576.
I actually used the trick a few years ago when we had real terminal
servers making up a SLIP connection and they had a hardcoded maximum MTU
of 1006. Then setting route MSS to 966 (i.e. MTU minus standard IP header
minus standard TCP header) used to cure the problems with TCP connections.
Most problems were actually seen when connecting to Solaris boxes within
our university network which had a broken PMTU implementation (almost
nobody heard of firewalls at that time).
> If "mss 576" works, which I doubt, from "man route":
You would have to use "mss 536", actually.
> mss M Set the TCP Maximum Segment Size (MSS) for connec-
> tions over this route to M bytes. This is normally
> used only for fine optimisation of routing setups.
> The default is 536.
>
> where 536 < 576, shouldn't the kernel just use MTU on the
> gateway for mss? Does it look like a kernel bug?
The documentation of `route' is wrong and should be fixed.
-- + Maciej W. Rozycki, Technical University of Gdansk, Poland + +--------------------------------------------------------------+ + e-mail: macro@ds2.pg.gda.pl, PGP key available +
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/