On Mon, Apr 29, 2002 at 04:56:05PM +0300, Dani Arbel wrote:

> Hi!
> To support GRE tunnel, you would need rules both in INPUT and OUTPUT
> chains, since conntrack do not track them.

Right. I knew that the conntrack wouldn't track them because
there is no ipt_conntrack_gre module which contains the GRE
intelligence. But, thanks. Now I know that INPUT/OUTPUT is the
starting/termination points.

Ramin

> Dani
> 
> On Mon, 29 Apr 2002, Ramin Alidousti wrote:
> 
> > On Sun, Apr 28, 2002 at 09:04:00PM -0400, Mark Orenstein wrote:
> >
> > > Quoting Ramin Alidousti <[EMAIL PROTECTED]>:
> > > > Yes. Once when it comes through the physical interface. There, netfilter
> > > > would see it as an IP packet with protocol 47 (GRE). And once when the
> > > > packets come out of the GRE tunnel. Here again, netfilter would see IP
> > > > packets but the protocol part would be TCP/UDP/ICMP...
> > > >
> > > > Try these rules to see the association:
> > > >
> > > > $IPT -A FORWARD -i <physical-interface> -p 47 -j LOG
> > > > $IPT -A FORWARD -i <gre-interface> -j LOG
> > > >
> > > > Ramin
> > > >
> > > Thanks very much Ramin, one more question though.  Would the first rule above
> > > actually be in the INPUT chain?  I'll be in school tomorrow morning, so I will
> > > be able to experiment to get a better understanding.
> >
> > A very good point Mark. I don't know but having thought about it, what
> > you said sounds absolutely right. The first rule might not work as those
> > packets are not meant to get routed. Please do test both and let me know
> > the outcome. Thank you.
> >
> > Ramin
> >
> > >
> > > Thanks again,
> > > Mark Orenstein
> > > East Granby, CT School System
> >

Reply via email to