On Mon, Dec 15, 2014 at 3:21 PM, Hunt Xu <mhun...@gmail.com> wrote:
>
> On Mon, Dec 15, 2014 at 2:48 PM, 邓尧 <tors...@gmail.com> wrote:
> > Hi,
> >
> > I need to connect two L2 physical networks together with openvswitch. GRE
> > tunnel works perfectly in my case, except DHCP. Both networks have their
> own
> > DHCP server, and distribute addresses from their own address pool.
> >
> > I need to keep both DHCP alive so old computers would still work like
> > before. If needed, a computer could connect the other network simply by
> > assigning an IP address of that network manually.
> >
> > BTW, I haven't enabled OpenFlow
>
> Do you mean that you use the standalone fail_mode of an OVS bridge, i.e.,
> with
> a rule of action "NORMAL" installed by default?
>

Yes.

BTW, I have done a lot homework in order to find out this simple answer,
I'm new to openvswitch ;p


>
> >
> > Any simple solution for my problem ?
> >
>
> If the answer to my question above is yes, then you can simply achieve
> what you
> want by installing another rule preventing ingress udp traffic with
> source port 68
> from the GRE port, with a higher priority than the default "NORMAL" one.
> This
> makes both networks not receiving DHCPDISCOVER messages from the other end,
> and so do the two DHCP servers.
>
I finally implemented these flows, and they work pretty well. But I run
into some other problems:

1. How to make the flows persistent across reboot ? Current I wrote some
shell scripts, and invoke them from /etc/rc.local. Is there any other
solution ?

2. How OpenFlow ports are mapped to OpenvSwitch ports ? Currently I can
only get the mapping by using command "ovs-ofctl show vs0", fortunately the
mapping is fixed across reboot. Can I remap those ports manually, so I can
assign a more meaningful (OpenFlow port) number to the  GRE tunnel port ?


>
> > Thanks
> > Deng Yao
> >
>
>
>
>
> --
> regards,
> hunt
>
_______________________________________________
discuss mailing list
discuss@openvswitch.org
http://openvswitch.org/mailman/listinfo/discuss

Reply via email to