Thanks for all the helpful input.

In my setup, I did have vlan300 and vlan800 created on top of eth0. Upon
the suggestion, I removed "eth0" from the xorp interfaces
configuration. After removing the configuration option
"default-system-config" from the interfaces, I now see that IGMP and PIM
are happier than before. I don't see the clueless error messages that some
configurations are not valid. It definitely works better now.

Still battling to get the multicast traffic routed end-2-end.

Frank
On Thu, Dec 8, 2011 at 5:52 AM, Ray Soucy <[email protected]> wrote:

> To be fair, I haven't been able to migrate to the current XORP yet, so
> these bugs are from 1.6 with patches; I may be speaking based on old
> information.
>
> If they've since been fixed then that's great.
>
> With 1.6 we do perform multicast (using IGMP, PIM-SM) just fine on
> 802.1Q interfaces in production for several connections.
>
> One note on VLANs and Linux; you will likely want to avoid using the
> native (untagged) interface for anything if you setup sub-interfaces;
> depending on your driver, and the way an application is implemented
> (raw sockets for example), the untagged interface may also see the
> tagged traffic as untagged.
>
> Also avoid the use of 31-bit prefixes for point-to-point networks like
> the plague.  You can configure them, but they are _not_ supported (use
> a 30-bit prefix instead); otherwise you will see some very odd (and
> hard to troubleshoot) breakage.
>
> On Thu, Dec 8, 2011 at 12:03 AM, Ben Greear <[email protected]>
> wrote:
> > On 12/07/2011 07:13 PM, Ray Soucy wrote:
> >>
> >> I haven't had good luck with default-system-configuration and VLAN
> >> interfaces; especially where PIM is concerned.
> >>
> >> You likely need to specify the addresses manually (in addition to
> >> creating the VLAN interfaces manually with vconfig).
> >>
> >> The XORP handling of VLANs has never been great; recent patches have
> >> been made to try an improve it; but it's still has work to be done
> >> before it's up to user expectations.
> >
> >
> > I don't ever test using default-system-config, but if you find
> > problems when using config from within Xorp, please report the
> > bug.
> >
> > In general, I'd suggest you drive all changes through xorp..ie,
> > tell it to set the IP instead of hoping it will figure things out
> > when you set it outside of xorp.
> >
> > You can script xorpsh using multiple -c 'command 1' -c 'command 2' ....
> >
> > It's not exactly user-friendly, but it works (for my uses, at least).
> >
> > And as always, patches are welcome, as are detailed bug reports
> > with xorp logs, OS & kernel version, etc.  Please submit such
> > bug reports to the bugzilla so we don't loose track of it.
> >
> >
> >
> >>> Being familiar with the Cisco routing CLI, these caveats seem quite
> >>> counter-intuitive... Could anybody help me understand if XORP was
> >>> deployed
> >>> outside the research/academia community?
> >
> >
> > Probably not so much.  We use it in our LANforge product, but we drive
> > xorp from our program (which builds networks with a GUI) using
> > programatically
> > created xorpsh commands and xorp config files.  Those
> > code paths are pretty well tested, but I'm certain bugs remain
> > to be uncovered.
> >
> > Thanks,
> > Ben
> >
> > --
> > Ben Greear <[email protected]>
> > Candela Technologies Inc  http://www.candelatech.com
>
>
>
> --
> Ray Soucy
>
> Epic Communications Specialist
>
> Phone: +1 (207) 561-3526
>
> Networkmaine, a Unit of the University of Maine System
> http://www.networkmaine.net/
>
_______________________________________________
Xorp-hackers mailing list
[email protected]
http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-hackers

Reply via email to