Re: Seeking clarification of "unregistered packet"
La Monte Yarroll <[email protected]> Wed, 20 May 2009 10:07:04 -0400
| Newsgroups | gmane.ietf.magma |
|---|---|
| Message-ID | <1242828424.8363.178.camel@kartuli> |
On Wed, 2009-05-20 at 08:55 +0530, Bharat Joshi wrote: > La Monte H.P. Yarroll wrote: > > I have a couple questions about the following material from RFC 4541, > > 2.1.2: > > > > > 3) An unregistered packet is defined as an IPv4 multicast packet with > > > a destination address which does not match any of the groups > > > announced in earlier IGMP Membership Reports. > > > > > > If a switch receives an unregistered packet, it must forward that > > > packet on all ports to which an IGMP router is attached. A switch > > > may default to forwarding unregistered packets on all ports. > > > Switches that do not forward unregistered packets to all ports > > > must include a configuration option to force the flooding of > > > unregistered packets on specified ports. > > > > Is the ability to "force the flooding of unregistered packets on > > specified ports" only important on switches which are only processing > > v1 and v2 packets (or v3 packets ignoring "include source" and "exclude > > source")? > > No. I think it is required even for V3 capable switch. Though I think > while defining unregistered packet, it would have been better if RFC > should have used 'destination address or a combination of source and > destination address which does not....' instead of just 'destination > address which does not..'. Thanks, this is the revised definition of "unregistered packet" which I was expecting. > > If it is relevant for a switch which processes fully general v3 Joins, > > the definition of "unregistered packet" seems a little ambiguous. > > > > Consider traffic from source S1 to group G which no Joins seen. This > > traffic (S1,G) is unregistered packets. If we then see a Join for > > (S2,G), we have now seen the group G mentioned in a Join. How am I > > supposed to treat the (S1,G) traffic? Is it still unregistered packets? > > Or am I supposed to stop forwarding that traffic to "specified ports"? > > If this switch understand IGMPv3 and there is no IGMP join for this > group [none of the versions], as per the rule, it looks like that > (S1,G) should still be treated as unregistered packets. Good. That's what I thought, but it's not exactly what the RFC says. > > My problem may be that I simply do not understand what kinds of devices > > I might have on "specified ports" that I may need to flood unregistered > > packets to them. > > If you would have noticed, the second paragraph says that the > unregistered packet needs to be forwarded to the ports where an IGMP > router is available. Alternatively it can be forwarded to all ports > except from which it came in as well. 2.1.2 1) tells us: > 1) Packets with a destination IP address outside 224.0.0.X which are > not IGMP should be forwarded according to group-based port > membership tables and must also be forwarded on router ports. So we are already told to forward all multicast traffic on router ports. We've been treating "specified ports" as something other than routers, perhaps other snooping switches? The difference between "specified ports" and "router ports" is that we stop forwarding multicast traffic to "specified ports" as soon as we have identified a valid receiver for that traffic, but "router ports" always get all multicast traffic. I should clarify that we're implementing "must include a configuration option to force the flooding of unregistered packets on specified ports" and not "may default to forwarding unregistered packets on all ports". > Do you see any specific issues in doing this? With your revised definition of "unregistered packet" it makes a lot more sense, but I'm still trying to understand why the handling of "specified ports" should differ from "router ports".