--- On Thu, 5/27/10, Brian Zill <bz...@microsoft.com> wrote:
> From: Brian Zill <bz...@microsoft.com> > Subject: RE: which interface to choose to send to destination link-local > address - any RFC? > To: "Parav Pandit" <paravpan...@yahoo.com> > Cc: "ipv6@ietf.org" <ipv6@ietf.org> > Date: Thursday, May 27, 2010, 10:24 PM > Hi Parav, > > >The IPv6 LL addresses are shared using out of band > protocol which works more at upper layer agnostic of L2 > properties and gives consolidated list of LL IP addresses > for all the VLANs. > > This is the part of your system that is broken. Your > out-of-band protocol should not be taking the link-local > addresses out of scope (i.e. losing track of what link they > are valid on). Instead, it should either (a) should > maintain the corresponding link information somehow, or (b) > use addresses that are valid in that scope (i.e. global > addresses). > > Typically, any reasonable protocol that uses link-local > addresses will learn any peer's addresses by receiving > packets from that peer on that particular link. The > protocol will then keep both the link-local address *and* > the scope-id for that address for later use in contacting > that peer. > > You should never need to resort to trying all links to > "find" the right one that has a machine with an interface > using that link-local address. Such an approach is > very error-prone. Since link-local addresses are only > guaranteed to be unique per link, you could have different > machines on different VLANs using the same link-local > address, and the one you find first may not be the intended > machine. > [Parav] This is the exact reason, I asked to double check on thought of doing ping6 on multiple interfaces. I was not convinced to perform ping6 on all the valid interfaces in unavailability of scope id. Let me revisit our out of band protocol and how can I modify it to ensure it goes out on the correct vlan. I really appreciate your inputs. Thanks. Parav > --Brian > > -----Original Message----- > From: ipv6-boun...@ietf.org > [mailto:ipv6-boun...@ietf.org] > On Behalf Of Parav Pandit > Sent: Thursday, 27 May, 2010 08:25 > To: Suresh Krishnan > Cc: ipv6@ietf.org > Subject: Re: which interface to choose to send to > destination link-local address - any RFC? > > Hi Suresh, > > Thanks for the script. > > The IPv6 LL addresses are shared using out of band protocol > which works more at upper layer agnostic of L2 properties > and gives consolidated list of LL IP addresses for all the > VLANs. > > Parav > > > --- On Thu, 5/27/10, Suresh Krishnan <suresh.krish...@ericsson.com> > wrote: > > > From: Suresh Krishnan <suresh.krish...@ericsson.com> > > Subject: Re: which interface to choose to send to > destination link-local address - any RFC? > > To: "Parav Pandit" <paravpan...@yahoo.com> > > Cc: "ipv6@ietf.org" > <ipv6@ietf.org> > > Date: Thursday, May 27, 2010, 8:04 PM > > On 10-05-27 09:28 AM, Brian Haberman > > wrote: > > > You will need the scope_id in order to use > > ping6. In order to send any > > > packet to a link-local address, you will need > the > > scope_id first. > > > > +1. So my question to you is, how did you get the LL > > address in the first place? That should give you clues > as to > > what interface you need the ping to go out on. If you > are > > really at a loss and you need to brute force probe, > you can > > write a small bash script to do this for you. > Something > > like > > > > for i in `ip address | grep "^[0-9]" | cut -d ":" -f > 2`; do > > echo "Pinging scope %${i}"; ping6 -c 1 > > <The_Link_Local_Address>%${i}; done > > > > Cheers > > Suresh > > > > > > > > -------------------------------------------------------------------- > IETF IPv6 working group mailing list > ipv6@ietf.org > Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 > -------------------------------------------------------------------- > > -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
