From: zzk (ccie_99@yahoo.com)
Date: Mon Dec 15 2003 - 11:09:37 GMT-3
The mroute shows the correct entries, i.e. R3's
incoming interface is R4...
They are all Cat6, so default is IGMP snooping.
any other possibility?
Btw, got another question even for the normal
operation:
When all the links are up, our NMS found the multicast
traffics from R1 both reach Core 1 and R2.
Show IP mroute shows, the traffics reach R2 and pruned
by it.
I feel the traffics only need to reach Core 1, but why
it reaches R2 also?
Thanks!
--- Craig Dorry <chdorry@yahoo.com> wrote:
> If the traffic is reaching R3, then the tree is
> being
> built correctly. Can you do a show ip mroute for
> the
> group on R3 and R4? Also, are you running IGMP or
> CGMP on the L2 switch and R3/R4 (if CGMP)?
>
>
>
> --- zzk <ccie_99@yahoo.com> wrote:
> > Hello
> > Thanks for the suggestion. The commands are
> already
> > there. :(
> > We followed the sample anycast RP config with
> MSDP.
> >
> > Acutally, we noticed the multicast traffics reach
> R3
> > during the failover case, but it somehow did not
> > reach
> > the client on the layer 2 switch.
> >
> > Regards
> >
> >
> > --- Craig Dorry <chdorry@yahoo.com> wrote:
> > > Yes, the problem is that you are hitting RPs.
> > Look
> > > at
> > > the source and distribution trees. The server
> is
> > > sending multicast traffic and R1 will register
> > with
> > > the closest anycast IP (Core 1 RP). So you
> should
> > > have a source tree from Core 1 back to the
> server.
> >
> > > The client will register with Core 2 RP (due to
> R3
> > > to
> > > Core 1 being down). This will result in the
> root
> > of
> > > your source tree and the root of your
> distribution
> > > tree being in different locations.
> > >
> > > A way to fix this would be to implement MSDP
> > between
> > > Core 1 and Core 2. This would cause Core 1 to
> > > advertise the multicast source (server) to Core
> 2
> > > which would result in Core 2 joining the source
> > tree
> > > (ie building S,G state over to core 1)
> connecting
> > > your
> > > source and distribution trees.
> > >
> > > Config example would be:
> > > ip msdp peer <loopback IP of remote device>
> > > connect-source Loopback0
> > >
> > > ip msdp cache-sa-state (This is on by default
> in
> > > some newer codes)
> > >
> > > ip msdp originator-id Loopback0 (Specifies
> > source
> > > of MSDP packets).
> > >
> > > Hope this helps.
> > > Craig
> > >
> > >
> > >
> > >
> > >
> > >
> > > --- zzk <ccie_99@yahoo.com> wrote:
> > > > server
> > > > |
> > > > R1---------------------R2
> > > > | |
> > > > | |
> > > > Core 1 (RP)-----------Core 2(RP)
> > > > | |
> > > > | |
> > > > R3---------------------R4
> > > > | |
> > > > | |
> > > > SW1---------------------
> > > > |
> > > > |
> > > > client
> > > >
> > > > The network is designed using Anycast RP setup
> > for
> > > > fast failover time.
> > > > The network works fine during normal
> operation.
> > > > However when two links are down, i.e. link
> > between
> > > > Core1 to R3 and link between
> > > > R2 and Core2. The multicast traffic stops.
> > > >
> > > > server
> > > > |
> > > > R1---------------------R2
> > > > |
> > > > |
> > > > Core 1 (RP)-----------Core 2(RP)
> > > > |
> > > > |
> > > > R3---------------------R4
> > > > | |
> > > > | |
> > > > SW1---------------------
> > > > |
> > > > |
> > > > client
> > > >
> > > > The show ip igmp group and show ip mroute
> still
> > > show
> > > > the correct entries. Because of anycast RP,
> both
> > > > Core
> > > > 1 and Core 2 are active RP in the network,
> will
> > > this
> > > > cause this type of problem?
> > > >
> > > > thanks
> > > >
> > > >
> > > >
> > > > __________________________________
> > > > Do you Yahoo!?
> > > > New Yahoo! Photos - easier uploading and
> > sharing.
> > > > http://photos.yahoo.com/
> > > >
> > > >
> > >
> >
>
This archive was generated by hypermail 2.1.4 : Sat Jan 03 2004 - 08:25:41 GMT-3