From: Venkataramanaiah.R (vramanaiah@gmail.com)
Date: Sat Dec 03 2005 - 18:38:49 GMT-3
Hi Josef,
I thought about using the tunnel, however when i tested it, the
multicast traffic does not seem to take the tunnel. It still flows through
the FR network natively and the situation does not change..
-Venkat
On 12/3/05, Josef A <josefnet@gmail.com> wrote:
>
> Thought of a tunnel between R2 and R3 ?
>
> Josef
>
>
> On 12/2/05, Venkataramanaiah.R < vramanaiah@gmail.com> wrote:
>
> > Hi,
> >
> > I have the following FR topology..
> >
> >
> > ______R2_____R4
> > /
> > R1--
> > \______R3_____R5 ( 239.5.5.5)
> >
> >
> > I have PIM Sparse mode running everywhere.. and of course with PIM
> > nbma
> > mode at R1.
> >
> > R2 is the RP and R1 is the MA. R5 has joined some group, Say
> > 239.5.5.5 (Will call this Grp5 henceforth).
> >
> > There are two issues in this topology. Let me discuss them one by
> > one.
> >
> > 1) When R5 sends PIM join to the RP (R2) via R3, the joins are sent
> > directed
> > to R2 from R3. This prevents the join to become successful. We can fix
> > this
> > by adding a static mroute at R3 for the RP pointing to R1. THis will let
> > the
> > shared tree to be built from the RP R2 via, R1-R3 and the receiver R5.
> > Hope
> > there are no other solutions for this problem.
> >
> > 2) When we ping the Grp5 from R4, the packets would reach the receiver
> > for a
> > while. After sometime, the last hop DR (i.,e R5) tries to switchover to
> > SPT
> > and this breaks the multicast flow in the FR cloud. We can probably fix
> > this
> > by using SPT threshold infinity at R5. Then the traffic would flow
> > smoothly
> > to R5. The main issue that i am unable to address is when we have some
> > group
> > say Grp3 joined at R3 and if we ping this group from R2. In this case,
> > setting spt-threshold infinity at R3, does not seem to help. Since the
> > source is on the same network address in the FR cloud, the ping works
> > for a
> > while through the shared tree via R1, but after sometime, R3 seem to be
> > tearing down the source tree because it thinks the source is directly
> > connected and it stops the mulitcast flow for a while. I am sure you
> > guys
> > would have faced this issue, so I am wondering if there is any solution
> > for
> > this problem..
> >
> > Thank you
> > -Venkat
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Mon Jan 09 2006 - 07:07:50 GMT-3