STP switchover

Hi guys!

I am looking at shortest path tree switchover...and got a question.

When the leaf router (router clostest to the receiver) gets its S,G entry finally and forwards the multicast packets it will try to switchover and make a shortest path tree to the source.

So far so good.

The leaf router sends a join message towards the first hop router instead of the RP. And when it comes to the point where there is a router in the chain that has a different interface pointing to the source than to the RP, this router will send the join message to the source of the multicast stream and a prune to the RP (so that the way to the RP is not used anylonger).

But what happens with the routers that already have a source tree from the first hop router to the RP. How do they get the prune?

Thanks in advance!

Regards!

Markus

Comments

  • But what happens with the routers that already have a source tree from the first hop router to the RP. How do they get the prune?

     

    The RP also sends Prune back towards Source and this Prune travels hop by hop until it reaches the first-hop router in order to remove this path from source to RP from the tree.

     

  • Nice one!

    Thanks for the fast reply and for clearification!

     

    Regards!

    Markus

  • You are welcome. [:)]

  • Okay I still got one here. Its going into details.

    What is if the RP sends a prune message to the first hop router to disable communication over the shared tree BEFORE the SPT from the first-hop router to the leaf router is built? Wouldnt there mcast packets missing or at least the multicast stream would stop for a moment?

    Is there any mechanism that ensures that the SPT between the first hop router and the leaf router is used before the rest gets pruned? I mean this should happen before the SPT to the RP or the shared tree fom the RP to first-hop router is torn down right?

     

    Thanks in advance guys!

    Regards!

    Markus

  • Is there any mechanism that ensures that the SPT between the first hop router and the leaf router is used before the rest gets pruned? I mean this should happen before the SPT to the RP or the shared tree fom the RP to first-hop router is torn down right?

    Leaf router receivers both flows just for a moment before it sents prune message upto the RP. Using debug commands we can verify this that prune sents towards RP only after the shortest path tree is built from leaf router to first hop router.

  • Okay! Perfect! That makes sense! But if the tie is not broken at the leaf router itself but on a different router in the chain some hops away (the point where the way to the RP and the way to the source is split) what then? I mean the pim prune is not sent by the leaf router in this case or am I wrong? Th epim prune is sent by the machine where the split comes in.

    Does this "split"-machine wati until it has both streams and the prunes of the path to the RP?

    Thanks for your patience in this case!

    Regards!

    Markus

  • Hi Markus,

    Rather than explaining myself, I like this, have a look on it:

    https://supportforums.cisco.com/thread/2031946

  • Okay! Perfect! That makes sense! But if the tie is not broken at the leaf router itself but on a different router in the chain some hops away (the point where the way to the RP and the way to the source is split) what then? I mean the pim prune is not sent by the leaf router in this case or am I wrong? Th epim prune is sent by the machine where the split comes in.

    Does this "split"-machine wati until it has both streams and the prunes of the path to the RP?

    In this case, leaf router has incoming interface for (S,G) and (*,G) are same so it can not send prune message to the upstream router however stp switchover always initiated by leaf router. Now the router in the path where incoming interface for (S,G) and (*,G) are not same will send prune to upstream router towards RP. This router put the interfaces in the incoming list when it receives the (S,G) join from the leaf router and send this join towards the source.

  • Rather than explaining myself, I like this, have a look on it:

    https://supportforums.cisco.com/thread/2031946

    Does it have what Markus is trying to ask ?

     

  • In this case, leaf router has incoming interface for (S,G) and (*,G) are same so it can not send prune message to the upstream router however stp switchover always initiated by leaf router. Now the router in the path where incoming interface for (S,G) and (*,G) are not same will send prune to upstream router towards RP. This router put the interfaces in the incoming list when it receives the (S,G) join from the leaf router and send this join towards the source.

    Hi!

    Thanks for the post! That should be good for me now. So the router where the path is split sends the prune. Perfect!

  • Hi guys!

    I wrote down the pim spt and pim register process in my blog. Would be nice if you could check that out...thats my understanding of it, maybe there is a fault in there, especially in the 5th step. Devil is always in the detail!

    http://chasingmyccie.wordpress.com/2012/03/03/pim-sparse-mode-pim-register-pim-spt-switchover-process/

     

    Thanks in advance for response!

    Regards!

    Markus

  • I wrote down the pim spt and pim register process in my blog. Would be nice if you could check that out...thats my understanding of it, maybe there is a fault in there, especially in the 5th step. Devil is always in the detail!

    http://chasingmyccie.wordpress.com/2012/03/03/pim-sparse-mode-pim-register-pim-spt-switchover-process/


    Nice affords putting it
    all together. Only this statement might be more prone for misinterpretation at least for me.

    5) The RP receives the prune message and prunes the traffic of the SPT to the last hop router.

     

     

     

  • Ah okay...

    Hmmm....

    Should it rather be like: the prune messages go upwards the rp and then the traffic is pruned. Furthermore the rp torns down the spt to the first hop router.

    Does that fit? Sorry i am asking, i am not focused anymore after 12hrs of leraning.

     

     

    Regards!

    Markus

  • Or better I should write that the first hope router sends a PIM prune also upwards the source-tree to the RP?

    Thats the last thing I need to know here.

     

     

    Thanks in advance!

    Regards!

  • Updated it!

    Should be fine now!

    Used "developing ip multicast networks" from cisco press to clarify 2 steps.

     

     

    Regards!

    Markus

Sign In or Register to comment.