Best way of troubleshooting MC OIL is null?

Hey guys,

Doing some revision of multicast. Right now I'm just doing some basic labs but still having issues. First I ran a simple dense mode topology and everything was fine.

Then I changed to sparse mode and a static RP. I'm using the Vol1 topology and made R4 the RP and R1 is the sender and SW2 is the receiver. I am running this in Dynamips.

When I ping from R1 I can see that it registers with R4 but the traffic does not go through to SW2. It stops on R5 which says that OIL is null even though SW2 has ip igmp join-group configured. It seems that R5 does not understand that SW2 wants to receive the traffic.

I can see that SW2 has joined the group by using show ip igmp groups. When I debug PIM and IGMP on R5 I don't see messages from SW2.

Maybe SW2 has problems switching to the SPT, I tried setting SPT threshold to infinity but SW2 still tries to switch to the SPT.

Comments

  • What does the reverse path show?  Do an mtrace from the receiver to the RP and then from then from the receiver to the sender. 

    Brian McGahan, CCIE #8593 (R&S/SP/Security)

    Internetwork Expert, Inc.

    On Feb 22, 2012, at 11:39 AM, "daniel.dib" <[email protected]> wrote:

    Hey guys,

    Doing some revision of multicast. Right now I'm just doing some basic labs but still having issues. First I ran a simple dense mode topology and everything was fine.

    Then I changed to sparse mode and a static RP. I'm using the Vol1 topology and made R4 the RP and R1 is the sender and SW2 is the receiver. I am running this in Dynamips.

    When I ping from R1 I can see that it registers with R4 but the traffic does not go through to SW2. It stops on R5 which says that OIL is null even though SW2 has ip igmp join-group configured. It seems that R5 does not understand that SW2 wants to receive the traffic.

    I can see that SW2 has joined the group by using show ip igmp groups. When I debug PIM and IGMP on R5 I don't see messages from SW2.

    Maybe SW2 has problems switching to the SPT, I tried setting SPT threshold to infinity but SW2 still tries to switch to the SPT.




    INE - The Industry Leader in CCIE Preparation

    http://www.INE.com



    Subscription information may be found at:

    http://www.ieoc.com/forums/ForumSubscriptions.aspx
  • I agree.  Mtrace is a very handy tool to have in your troubleshooing tool bag.  I have had it help to identify issues many times.  

  • mtrace from receiver to the RP and towrds source. If mtrace is ok then

    debug ip mapcket on all routers in the multicast path from receiver to RP to check whether join messages reach to the RP then

    further troubleshooting steps.....

     

  • Hi Daniel! Keep cool :) I had issues with bmulticast and dynamips somehow concerning the data plane. Dont know exactly why but probably dynamips has sometimes issues in replicating the traffic correctly. Do you have the possibility to set this up in a rackrental?

    Or already burned all your tokens? If yes you can have some of mine if you like as your lab exam is not far away and I know the feeling if time is running out and something is not working.

     

    Regards!

    Markuss

  • Yes, I read Petrs post on multicast troubleshooting and tried some stuff. Here is mtrac from SW2.

    Rack1SW2#mtrace 150.1.4.4
    Type escape sequence to abort.
    Mtrace from 150.1.4.4 to 155.1.58.8 via RPF
    From source (?) to destination (?)
    Querying full reverse path...
     0  155.1.58.8
    -1  155.1.58.8 PIM  [150.1.4.4/32]
    -2  155.1.58.5 PIM  [150.1.4.4/32]
    -3  155.1.0.4 PIM  [150.1.4.0/24]
    -4  150.1.4.4

    150.1.4.4 is RP.

    Mtrace to sender.

    Rack1SW2#mtrace 155.1.146.1
    Type escape sequence to abort.
    Mtrace from 155.1.146.1 to 155.1.58.8 via RPF
    From source (?) to destination (?)
    Querying full reverse path...
     0  155.1.58.8
    -1  155.1.58.8 PIM  [155.1.146.0/24]
    -2  155.1.58.5 PIM  [155.1.146.0/24]
    -3  155.1.45.4 PIM  [155.1.146.0/24]
    -4  155.1.146.1
    Rack1SW2#mtrace 155.1.0.1
    Type escape sequence to abort.
    Mtrace from 155.1.0.1 to 155.1.58.8 via RPF
    From source (?) to destination (?)
    Querying full reverse path...
     0  155.1.58.8
    -1  155.1.58.8 PIM  [155.1.0.0/24]
    -2  155.1.58.5 PIM  [155.1.0.0/24]
    -3  155.1.0.1

    Seems there are different paths depending on source IP. If I do a ip igmp join-group on R5 Fa0/0 which is connected to SW2 then traffic flows even down to SW2 since OIL is no longer null. What am I missing here? I'm expecting to see more IGMP and PIM packets here. This what I am expecting:

    SW2 send IGMP join to R5 or PIM join depending on who is DR. SW2 gets rooted to RPT on R4. R1 registers with R4 as a source and then gets register stop. SW2 learns of source and prunes itself off the RPT and joins the SPT instead. I tried setting SPT threshold to infinity but it still switches over.

    However, if I debug on R4 I don't see any PIM join messages.

  • Thank you Markus, that is a very generous offer. I do have some few tokens left. No session free tonight but booked one tomorrow just in case to see if it is Dynamips issue or not.

    I do know that Dynamips does have issues with multicast but I don't want to be too fast to blame it on Dynamips if there is a risk that my knowledge is lacking. Also it is good practice for the TS to not have only smooth sailing when labbing :)

  • Seems there are different paths depending on source IP. If I do a ip igmp join-group on R5 Fa0/0 which is connected to SW2 then traffic flows even down to SW2 since OIL is no longer null. What am I missing here? I'm expecting to see more IGMP and PIM packets here. This what I am expecting:

    I think you are doing great, try on real devices rather than troubleshooting dynamips itself at least now. All the very best.

  • Mtrace is definitely a tool you wanna have when tshooting mcast issues. Petr post on the blog.ine.com is also a very very very good resource. I go back to it every once in a while to brush on tshooting mcast.

    Just my 2 cents.

Sign In or Register to comment.