multicast over frame-relay issue !!

Unanswered Question
Feb 16th, 2009

i have simple hub and spoke tpology

the hub configured with to map statments of frame-relay point to both spoks with broadcast word

spoks has one with broadcast word for the hub other one only map the other spok and point it to the local DLCI

simple and everyone ping each other

now i run ospf over this network

and we know by defualt it is non-broadcast then i add the neighbor command on the hub point to each spoke

now evrthing works fine as well

Multicast

the hub is the RP and discovery comand entered as well

one router behind the sopks joined an igmp group

each router run pim sapre-mode and the hub in addition to that has pim nbma-mode

the problem is

everything OK every on ping the joined igmp group

ecept from the other spoke

when i ping forst packet get replay and anything after first packet NO

in the hum and the RP as well what i noticed

first ping there is outgoing interface for that spoke for the subsequent the outgoing interface shows as NULL !!!!

however when i run ospf as poin-to-multipoint works fine

with EIGRP the same issue (split-horizon disabled)

any Idea Guys !!!

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 4.7 (9 ratings)
Loading.
Edison Ortiz Mon, 02/16/2009 - 14:29

Please post the output from:

show ip mroute

show ip pim rp mapping

show ip pim interface

show ip pim neigh

show ip os neigh

from the 3 routers.

__

Edison.

Marwan ALshawi Mon, 02/16/2009 - 14:42

hi Edison

i dont have this output now

but i can give what you looking for frothis show comands

for mroute

spoke show the outgoing interfac as the FR interface

but the issue with the hub

although it has ip pim nbma-mode

as i mention firt time with show ip mroute

the hub shows

for the sopke i ping FROM the ip address in S and the destination ijmp grup in the G

and the incoming and outgoing interfaces correct

howver fter first success ping

on the hub the outgoing interface for this spok ( the one i ping from ) the outgoing interface shown as NULL

here why i am confsed

even if i put he command ip pim spt-thrshould infinity no lck !!!

for mappping they all see the hub as the RP

becuase i did in the spoke autorp listner !!!

if you need anymore details let me know ?

but why the first packet go then the outgoing interface in the hub change to NULL ??

thanks again

Edison Ortiz Mon, 02/16/2009 - 15:00

I can't duplicate your problem here and it could be an underlying problem with the IGP.

Here is my lab, R1 (Hub) R2 and R3 (Spokes)

R1:

interface Serial1/0

ip address 192.168.123.1 255.255.255.0

ip pim nbma-mode

ip pim sparse-mode

encapsulation frame-relay

no ip mroute-cache

serial restart-delay 0

frame-relay map ip 192.168.123.2 102 broadcast

frame-relay map ip 192.168.123.3 301 broadcast

no clns route-cache

R2:

interface Serial1/0

ip address 192.168.123.2 255.255.255.0

ip pim sparse-mode

encapsulation frame-relay

ip ospf priority 0

serial restart-delay 0

frame-relay map ip 192.168.123.3 201

frame-relay map ip 192.168.123.1 201 broadcast

no clns route-cache

R3:

interface Serial1/0

ip address 192.168.123.3 255.255.255.0

ip pim dr-priority 0

ip pim sparse-mode

encapsulation frame-relay

ip ospf priority 0

no ip mroute-cache

serial restart-delay 0

frame-relay map ip 192.168.123.2 103

frame-relay map ip 192.168.123.1 103 broadcast

no clns route-cache

end

__________

I made sure R1 was the DR for that segment by changing the OSPF priority on the spokes.

I configured R1 with Auto-RP by using loopback and here are the mappings all routers:

PIM Group-to-RP Mappings

Group(s) 224.0.0.0/4

RP 150.1.1.1 (?), v2v1

Info source: 150.1.1.1 (?), elected via Auto-RP

Uptime: 00:39:54, expires: 00:02:49

_____

Now, I'm going to do a join on R3 and do a ping from R2..

R3(config)#int lo0

R3(config-if)#ip igmp join-group 229.3.3.3

R3(config-if)#do show ip mroute 229.3.3.3

(*, 229.3.3.3), 00:00:07/00:02:52, RP 150.1.1.1, flags: SJCL

Incoming interface: Serial1/0, RPF nbr 192.168.123.1

Outgoing interface list:

Loopback0, Forward/Sparse, 00:00:07/00:02:52

_____

Verify the join was sent to the RP:

R1#sh ip mroute 229.3.3.3

(*, 229.3.3.3), 00:01:42/00:02:47, RP 150.1.1.1, flags: S

Incoming interface: Null, RPF nbr 0.0.0.0

Outgoing interface list:

Serial1/0, 192.168.123.3, Forward/Sparse, 00:01:42/00:02:47

______

Ping from the other spoke:

R2#ping 229.3.3.3 repeat 10

Type escape sequence to abort.

Sending 10, 100-byte ICMP Echos to 229.3.3.3, timeout is 2 seconds:

Reply to request 0 from 192.168.123.3, 152 ms

Reply to request 1 from 192.168.123.3, 92 ms

Reply to request 2 from 192.168.123.3, 128 ms

Reply to request 3 from 192.168.123.3, 136 ms

Reply to request 4 from 192.168.123.3, 140 ms

Reply to request 5 from 192.168.123.3, 132 ms

Reply to request 6 from 192.168.123.3, 116 ms

Reply to request 7 from 192.168.123.3, 104 ms

Reply to request 8 from 192.168.123.3, 192 ms

Reply to request 9 from 192.168.123.3, 124 ms

_______

Check the hub mroute table now:

R1#sh ip mroute 229.3.3.3

(*, 229.3.3.3), 00:03:54/00:02:32, RP 150.1.1.1, flags: S

Incoming interface: Null, RPF nbr 0.0.0.0

Outgoing interface list:

Serial1/0, 192.168.123.3, Forward/Sparse, 00:03:54/00:02:32

(192.168.123.2, 229.3.3.3), 00:01:04/00:02:54, flags: T

Incoming interface: Serial1/0, RPF nbr 0.0.0.0

Outgoing interface list:

Serial1/0, 192.168.123.3, Forward/Sparse, 00:01:04/00:01:55

HTH,

__

Edison.

Marwan ALshawi Mon, 02/16/2009 - 15:50

Hi Edison

first Thank you for this config i will give it try tonight

however i did the same as u did mostly except i did not put

th following command on the hub inteerface

no ip mroute-cache

dose this make deffrence ??

also for rp in the hub i did it like

int lo0

ip pim spare-mode

ip pim send-rp-announce looback 0 secope 16

ip pim send-rp-discovery loopback 0 scope 16

in addition to that

IF I DONTPUT the AUTORP LISTNER on SPOKES they will not know about the RP

becuase u know there is know dense ???!!!

let me know if something missing

as i mentioned

why it wor when i have it as ospf poin-to-multipoint

but with non-broadcast and eigrp Not ??!!!

Thanks again

Edison Ortiz Mon, 02/16/2009 - 15:56

however i did the same as u did mostly except i did not put

th following command on the hub inteerface

no ip mroute-cache

dose this make deffrence ??

No, I did it for the debugs.

It works the same w/o it.

IF I DONTPUT the AUTORP LISTNER on SPOKES they will not know about the RP

becuase u know there is know dense ???!!!

Correct.

let me know if something missing

Try forcing the hub to be the PIM DR as well as the OSPF DR.

as i mentioned

why it wor when i have it as ospf poin-to-multipoint

but with non-broadcast and eigrp Not ??!!!

point-to-multipoint does not use the concept of the DR and that's the reason I told you, something is wrong with the underlying IGP.

I'll check with EIGRP later on and see if I can duplicate the problem.

__

Edison.

Marwan ALshawi Mon, 02/16/2009 - 16:13

Try forcing the hub to be the PIM DR ?

how can this be done and what is the defrence from from RP role ?

by the way sometimes they mentioned about Mapping agent MA is that means the router with the discovery command ??

which we have it in my case on the same router !!??

also if i have question say this router need t be the group maping

sometimes i see it solved as above with discovery and filltering if required HOEVEVER somtimes i see it used as BSR ??

why and how i can kno which one to use

Thanks again mate

Marwan ALshawi Mon, 02/16/2009 - 16:18

i will try it tonight :)

by the way why u did used

neighbor 192.168.123.3 Serial1/0

neighbor 192.168.123.2 Serial1/0

while we done need to send roting as unicast

it should work without it right ?

Marwan ALshawi Tue, 02/17/2009 - 02:37

Hi Edison

i have attached the router config i have tried th ething u have mentioned but no luck

R1 hub

R2 and R3 spokes

R4 just a router connected to R3 and has defualt route to R3 and i joined it to an ijmgp group for ping test

first ping from R2 go and the subsequent NOT

other routers ping normally

also if u see in router 1 the outgoing interface for R2 to the gorup i use is Null

in router 4 for all other router shows null but they can ping except R2

any idea will be helpful cause i got confused all configred ok but not luck !!

thank you

Attachment: 
Edison Ortiz Tue, 02/17/2009 - 07:08

Unable to duplicate.

Based on your attachment, the Hub is Pruning the S,G although you have a pim nbma on the interface so it seems to be a bug.

Try with a different IOS.

Posting configs with R4 attached to R3.

HTH,

__

Edison.

Attachment: 
Marwan ALshawi Tue, 02/17/2009 - 13:19

hi Edison

based on my config it should work i mean the config is ok ?

because i just get confused about it

about the comment bellow from mohamed

the RPF not the issue becuase it is just simple hub and spoke

even not full meshed network

and the traffic should go R1 !!!

Thanks Edison, by the way if you think my config OK becuase as i told you only with OSPF non-broadcast and eigrp dose not work !!

do you think becuase i use GNS3 for this ?!

Thanks again

Edison Ortiz Tue, 02/17/2009 - 13:48

I will let Mohamed discuss his logic on RPF since I don't see any RPF problem on the topology.

As you see from my configs, EIGRP & OSPF worked identically.

BTW, my lab was also done on GNS3, using 7200 routers and 12.2(25)S15 image.

HTH,

__

Edison.

Mohamed Sobair Tue, 02/17/2009 - 03:06

Marwan,

what is 225.0.0.1 represets? If you are Pinging from IGMP group assigned on router-2 interface to the Source which should be 150.x.x.x should be successful and its only a source of check.

The RPF method state the following:

If the incoming Interface for the multicast packet doesnt lead back to the source which is (R1) in your case,then the RPF fails and multicast packet is dropped.

Make sure that R1 source of the multicast packet is reachable via the same Interface of the recieved incoming multicast packet. (You will need to source the Ping from that Interface) OR you have to configure the following:

IP mroute "Source addresss" "subnet Mask" exit interface.

HTH

Mohamed

Mohamed Sobair Tue, 02/17/2009 - 14:06

Edison,

The original poster has claimed that he can perform ping from all routers except R2? and I have suggested the following:

Since all routers recieved their mapping information and they have route back to the RP, then the possible cause is the Source of the ICMP packet from R2 is not the incoming interface for the multicast packet. I have also suggested him to configure Static IGMP Group on one of R2's Ethernet Interface and perform normal ping sourced from the serial Interface while making sure the outgoing interface for the ICMP packet is an RPF interface.

Another question is why Marwan is performing ping to 225.x.x.x Group from R2? what is this going to accomplish?

Mohamed

Marwan ALshawi Tue, 02/17/2009 - 14:22

Hi Mohamed

i do this for CCIE LAB !

and if it is RPF problem then in this case i could not have been ble to get any response or reply when i ping

while i do have first packet with replay and the others failes which exclude RPF!

Edison i use the same as yours but not sure what IOS version exactly i will check tonight

by the way my config same as yours then can i consider my config right? and the logic i do my multicast over FR with NBMA is right as well ??

one more thing why u used neighbor command with eigrp without it it should work as well ??

thank you

Edison Ortiz Tue, 02/17/2009 - 15:12

Marwan,

I tried to match your config as close as possible and I'm unable to duplicate what you are seeing.

Per your config output, you get the join:

From R3

(*, 225.1.1.1), 00:13:38/stopped, RP 150.1.1.1, flags: S

Incoming interface: Null, RPF nbr 0.0.0.0

Outgoing interface list:

Serial1/0, 10.1.1.3, Forward/Sparse, 00:13:12/00:02:35

Then the ip pim nbma-mode isn't doing what is supposed to do then the router prunes the S,G:

(10.1.1.2, 225.1.1.1), 00:00:05/00:02:55, flags: PT

Incoming interface: Serial1/0, RPF nbr 0.0.0.0

Outgoing interface list: Null

With my config, I get the join from R3:

(*, 225.1.1.1), 00:02:47/00:02:43, RP 150.1.1.1, flags: S

Incoming interface: Null, RPF nbr 0.0.0.0

Outgoing interface list:

Serial1/0, 192.168.123.3, Forward/Sparse, 00:02:47/00:02:43

and the nbma mode kicks in by listing S0/1 with the unicast information from each router of the link:

(192.168.123.2, 225.1.1.1), 00:01:21/00:02:45, flags: T

Incoming interface: Serial1/0, RPF nbr 0.0.0.0

Outgoing interface list:

Serial1/0, 192.168.123.3, Forward/Sparse, 00:01:21/00:01:39

As you can see, the ping is coming from R2 to R3.

________

As for the EIGRP neighbor statement, it does not factor into problem. It works the same with or without.

__

Edison.

Marwan ALshawi Tue, 02/17/2009 - 15:28

Ok upon this

i can guess it is more IOS issue as said rather than config

becuase my config since first time sae as yours.. and with Mcast as you know there is no much config it just concepts need to be applied correctly

i think i need to change the IOS otherwise it will not work for ever becuase no way to configure it other than this way right ..!!

thanks for you help

Edison Ortiz Tue, 02/17/2009 - 16:24

Marwan,

I have good news. I was able to duplicate your problem.

Try creating a loopback in R2.

I deleted the loopback in R2 and the hub pruned the S,G like in your lab.

Very interesting and I wonder if this is a GNS3 behavior. I don't have real routers to test right now but this is very strange.

__

Edison.

Edison Ortiz Tue, 02/17/2009 - 16:43

Further testing...

It seems the behavior (at least on GNS3), you need to have more than one Layer3 interface in up/up state if you have the router acting as a Mcast Server.

I did the opposite and setup R2 as the Mcast client with R3 being the server.

R3 is connected to R4 so it has two Layer3 links hence there is no need for loopback.

Multicast is working fine on this scenario.

__

Edison.

Marwan ALshawi Tue, 02/17/2009 - 19:10

Hi Edison

really interesting and strange !!!!

i will try it

by the way do i expect it in the CCIE LAB as well :)

i think this is only in gns3 because i have never seen a document or bock mentioned about this point !!

i think the copy right for this will be yours :)

thanks again i will let you know after try it

by the way i have to make this loopbackaddress Mcast enabled ??

Edison Ortiz Tue, 02/17/2009 - 21:10

You won't expect that kind of behavior in the CCIE lab as most routers will have multiple L3 interfaces.

The fact that we created a router with just one interface and in your case without a loopback, created the issue.

Routers aren't that good for Mcast server/client function, their function is to be Mcast routers.

No, you don't need to enable PIM on the loopback.

HTH,

__

Edison.

Marwan ALshawi Wed, 02/18/2009 - 03:23

actually i did what you have notced

but it was good to some extend

ping from R3 to R2 as R2 joind a group dose the ping ffrom 8 i get about 3 replyes whihc means ok but somthing wrong maybe with GNS3 !!!!

from R2 to R3 or R4 the same problem

anyway i think this is the only way to configure the Mcast over FR

thank you for your help

Marwan ALshawi Wed, 02/18/2009 - 12:49

Hi Edison

i have somthing confused me

in multicast section

i see in internetworkexpert workbooks sometimes when the question say this router will be resposible for RP to group maping

they solved as RP agent with discovery comaand

and in others useing BSR

when i have to use BSR

thanks

Edison Ortiz Wed, 02/18/2009 - 13:07

Why people fail in the CCIE Lab is mostly due to interpreting what is asked.

It's very hard to tell you when to choose one technology over the other without reading the whole workbook.

Often the answer is tied to another underlying technology that is part of that Lab.

Other times, the answer is there with some hints like:

Use PIM version 2.0

or

Use PIM version 1.5

or

Use PIM described on RFC 5059

Understand?

I'm afraid, you need to practice and practice. There isn't a straight answer to this type of query.

__

Edison.

Marwan ALshawi Wed, 02/18/2009 - 14:19

Hi Edison

i agree with you

because with multicasting and IPV6 there is no alot practice i think i need to cover all posible case as i can

for example the hints you have mentioned very nice

thats why i need to know what hints should i see to use this one or that one

for example maybe the type of network effect the choice with frame-relay and ospf

we need to see what frame-relay we hav P2P or P2multipoint and so on then we can decied

i am aware aboutthat thanks

but for example i will give now somthing i am thinking about confusing HSRP solusion

there is to labs in he IE workbook

both of them asked to track the connection to the FR cloud OK

buut one of them solved with normaway which is track interface serial x/x

the other one solved in a way i liked it

lets say the 2 spokes are HSRP peeers and they have to track the connection to the FR cloud i mean the active one as above

but what they did

the created a tunnel interface between the active HSRP(spoke) and the hub router and they put the tracked interface as the tunnel

it is interesting one i tried it and worked fine

but which one rom your experince will be considered the right one ( or both works)

the question ask the same and the case mostly the same

thanks again :)

Actions

This Discussion