Comware

Hello,

I have worked on this on and off for weeks but the solution escapes me, maybe it is not possible.

We have a multicast source on VLAN 93 - the routers are WAN connected with static routes on VLAN 400 - at another phyiscal site the multicast receiver is on VLAN 141.

I have turned on IGMP and PIM-sparse for all VLANS and it shows PIM neighbors on VLAN 400 (the main router for each physical site) - but a PIM mroute will not form to the desitnation device.

Does anyone have any possible advice on this? I will work on posting the configs in a secure manner, with identifying information removed, but I thought that I would start asking around,

Thanks,
Dave

Hello @DPullin 

For PIM sparse mode to work it is not enough to just enable IGMP and PIM on all VLANs where sources and receivers reside. You need to define a Rendezvous Point (RP).  In PIM-SM the RP for all or specific multicast group is the distribution point for the multicast forwarding.  The designated router of the source will start sending the multicast to the RP and the designated routers of the receives will send their joins also to the RP.  

The RP can be defined statically on every switch in the PIM domain or can be elected using boot strap routers and candidate RPs.  The static method is simple but doesnt scale and doesnt allow redundancy. The dynamic is a bit more complex to configure but scales better in large environments.

You can find more details in the documentation but you didnt provide the models of your switches/routers. If it is an Aruba switch here Chapter 4 PIM-SM page 85

https://support.hpe.com/hpesc/public/docDisplay?docId=a00091313en_us

For Comware switches here for example the IP Multicast Guide of 5940 series., page 120.

https://support.hpe.com/hpesc/public/docDisplay?docId=emr_na-c05212291

 

Thank you for the reply, I really appreciate it!

I have turned on dynamic BSR and candidate RP for each of the routers in question, I have tried setting priority for each task also, and it still does not work - there are pending PIM join requests on the destination and the source routers but I don't know how to address these.

This is a small scale deployment with only 1 device needing the multicast - so static config would actually be fine, but I have tried and no luck so far.

I will post the configs of each router tomorrow - maybe that will clarify why it isn't working. The source router is a 3500yl, the central router is a 3810M, and the destination router is also a 3810M.

Thank you again, if you could please take a look at the config files after I download them tomorrow I would really appreciate it.

Thanks,
Dave

Below are the configs, thank you!

*** Since this question has been answered I wanted to remove the configs from public view *** 

*** I've sent them over DM to interested parties ***

The solution is still accurate though - turn off IGMP on ALL switches in the domain - not just the target PIM switches

Thank you for all of your help!

Hello 

Here is what I can add since I am also in this topic. @akg7 may add his view later.

Based on the configuration we see that source router 192.168.254.249 is the elected BSR and based on the configuration it should be also the RP for all multicast groups.  But did you check with "show ip pim rp-set" if all routers see it as RP?

The command show ip pim pending is showing us that the source routers has received IGMP joins in VLAN 93 and the receiver router in VLAN 141. Pending routes in SM however should have upstream and downstream interfaces. The receiver router should select the upstream interface with the best route to the RP. The source router should be able to show downstream interfaces if some PIM neighbors requested to receive the stream so we can see which routers have joined the tree. It would be helpful to check with the following command how exactly is seen his mroute on source and receiver router.

show ip mroute [group-addr ][ source-addr]

In my eyes the configuration should be actually OK. I see that you have enabled IGMP in VLAN 400 on all 3 routers. I understand that there are no mcast receivers in VLAN 400 so IGMP is not necessary there. I am not quire sure if having it can cause any harm. Because this command enables L3 and L2 IGMP at the same time and L2 IGMP will block multicasts if there are no IGMP joins in particular VLAN. (which is the case for the central router). Maybe it wouldnt be so difficult to test what happens if IGMP is disabled on VLAN 400 on all 3 routers.

@Emil_G - thanks for the reply, I really appreciate it!

Only the "SOURCE" router 192.168.254.249 shows an mroute, below is the command output:

 

hostname(pim)# show ip pim mroute

IP Multicast Route Entries

Total number of entries : 1

Group Address Source Address Neighbor VLAN
--------------- --------------- --------------- ----
239.10.0.104 172.20.93.30 172.20.93.30 93

hostname(pim)# show ip pim mroute 239.10.0.104 172.20.93.30

IP Multicast Route Entry

Group Address : 239.10.0.104
Source Address : 172.20.93.30
Neighbor : 172.20.93.30
VLAN : 93
Up Time (sec) : 85316
Expire Time (sec) : 300

Multicast Routing Protocol : PIM-SM
Unicast Routing Protocol : connected

Metric : 1
Metric Pref : 0
Assert Timer : 0
RP tree : No

Downstream Interfaces

VLAN State Up Time (sec) Expire Time (sec)
---- ---------- ------------------ ------------------
93 forwarding 85316 202

 

--- this output above doesn't look like it's connecting / forwding to VLAN 400 ... 

 

All 3 routers show the exact same thing for rp-set, the output is below:

 

hostname(pim)# show ip pim rp-set

Status and Counters - PIM-SM Static RP-Set Information

Group Address Group Mask RP Address Override
--------------- --------------- --------------- --------

Status and Counters - PIM-SM Learned RP-Set Information

Group Address Group Mask RP Address Hold Time Expire Time
--------------- --------------- --------------- --------- ---------------
224.0.0.0 240.0.0.0 172.20.141.1 150 133
224.0.0.0 240.0.0.0 192.168.254.246 150 133
224.0.0.0 240.0.0.0 192.168.254.249 150 133
224.0.0.0 240.0.0.0 192.168.254.253 150 133

 

Thanks for your help,

Dave

 

Hi,

Does anyone happen to have any additional information on this? There REALLY should be better documentation from HP on how to implement & troubleshoot PIM - I've read all of the manuals for our hardware, and the guides are nothing more than basic commands - and many of the troubleshooting commands that are supported on Cisco are non-existant on HP.

To make matters worse, something in the PIM configuration is causing total loss of connectivity to some switches - how is this possible? You'll see from my configs above, they are simple static routes between devices, but certain PIM configurations cause total loss of connectivity after a few hours - maybe it is a multicast packet storm? I did not think that this was really possible, especially without warning, but it is strange.

Any further help with this would be greatly appreciated, I never thought that a simple static PIM-SM implementation would be so difficult with this HP hardware.

Thanks,
Dave

Hello,

sorry for the delayed response. 

Did you try what happens if you disable IGMP on VLAN 400 on all the routers. You dont have mcast receivers there so IGMP is not needed.

I see that IP 172.20.141.1 is in the RP set. This is the IP of the destination router in VLAN 141 which is also the VLAN of the mcast receivers. The RP candidates were configured with source VLAN 400 so I am not sure how exactly this happened. Probably you tested something in between and changed the config. Maybe you can make sure that all RP candidates use the IP in the VLAN in which they have the PIM neighborhood, 400.

Anyway from what I see the static routing should be working. I assume that you have tested the ping between the multicast source and mcast receivers.

It is hard to tell if your connectivity issues are caused by PIM or something else. PM-SM should by default not route multicast traffic unless it is requested. IGMP on the receiver VLAN should make sure that multicast is only forwarded on the ports where multicast receivers are connected.

Thanks for the reply, we really appreaciate it.

I replied to the private message and we will begin compiling the requested information.

But, I mentioned this in the PM, we defaulted all the configurations and turned on PIM from the ground up again, and a new mroute did form! I need to test this now, but I will report back - we had tried with IGMP off before, but not since going back to default config for PIM (off) and starting over.

Thanks again,
Dave