cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
5542
Views
0
Helpful
25
Replies

EtherChannel using L3

Jeff Harned
Level 1
Level 1

Hello All,

Could use some help.  I am looking at using EtherChannel to route multicast data from a source to a destiantion unit.  I would
like to try using L3 instead of L2 using two 3560x switches.  I have made several attempts to get data to the end device but have not succeded.
I have enclosed a short version of the configuration for the two switches including a block diagram.

Thanks for any help!

 

Regards,

Jeff

 

25 Replies 25

Jeff

Curious though thatwhen using Inter-Vlan Multicast Routing things were working!

If you mean everything was on one switch then it would.

If you look at your routing tables for the switches they each know about any directly connected networks and they will automatically route between them, assuming "ip routing" is enabled.

So when everything was on one switch you didn't have to worry about routing.

But now you have two switches and the source and receiver subnets are on different switches and each switch needs to know about both subnets to be able to forward packets.

Jon

Hi Jon

I did not want to continue without commenting on your last reply.  My previous adventure with multicasting invloved around as you mentioned with the one switch (Cisco 4) set up to Inter-VLAN Multicast routing.  The muticast inputs to it are the same, so that has not changed tesing with EtherChannel.

The bottom switch (Cisco 6) with fed with this multicast data to this switch configured as a Private Vlan switch with a combination of community and islolated ports to multiple receiver units.  So in this case I did not have "routing" to be concerned about just VLANS.

So what you are saying makes sense.  So this indeed is new territory for me and I
will continue on and let you know how things turn out.

Thanks.

Regards,

Jeff

Hi Jon,

I took your suggestion and used static routes in the configurations.  Multicast data is
now being received !! The EtherChannel is working since the multicast data continued to flow after pulling one of the port-channel cables out!

I have included attachments for you to view and comment on, including an updated block diagram.

This has presented some questions;

1)  I only had to use the ip route command on the bottom switch (Cisco 6) for
     multicast data to be received?
2)  In my switch configurations I used the command port-channel load-balance
     src-dst-ip,
is this necessary?
3)  It did not appear to make a difference to what I set the receiver's ip address to,
     data still was received?  I would of thought the ip address would at least have to be
     in the same subnet as the port?

Thanks.

Regards,

Jeff
    

Jeff

1) it's to do with the RPF (Reverse Path Forwarding) check. The destination IP of the multicast stream is a group address ie. it is not the receiver so the top switch does not need a route to the receiver.

Because the switches are PIM neighbors the top switch simply sends the multcast stream to the bottom switch.

However the bottom switch has to do an RPF check which means when a L3 device receives a multicast packet it looks at the source IP of the packet ie,. the sender and then it checks it's IP routing table for a route to the source. That route will tell it which interface to use to get to the source.

If that interface is the same interface the packet was received on then it passes the RPF check and the L3 device can forward the packet.

So when you added the route to the bottom switch the RPF check was passed and the packet could be forwarded.

2) it is a good load balancing algorithm to use and yes it should provide better distribution than src-dst-mac.

3) I can't test this but if the receiver is connected directly to the L3 port on the bottom switch then it may well work simply because the IGMP is a multicast and so the bottom switch will see the IGMP from the client.

You could check this with "sh ip igmp membership" on the bottom switch.

All the above said I would make the configurations standard ie. add a route to the top switch for the receiver subnet and also put the receiver in the same IP subnet as the L3 port it is connected to.

What you have currently may be working but it isn't recommended really ie. you should have IP routing setup properly before you then add multicast routing and clients should always be in the same IP subnet as their gateway on the L3 device.

Jon

 

Hi Jon

Yes, I do want to do the configuration the correct way!  The latest configurfation that I previously sent, the four receivers are directly connected to the bottom switch (Cisco6).

If I understand than I should route to the top switch (Cisco4) from the bottom switch?

I have included the ip igmp membership file for your inspection.

Thanks.

Regards,

Jeff

Jeff

Apologies for delay in getting back.

Basically you need to setup IP routing first so that you have connectivity between all devices ie. those connected to the top and bottom switches.

You need to setup the routes on both switches as discussed previously ie. both switches need to know about all non directly connected networks.

Up to you exactly how you do that ie. see previous post and if you have more queries please come back.

Then your multicast routing just uses the IP routing table and it should all work correctly.

Jon

Hi Jon

No apology is necessary!!!  You have been a great help!

I did some reconfiguartion with the LAN and as far as I can tell all is working great.
Multicasting is working just fine.  I ran  the show ip route command on the bottom switch (Cisco 6) and have a question about the results.  The  L & C codes are shown for all four ports that are connected to the receivers. What are these two codes actually telling me?  Not clear on the "C" code. It shows classless ip addressing
x.x.x.x/24 and x.x.x.x./32 per port.

I will be looking next at dynamic routing and will let you know how that goes!

Thanks.

Regards,

Jeff

Hi Jon,

Ok, I took a look at dynamic routing and tested using OSPF.  It looks like things are working!  I have alot going on at once here and if you would look at my configuration
for the two switches and the four receivers and comment? I also included both "show ip route" data for both switches.

If I understand correctly the routing table(s) would automatically update when there is a change to the network?  When I configured another port on the Cisco6 switch I noticed that the Cisco4 switch automatically udated its routing table to reflect the change on the Cisco6 switch.

Thanks.

Regards,

Jeff

 

Jon

Here are the other attachments.

Regards,

Jeff

 

 

Hi Jon

My error here. I pinged from the soure to the ip address for eth0 which is used for admn (Vlan1).  I use eth0 for for all admn comms. I use eth1 for data on the receiver end which is set at 10.1.10.2 as stated before.  No, I can not ping from the multicast source (Port 5/Vlan2) to the receiver port (10.1.10.2).

 

Regards,

Jeff

 

Okay, do you have routing setup on the switches ie. static routes or a dynamic routing protocol ?

Jon