Routing issue

[Steve Coady]

Hello

A while back my company swapped out 6509 for Nexus 7k's. Several devices are intermittently having connectivity over Multicast to local servers. Other devices

I am getting the following traceroute from a switch, connected to Distro 7K's via GigabitEthernet1/0/1 and GigabitEthernet1/0/2 (Routed links). No trunks.
 

3750#traceroute 10.20.96.80

Type escape sequence to abort.
Tracing the route to 10.20.96.80

  1 10.99.222.110 0 msec
    10.99.224.110 0 msec
    10.99.222.110 0 msec
  2 10.99.224.232 0 msec
    10.99.224.234 0 msec
    10.99.224.228 0 msec
  3 10.224.8.240 0 msec
    10.224.8.200 0 msec 0 msec
  4 10.20.254.235 8 msec
    10.20.254.229 8 msec
    10.20.254.235 0 msec
  5 10.20.96.80 0 msec 0 msec 0 msec

3750#traceroute 10.20.96.82

Type escape sequence to abort.
Tracing the route to 10.20.96.82

  1 10.99.224.150 0 msec
    10.99.222.110 0 msec
    10.99.224.150 0 msec
  2 10.99.224.230 0 msec
    10.99.224.228 0 msec
    10.99.224.234 0 msec
  3 10.224.8.240 0 msec 0 msec
    10.224.8.200 8 msec
  4 10.20.254.235 8 msec
    10.20.254.229 0 msec
    10.20.254.233 0 msec
  5 10.20.96.82 0 msec 0 msec 0 msec

EIGRP CONFIG:

router eigrp 10
 network 10.0.0.0
 eigrp stub connected
 no auto-summary
 passive-interface default
 no passive-interface GigabitEthernet1/0/1
 no passive-interface GigabitEthernet1/0/2

!
interface GigabitEthernet1/0/1
 no switchport
 ip address 10.99.224.151 255.255.255.254
 ip pim sparse-dense-mode
end

interface GigabitEthernet1/0/1
 no switchport
 ip address 10.99.222.151 255.255.255.254
 ip pim sparse-dense-mode
end

The traceroute does not know which link to use.

The directly connected Distro device traceroute has no issues. 4 lines and it is there.

Should EIGRP be changed to prioritize one uplink over the other?

[procopius1980 · Accepted solution]

The ip addresses on your G1/0/1 and G1/0/2 don't always match what you are getting on hop one in the traceroutes.  Shouldn't you see 10.99.224.150 and 10.99.222.150 only?  Why do you see 10.99.222.110?

Also, I have never seen anyone use a /31 subnet mask.  I will have to look that one up.

As to your question, I am reasonably positive that running a traceroute on a device that has two equal cost links will show the traceroute going over both links.  I am fairly sure that is normal.  I don't think you need to do anything, but I am unfamiliar with routing multicast between one layer 3 device to another.

[Alexander Cousins · Accepted solution]

Hi

Looking at your traces it looks like you have equal cost multipath (ECMP) routing and whilst the output of traceroute can confusing the the traceroutes may be valid (I guess there is no loss of IP connectivity). I would not look to change the routing solution and try and fix whatever in multicast is failing. Here are a few suggestions which may (or may not) be of use.

By default all multicast should be using your G1/0/1 interface because of its higher IP anyway. The issue could be with return traffic failing the RPF check when received if received on your other interface.

Check where the RPF path is using 

show ip rpf (should see gi 1/0/1)

show ip mroute count (look for RPF failed frames)

Other: 490/0/9 

If they dont get you any further You could confirm it its an ECMP issue by dropping the connection with the lowest IP address and seeing if the problem persists.

You could try issuing the  ip multicast multipath Command (just googled it myself) which may address any RPF issues.

HTH