So I need help. I’m not at work, so posting config isn’t really an option, but here goes:
I have a 4500 with a port channel to two Nexus 5k’s. There are two ports to each 5k all in the same port channel. Let’s call it po13.
On the 5k’s, I have two ports on each going back to the 4500, all in po13 and vpc’d.
All channel group members are mode active.
I have a vpc domain on both 5k’s. Let’s call it vpc domain 1. There is also a po1 between the 5k’s that is the vpc peer link which is po1.
5k1 has been in place for a long time (it originally had a plain trunk to the 4500). So, I copied the config from 5k1 to the new 5k (5k2), and simply changed the interface vlan ip. Then I proceeded to setup all the said port channels and vpc’s.
I basically mirrored the portchannel, vpc, peer link, and mgmt0, from one of our other datacenters that have the exact same physical architecture. I only changed the portchannel numbers, vpc domain ids, and mgmt0 IPs.
5k1 pings out fine using its interface vlan ip. I can remotely manage it. All is well.
5k2 is not. I can ping across mgmt0, but I cannot ping out from the interface vlan. It says, “no route to host.” I have entered the default route 0.0.0.0/0 x.x.x.x (my upstream default gateway). Show ip route returns nothing because I can’t ping my default gateway.
The weird part is, ALL portchannels on ALL THREE switches are up. On 5k2, show cdp neighbors shows 5k1 AND the 4500. If there was something wrong with the po13 between 5k2 and the 4500, I would think that traffic would route through 5k1.
I guess I’m puzzled as to how the portchannel could be up and passing cdp, and apparently lacp traffic, yet I cant ping out. 5k2 is useless at the moment. Any initial thoughts on things I can check? Should I just wipe the 5k2 config and build it from scratch?
When I go back to work, I’m going to unplug the peer link and set up a simple trunk from 5k2 back to the 4500 and see if I can ping out then...still, any initial thoughts would certainly be welcome.