0

I have a metro ethernet link between two offices with Cisco C3560X's on each end. Actually, I have two links. (We are switching from one provider to another.)

I can't get the new link to work. Pinging the vlan ip addresses yields no response. I'm running a packet capture on one end, and I'm seeing CDP, LLDP, and EIGRP HELO broadcasts from both ends. I'm also seeing ARP requests from the local end when I try to ping the far end. I'm NOT seeing any ARP requests from the other end, or any responses to the HELO's.

As a test, I added a static ARP entry on both ends. And now it works!

So either the WAN provider is blocking ARP, or some misconfiguration on my switches is preventing ARP from working.

What configuration could I have that prevents ARP from working on specific interfaces? It's not preventing ARP from working on our other WAN interface.

Or, is it more likely my WAN provider is blocking ARP?

Here are my configurations:

Switch S:

arp 10.0.100.89 30f7.0d5c.6bd1 ARPA

vlan 88
 name wan2
end

interface Vlan88
 description wan2
 ip address 10.0.100.92 255.255.255.248
end

interface GigabitEthernet0/12
 description wan2
 switchport access vlan 88
 switchport mode access
 spanning-tree portfast
 spanning-tree bpdufilter enable
end

Switch T:

arp 10.0.100.92 7426.ac7c.d049 ARPA

vlan 88
 name wan2
end

interface Vlan88
 description wan2
 ip address 10.0.100.89 255.255.255.248
end

interface GigabitEthernet0/48
 description wan2
 switchport access vlan 88
 switchport mode access
 priority-queue out 
 mls qos trust cos
 auto qos trust 
 no keepalive
 spanning-tree portfast
 spanning-tree bpdufilter enable
end
5
  • You've tried without the static ARP entries, haven't you?
    – Zac67
    Commented Sep 2, 2022 at 7:12
  • Yes, "As a test, I added a static ARP entry on both ends. And now it works!". Without the static ARP, it doesn't work.
    – longneck
    Commented Sep 2, 2022 at 13:19
  • 1
    It is probably the WAN blocking it. We have had various ISPs drop similar packets. We had to do a packet capture on both ends of the link at the same time to prove it. In our case it was spanning tree that wasnt getting through so our two WAN links were causing loops. Trying to explain the significance of this to the ISP helpdesk proved too painful and we ended up having to slightly redesign the network. Commented Sep 2, 2022 at 14:00
  • 1
    Blocking spanning tree I get. But ARP? That's just nuts.
    – longneck
    Commented Sep 2, 2022 at 14:28
  • @longneck Sorry, I wasn't sure of the actual config you tested with. I'd run a packet trace on each side to document the problem and then ask the ISP to solve it. You could work around it using static ARP but I wouldn't really do that. If it's supposed to be a transparent link (à la MPLS) then ARP needs to work.
    – Zac67
    Commented Sep 2, 2022 at 19:12

1 Answer 1

1

The vendor's NID on one end was not behaving correctly when the vendor logged in. Among other thigs, for example, the interface showed "not connected" even though it was clearly passing traffic. A reboot of the device solved all of the issues.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.