ASR1002 odd layer 2 behavior

[axl8227]

I am facing an odd layer 2 behavior between 4 Cisco Router ASR1002-HX connected at the same Cisco Catalyst 2960 within the same vlan. Randomly each router can't reach the others within the same local vlan.

No layer 1 issues are present.

We face the same issue trying change L2 switch, sfp, cables, vlan, ip addresses, removing any L3 protocol and static route, clearing arp and setting static arp entries.

Updating the firmware from 16.3.3 to 16.3.6 did not solve.

Did someone ever face the same issue?

[Daniele Giordano]

What is the issue? Do you have a ping loss? how many pings? How many times? It's jus a ping or the routers are unreachable for minutes?

Regards.

[axl8227]

The issue is related to any type of traffic. Sometimes ping works, sometimes not. Furthermore iBGP sessions flap.

[Daniele Giordano]

Is it a flap of 1 second or the duration is higher?

[axl8227]

iBGP sessions survive for about 1 minutes then restart because of the lack of connectivity.

[Daniele Giordano]

During the issue, can you see the mac-address of the routers on the switch?

Regards.

[axl8227]

Yes, I can see the mac-address.

Probably I've found the mistake but I've got no idea how to fix it.

Routers interfaces that have that issue have an ip addresses within the 192.168.0.0/22 network.

I'm announcing that /22 prefix to some eBGP peers, but I would like also to announce 192.168.0.0/24.

To reach that goal I have configured a static ip route /24 out of that same network.

If I delete that static route everything seems to working fine, but the /24 is not announced to bgp peers because it is not within the routing table.

How can avoid this behavior?

[Daniele Giordano]

Try this trick:

ip route 192.168.0.0 255.255.255.0 null0

Let me know what happens.

Regards.

[axl8227]

Same behavior because the network is matched at the same way.

ip route 192.168.0.0 255.255.255.0 GigabitEthernet0/0/0

Has the same result of

ip route 192.168.0.0 255.255.255.0 null0

[Daniele Giordano]

mmm, you must have a route installed in your global routing table in order to announce it...

what is the pourpose of this route?

[axl8227]

I'm trying to  announce the aggregate /22 and the 4 more specific network /24 from each of the 4 routers.

Each router has an ip address of the /22 and the subnet mask /22.

[Jason M.]

Your original post stated that you were not able to ping across the VLAN. Does that mean all of your routers are on the same subnet and can't ping the other routers on that subnet? Or are you saying you can't ping other interfaces on the those routers like loopbacks and IP address on the non-directly connected interfaces?

If you want to assign a /24 to each router and there are no other paths to those subnets then you should still use the null0 so the advertisement isn't dependent on the single interface being up.

[axl8227]

I have 4 routers. Each router have an interface in a /22 subnet. Randomly ping does not work between themselves. Those 4 interfaces are in the same vlan within the same layer 2 switch. I was guessing it was a routing issue due to the /24 static route for bgp purpose, but if I remove it that odd issue is still present. Do you know any unknown feature on the Cisco ASR2001HX that can be compatible to this kind of behavior?

[Jason M.]

I'm not sure about the answer to that question.

I would suggest performing a long duration ping and make note of the intervals of when it responds and when it does not. Perform it from each of the routers to see if they stop at the same time. That can provide clues to what the source of the problem is.

Prior to starting those pings check to see if you already have an ARP entry.

Do those routers go to a single switch?

[Richard Burts]

I believe that checking for arp entries is a good place to start. It would also be helpful if the original poster can provide some details of how these routers are connected and the details of the configuration of the router interfaces.

HTH

Rick