Fallback from failover (HSRP) static ARP fail, known issue?

[Bas van Delft]

Hi all,

I have this situation that we need a static ARP entry for Microsoft NLB. In our HSRP-router setup we have to configure this static ARP entry on both routers (of course). During a failover the second router will provide communication to the NLB cluster without problems. However, when we returned back to the primary HSRP-router, communication was failing. We had to remove the Static ARP and re-apply this configuration entry to have it work again. Is this a known issue that sometimes happens? It's quite hard to recreate this situation in a lab (cost-wise) and we are not that keen on just testing this in a live environment either.

Would this be a one time thing, that just happened to strike us, or is this a bug that someone else here has encountered as well? For completeness, we're running two 2851's with IOS 12.4-3i Adv Sec. image.

The routers are connected to a switched environment, on different switches.

BR,

Bas van Delft

[Amit Singh]

Hi Ban,

Are you using a multicast MAC NLB deployments on the servers? If yes the you have to use this staic ARP entry. I had this issue once when I used 3560 switches for the same deployment. I didnot use the Multicast MAC and intrun relied on unicast deployment. Everything worked for me during that testing after that.

How are you servers connected? Which switches are your using? Why you dont have this arp entry on L2 switches where servers are connected. Please use the link below as an example :

http://www.cisco.com/en/US/products/hw/switches/ps708/products_configuration_example09186a0080a07203.shtml

Cheers,

-as

[Bas van Delft]

Hi Amit,

Yes, we are using Multicast MAC NLB. therefore we use the static ARP, to route to other subnets (VLANs). That's why I don't follow why I should configure a static ARP on the L2 devices (which are 3Com by the way).

But that is another issue we are talking about, which I solved with this static ARP on the routers.

My issue is that failover worked OK, but when the first router came back up, the ARP entry did not work somehow as the NLB address wasn't pingable from other subnets. We had to delete the static ARP entry and re-apply it to get the NLB working again.

Any experience in such case?

BR,

Bas