01-23-2012 06:22 AM
We're replacing our older catalyst switches with new SG300 family switches and have a Microsoft NLB cluster for some services that run in multicast balancer mode.
We currently do L3 routing to the network with the cluster and have the following IOS configuration line in the specific switch to let users on other subnets to access the services.
arp 10.20.1.226 03bf.0a14.01e2 ARPA
How do we replicate this using the SG300 series in L3 mode? Whenever I try to add a manual ARP entry I get an error message that says that the MAC address is not a valid unicast address?
Greatful for any tips.
03-07-2013 11:06 AM
I also need to add a static ARP entry for Multicast with IGMP. I get the same message on a SG500-28P.
Can we draw the conslusion that it is not supported on SG300 or SG500 series switches?
03-07-2013 12:36 PM
Hi, the switch distinguishes by the first byte if it is a unicast address or a multicast address. The first 3 bytes are the OUI. The first byte of the OUI is the most significant. While the last 3 bytes are vendor assigned the last, 6th byte is the least significant.
A fundamental instance of a unicast MAC would be 0001.3344.5566. While a fundamental multicast MAC would be 0100.AAAA.BBBB.
Notice the bolded numbers. Those are the most significant bits in the most significant bit.
On the screenshot posted the 03:bf:0a:14:01:e2 does not work. However 02:bf:0a:14:01:e2 will work. The reason is, 03 represents multicast therefore is an illegal value.
You can set up multicast MAC group
bridge multicast filtering
interface vlan 1
bridge multicast address 03:bf:0a:14:01:e2
-Tom
Please mark answered for helpful posts
03-26-2013 12:15 PM
Tom,
Are you saying the switch is able to suport NLB in mulicast (with IGMP) mode?
If so, can you please explain this a bit more?
Sent from Cisco Technical Support iPad App
03-26-2013 06:06 PM
Hi Boudewijn, I don't see why it couldn't. NLB in multicast mode just adds a multicast address to the cluster which allows the inter host communication. But the limitation may be when the ARP reply is sent by a host in the cluster in response to the ARP request it will map the unicast address to the multicast address. The switch may reject this which is why I suspect the original poster is making the inquiry.
In hopes to make this work (which I do not know if it can...) you can try to make a bridge address such as my example for the multicast settings. You can try to become creative with a static ARP entry for your unicast address' as well since the table will support the unicast address.
I don't think the switch will have any problems with NLB in unicast mode so long as you have a second NIC to permit the inter host communication.
-Tom
Please mark answered for helpful posts
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide