Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 
Community Member

email cluster and EIGRP

I have a confusing situation and I need some advice. We have two mail servers running as a cluster. And we have layer three switches (3750) as access switches and the 6509's as core. At the 3750's, we are running EIGRP and configured the access VLAN. Now the problem is as soon as we configure the cluster, we loose connection between the cluster and the Microsoft outlook from the users PC's but we can ping the cluster from the same PC's. with the same switch configurations, we can access the mail should we remove the cluster (use one exchange box). However, should we remove the routing, move the VLAN to the core and configure the 3750's like normal layer 2 access switches with , everything works perfect. we have similar problem with users accessing the exchange server from the WAN.

Any advice will be appreciated

Hall of Fame Super Silver

Re: email cluster and EIGRP

Hello Emmanuel,

may you provide some more details about how is the cluster done.

It is not clear how the issue you see can be related to EIGRP routing protocol.

Are the single servers in different IP subnets ?

How are the switches configured about the cluster ?

Do they know about the virtual server if any ?

You say they are able to work if the servers are put in the same IP subnet using the 3750 as L2 switches for that vlan.

The difference can be any form of broadcast or multicast communication between the real servers that is possible in the same L3 subnet but it would require specific config to be supported on different Vlans.

To be sure when placed in the same subnet try to sniff traffic from servers ports using a monitor sessions or search info on MS web site that can explain how the servers communicate between them.

Hope to help



Re: email cluster and EIGRP


Im not sure how clustering is performed, but given your description of the proble, it seems as though both email servers are on the same vlan and a cluster virtual IP is what is advertised to the Outlook clients.

If thats so, is there L2 adjacency bewteen the 2 3750s? It's a routed access layer, which means that you cannot span the vlan across both switches; each vlan is confined to one switch, unless, of course, you have an L2 trunk between the access switches. Doing that would simulate what you were talking about with regard to deploying the legacy L2 access layer with trunks up to a routed distribution/core layer with an L2 trunk between them to support the HSRP configuration.



CreatePlease to create content