07-15-2002 04:48 AM - edited 03-01-2019 11:53 PM
We have 2 Clustered File & Print servers setup using Legato CoStandby Server, each with an Intel Pro/100S Dual Port Server Adapter and a dedicated link between the servers. The ports on each card are configured as a Team, and are forced at Full duplex, 100 MB. On the Cisco 3524XL Switch, the ports are configured at Full duplex, 100 MB - but are not configured in a Port Group. With this configuration we were experiencing MAC Address Flapping on each switch that the servers were connected to. We also tried connecting each port to a different switch and received similar errors. I believe that our problem is due to the fact that the MAC address is the same for both ports on the dual port adapters, which was causing a loop between the 2 ports on the switches that the servers were connected? If this is true, could we hardcode the MAC address on both ports of the switch - or would we be better off scrapping the dual port adapter and using 2 separate NIC's?
07-15-2002 05:09 AM
A MAC address can only exist on a single port. Configuring static macs will not help as the CLI will not let you assign a unicast mac to multiple ports.
Why haven't you configured the switch for Port Grouping such that the 2 ports will act as a single port as you have done on the Intel NICs??
07-15-2002 06:13 AM
We attempted the Port Group option in a test environment and noticed performance degradation on the servers so we avoided it. Do you know of any performance issues when creating a port group? Should spanning-tree portfast be enabled? We would also like to avoid plugging both connections into the same switch to provide redundancy.
07-15-2002 06:24 AM
My mistake, you should not enable port-group on different switches, if the same switch, port-grouping should work fine.
The bottom line is the switches cannot have the same MAC address on 2 different ports in the same VLAN at the same time.
07-15-2002 06:48 AM
Thanks for your assistance. I think that we will probably get rid of the dual port cards and use 2 separate cards per switch to avoid the MAC issue.
07-15-2002 08:07 AM
You may still have a problem if you have a single IP bound to two different MAC addresses.
07-15-2002 10:39 AM
We presently have other servers that are using (2) 3Com 3C980 NICs that are bound to a virtual ip address and have not had any issues. Have you experienced any problems with this type of environment?
07-16-2002 01:20 PM
I believe that if you want link redundancy AND bandwidth teaming, you'd need two dual NICs.
One dual NIC would be configured for aggregating bandwidth connected to switch "A" as the primary (operating) channel, the second dual NIC would be configured for bandwidth teaming and set as a redundant (non-functioning until primary failure) connection to switch "B."
My understanding is that for any pair of ports, they can be configured for aggregation (like Etherchannel) or redundancy, but not both concurrently.
FWIW
Scott
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