Working with a customer today, the idea of Address Pools came up during a discussion about Cisco UCS. During this discussion we tossed out the idea of using a Address Naming Convention in the Cisco UCS environment. The purpose being, given a MAC, World Wide Port Name (WWPN) or World Wide Node Name (WWNN); I should be able to identify what Cisco UCS Domain, UCS Fabric, NIC and Blade number.

Here is what we use internally, it is what we shared with the customer:

For MAC Addresses:

Sample MAC address: 00:25:B5:01:A0:01
In blue, 00:25:B5, bytes 1-3 represent the vendor identification of the MAC Address (Cisco).
In violet, 01; byte 4, represents the UCS Domain identifier in HEX, giving us 256 unique Domains.
In green, A0l; byte 5, represents the interface identifier. This is displayed in HEX (0-F)
In red, 01; byte 6, represents the blade number in decimal format (1-8).

This address naming convention can then be applied to the WWPN and WWNN addresses throughout the Cisco UCS environment. Resulting in the ability to quickly identify (and hopefully troubleshoot) your UCS environment if needed.

xx:xx:xx:xx:xx – Vendor specific bytes 1-3 for MAC and bytes 1-5 for WWPN and WWNN (Cisco)

For WWPN:
WWPN address: 20:00:00:25:B5:01:A0:01
20:00:00:25:B5 – Vendor specific bytes 1-5 (Cisco)

For WWNN:
WWNN address: 20:00:00:25:B5:01:A0:01
20:00:00:25:B5 – Vendor specific bytes 1-5 (Cisco)

 

After hours and hours and hours of troubleshooting, we’ve finally figured out why our brand new C260 ESX hosts that we’re installed in a VPC were not functioning correctly. At the most, only 3 interfaces were ever able to come online, bringing the fourth would take the entire host down.

The problem?

The C260 uses one of the physical MAC addresses for vmk0 instead of a traditional ESX virtual MAC like 00:50……, this throws the whole gammet out the door.

After hours of digging I found this… http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1031111

The workaround? re-create vmk0, which is quite simple.

Delete vmk0

Re-create vmk0

Reboot the host and away you go!

with the C220’s, we were also forced to re-create vSwitch0 before everything would work.. yuk.