Nic Teaming problem UCS (blade) with Vmware

[Indika Wasala]

All,

We have below scenario,

There are two VMs in separate two ESXi hosts (blade servers). Each VM is having two nics connecting to two port groups; production (prod) and heart beat(hb). In UCS, two nics are added to each host and are in active-active configuration for all the port-groups in vmware standard switch. We have created prod and hb vlans in fabric interconnects and allowed them in the two connected nics to hosts. Ping does not work between two VMs for the heart-beat network. But when we configure only one nic as active for the heart-beat port group in vmware standard switch, it started working fine. Why it is not working in active-active configuration..

Thanks,

[Walter Dey]

How is load balancing on vswitch configured ?

what is your North bound switch configuration; e.g. 2 separate switches, connecting to either FI ? and do the switches have a trunk with the proper Vlans of the UCS domain. (maybe you have a vpc)

Could be, that on the same vlan, egress traffic of VM-1 is in fabric A, and ingress to VM-2 is on fabric B; therefore although this is L2 end to end (within in Vlan), the traffic has to exit the UCS domain, and enter the domain on the other fabric.