Hi,
We have an IBM Blade Chassis with 2 integrated BNT Layer 2/3 Copper Gigabit Ethernet Switch Modules. Latest firmware applied. Each integrated switch is connected to one upstream switch using one cable. 3 VLANs are trunked down to Blade Chassis through each link. Blade servers are running ESXi 4.1. vSwitch0 has Management network and 3 VM port groups connected to their target VLANs. The vSwitch0 is connected to vmnic0 and vmnic1. I can remove and add vmnic0 and vmnic1 from vSwitch0 and it all works fine, traffic fails over seamlessly as it should.
Now, the problem: Integrated switches have 6 external and 14 internal ports (one per blade). Loss of link on external port (cable unplugged, upstream switch dead etc.) will not result in link down state on internal port. In turn, ESXi running on blade still shows vmnic0 and vmnic1 as up and running and will not detect failure of upstream link and will not fail over the traffic.
I know there is a way to configure integrated switch to disable internal ports in case of external port link down situation, which would correctly inform ESXi that vmnicx is down and allow it to fail traffic to the other vmnic. I can't figure out where this setting is. I looked through available documentation and can't find anything.
The situation I have and mechanism for failover / informing internal ports in case of external port failure is nicely diagramed here, Figure 16-1. The only difference is that we have only one cable going from integrated switch to upstream switch, the rest is the same:
I know ESXi failover can be configured for Beacon Probing, but I'd much rather have internal port on the switch fail when external uplink fails, because I know this failover will work 100% reliably.
I'd appreciate if someone has the same config and can post what needs to be enabled for this to work.
Thanks