To test SLB, I setup a new vlan 111 (and interface vlan 111) and put the VIP on the range of IP addresses associated with this vlan. A second switch was setup with an identical configuration, vlan 111 was trunked between the two switches, and the VIP was then configured on both sides with the "inservice standby vlan-111-hsrp" command. HSRP was obviously configured on the "interface vlan 111" on both switches.
The configuration works with the loadbalancing etc. under normal conditions.
What happens when the link between the two switches is cut is that the vlan interface goes down. I believe that IOS thinks that vlan 111 is unused because there are no longer any physical interfaces in the vlan when you do "show spanning-tree vlan 111", and so it drops the vlan interface. HSRP then fails completely, which means that the VIP on both switches goes out of service.
We do run spanning-tree in "transparent" mode.
I would like to know if there is an elegant way to keep vlan 111 active in this scenario?
I could always trunk this vlan on other links where it is not actually required, but this sounds untidy to me.
I am also wondering if I really need a separate vlan for the SLB definitions? Would things work better if I deleted the new vlan completely (retaining the VIP address, though) and did that inservice standby story on another vlan which does have active servers? I want to use this for DNS so I really want to keep the VIP off the IP subnets of any other vlans.