all clients disconnect on failover

[Sebastian Vieira]

On 10/10/06, Graeme Fowler <graeme at graemef.net> wrote:
>
> I have a theory here, though... you're doing NAT. Therefore the director
> is the default gateway for the realservers - is the address of their
> default gateway the same as the VIP?
>
> If not, you need another vrrp_instance defined which makes the default
> gateway fail over to the backup director.
>
> Graeme



Okay, i was able to check out the settings for the realservers. It's a bit
complicated (for me), so let me just give you my findings:

realservers:
ip: 192.168.14.x
gw: 192.168.14.254
netmask: 255.255.255.0
they're in a seperate vlan

director:
ip: 192.168.50.110
gw: 192.168.55.254
netmask: 255.255.248.0
seperate vlan

I've tried to change the gw on the realserver to the ip of the director, but
then i get a network unreachable error. And even if that was possible, i've
learnt that the realserver go back to their default settings upon reboot
(which happens every now and then), so changing them is not an option. What
i could do is put the director in the same vlan as the realserver and
configure the director to use lvs-dr instead. I've no idea if that will make
a difference though. Is LVS-NAT the only way to go here?

Changing the mac address is not an option since i have to bring the entire
interface down before i can do that, in which case keepalived sets the node
in fault modus and renders itself useless.


regards,

Sebastian