|
On 2/8/06, Graeme Fowler <graeme@xxxxxxxxxxx> wrote:
>
> Nothing at all happens, right? What version of keepalived is this?
Yup, no file is created and the interface does not go down.
~# keepalived -v
Keepalived v1.1.11 (01/23,2006)
> Thinking a bit deeper about this, I can see one problem straight away
> (probably unrelated actually, but go with me here): having all your
> announcements on eth0, but your VIP on eth1, means that if eth1 goes
> down (say a switch fails or someone trips over the cable) then there is
> nothing in your config to monitor it.
After you said that, I tested when just eth1 was down, and as you said, it
is not monitored.
> I use the "track_interface"
> vrrp_instance option for this, as follows:
>
> vrrp_instance DNS-FRONT01 {
> state MASTER
> interface bond0.500
> track_interface {
> bond0.99
> }
> ...rest of config
> }
>
> In which case if *either* bond0.500 or bond0.99 go down, the state will
> change. Maybe you also need something like that?
Im not understanding the bond0 concept. I dont see it in the user guide
either. Could you explain it a little?
Thanks,
Matt Chan
|
