|
> so it's listening to 0.0.0.0 on one interface (say eth0)
on 0.0.0.0 proto VRRP, this is the design of multicast.
> I have just the vrrpd part of keepalived running on a pair of routers.
> These machines are also running dhcpd (and a few other services) in failover
> mode each with their own failover mechanism. It would be nice to have them
> all failover together in from a central monitoring process, rather than
> each application failing over independantly. I know this isn't your problem,
> but it would be nice if each application didn't write its own failover.
Why not letting VRRP control dhcpd. I mean, dhcpd is brought by VRRP
notify script. That way you will benefit VRRP takeover speed for your
application. Kind of, failover is handled by VRRP, that way application
configuration stay as simple as possible.
> it would be nice to be able to ask the state in a programatic way.
you can also run ip address list and see if VRRP VIP are present
for quick diagnostics.
> would it be better to move subsequent questions to the keepalived mailing
> list?
yes, that way other people with complexe VRRP use will be able to join the
discussion.
regards,
Alexandre
|
