|
On Sun, Aug 04, 2002 at 09:11:28PM +0000, Juri Haberland wrote:
> carla quiblat <carlaq@xxxxxxxxxxxxxxxx> wrote:
> >
> > Hi there,
> >
> > I'd like to ask how the back-up balancer takes over the haresources of
> > the main balancer? Do both balancers contain the same haresources when
> > heartbeat is started? Which part of ha.cf says that it's a master or the
> > slave?
>
> You are supposed to have identical haresources files on both balancers.
> So you have the _same_ node name in both files. This name determins the
> master node.
>
> > The lvs script that I wrote based on examples from the
> > documentations already configures the VIP of the machine, and this lvs
> > script is called or used by the haresources file. So that if both
> > heartbeat are ran in both balancers, they both use the same
> > VIP at the same time which causes a conflict. However, if I don't include
> > it in lvs script, it doesn't get configured by heartbeat anyway. I get
> > this error:
>
> Heartbeat should start the resources only on one host at a time.
> If it does something different then there is a configuration error.
>
> > heartbeat: 2002/08/05_02:26:37 ERROR: unable to find an interface
> > for 192.168.0.1
> >
> > Are there other options for the "IPaddr" portion of the haresources? Like
> > maybe IPaddr::192.168.0.1::eth0:110?
> >
> > Where do I tell heartbeat or ldirectord to use this certain interface for
> > the VIP (e.g. eth0:110)?
>
> I'm not quite sure whether the IPaddr script can do this. But you can
> use your on script. However, it is important that the first script after
> the node name returns "running" or "stopped" if called with the
> parameter "status"!
You can specify which interface to use (eth0, eth1, etc...) but not
which virtual interface (e.g. eth0:110). The IPaddr script
works this out for itself by searching for a free virtual interface
to use.
--
Horms
|
