David Miller wrote: > From: Al Boldi <a1426z@xxxxxxxxx> > > David Miller wrote: > > > From: Yanping Du <ypdu2001@xxxxxxxxx> > > > > We got complaints on this Linux behavior, since > > > > customers feel it confusing, and don't want to see > > > > ICMP replies for an eth ifc if it's down. > > > > > > We're not replying to the "eth interface" we're replying > > > to the IP address which is still assosciated with this host > > > even though the interface it is assigned to is down. > > > > Down implies a temporary inactivation. > > The interface is going down, the address hasn't been removed, they are > seperate and distinct operations. Exactly. > If you remove the address, we'll > stop responding to it. Removing the address implies a permanent inactivation. > > You are effectively disabling functionality by implementing the host > > based model on top of the interface. > > "/sbin/ip addr del xxx" works perfectly fine. Imagine two apps, one down'ing the interface, the other up'ing the interface. With your scheme this isn't possible, unless app2 knows the ipaddr. > And it's not "on top of", it's "in conjunction with". Not really, as you are forced to remove ipaddr data to get the ifbased model, which implies a reduction in features. > Look, the default is not changing, that's what existing Linux systems > do and have done for more than 10 years, and therefore we'll break > things everywhere if we change. No need to change the default; all we need is a compatibility option, like: # echo ifbased > /proc/../conf/all/compat # echo hostbased > /proc/../conf/all/compat # echo OSIbased > /proc/../conf/all/compat Thanks! -- Al - To unsubscribe from this list: send the line "unsubscribe linux-net" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
