Quoth Alan Maguire on Fri, Aug 11, 2006 at 09:45:13AM -0700: > makes sense. so anytime routeadm is run, it should check > the general/enabled property of each v4 routing daemon. if > these values conflict with the persistent ipv4-routing value, > (e.g a daemon is enabled but ipv4-routing is disabled or vice versa) > it is updated accordingly. additionally, if ipv4-routing is > enabled say, yet a daemon specified on the "ipv4-routing-daemon" > list is disabled (through "svcadm disable" presumably), that daemon > is removed from the list (i.e. we do all synchronization in routeadm, > removing the need for each daemon start/stop method to deal with > these issues).
Why not go all the way? Why keep a separate ipv4-routing value if it can conflict with the general/enabled properties of the services themselves? Why not report ipv4-routing as true if and only if any of the services are enabled, and if the user asks to enable it, enable the chosen service? > > This may also eliminate the need for routeadm -u on boot. > > the problem there is that routeadm settings may be changed > during boot on the basis of probing network configuration that > may have changed since last boot, so i think we're stuck with it. Please elaborate. David
