hi folks i've updated the design document with SMF-related changes suggested during this discussion. key points are:
- use of temporary disable/persistent enable to implement semantics of "enable at next boot" associated with routeadm -e. this removes the need for a call to "routeadm -u" at every boot. running "routeadm -u" causes temporay disable to be removed, so that current config becomes "next-boot" config - use of new variable "routing-svcs" so users can specify routing services via routeadm. the v4/v6 split thus is not perpetuated. as an example, configuring in.routed/in.ripngd as services to be enabled at next boot becomes: # routeadm -s routing-svcs="route:default ripng:default" # routeadm -e ipv4-routing - use of legacy-routing:ipv4/ipv6 services for routing daemons that have no smf counterparts - removal of /etc/inet/routing.conf. properties therein are distributed among relevant routing services - state of ipv4/ipv6 routing, current and persistent, reflect daemon states - ipv4 routing is currently enabled if any ipv4 daemon services are enabled presently(either general-ovr/enabled=true or general/enabled=true). ipv4-routing is persistently enabled if any ipv4 daemon service has general/enabled=true. - each daemon has separate properties for each option. on upgrade, the daemon start method migrates the daemon arguments to individual properties. see http://www.opensolaris.org/os/project/quagga/quagga-design.pdf note that it does not address moving static routing configuration to smf, for the quagga project the priority is getting dynamic routing sorted. we'll file an RFE for the static routing work with the suggestions provided of course. huge thanks to everyone who participated! alan This message posted from opensolaris.org
