Loopback netmask

Question

I am trying to implement IP network redundancy for a zone using OSPF and zebra. The setup is similar to the configuration example included with the zebra software in /etc/sfw/zebra/README.Solaris. Everything is working with one exception. I can't seem to get the loopback address for the zone to be plumbed as a /32. I've tried using zonecfg like this:
zonecfg:my-zone>add net
zonecfg:my-zone:net>set address=w.x.y.z/32
zonecfg:my-zone:net>set physical=lo0
zonecfg:my-zone:net>end
This method works fine for the physical networks, but the loopback address always comes up as a /8. I have to manually ifconfig the address after the zone comes up to make the routing work. That would be fine if a reboot always happens from the global zone using zoneadm, but if the non-global zone reboots on it's own the networking will be broken. Anyone have ideas on a workaround?
I'm running Solaris Express 11/04
Thanks
carisdad

Answer 1

I am having this issue as well. We use Foundry ServerIrons for load-balancing and pool management. Since a virtual source address is used for the src/dst IPs, we need that virtual ip on the server's loopback. No matter what I set the netmask to in the zonecfg or /etc/inet/netmasks, it always sets to a /16, 255.255.0.0 mask. The IP is in the 140.0.0.0 block.
Anyway, this is a pain for zoneadm to completely ignore the netmask given in zonecfg for a lo0 interface. I will try the vni interface now, but it would be seemingly much simplier if the lo0 code just worked.
Thanks!