Sstrhsrtj

I have a corner-case where I need to leave the family inet6 configuration under IFL present while hosts in that network should not use IPv6. This means, that the logical interface of a router will contain a family inet6 and a link-local address:

root@r1> show interfaces ge-0/0/0.10   

  Logical interface ge-0/0/0.10 (Index 332) (SNMP ifIndex 534)

    Flags: Up SNMP-Traps 0x4000 VLAN-Tag [ 0x8100.10 ]  Encapsulation: ENET2

    Input packets : 0

    Output packets: 146

    Protocol inet6, MTU: 1500

    Max nh cache: 75000, New hold nh limit: 75000, Curr nh cnt: 0, Curr new hold cnt: 0, NH drop cnt: 0

      Addresses, Flags: Is-Preferred

        Destination: fe80::/64, Local: fe80::206:a00:a0e:fff0

    Protocol multiservice, MTU: Unlimited



root@r1> 

This should mean, that when hosts manually configure global unicast address, then in theory, they can reach the destination, but the packet is not routed back to them. However, this is not even viable because of RPF check. Also, hosts can reach the router over IPv6, but I don't see this as a problem.
Are there any other risks/disadvantages with this setup?

In my mind, this is unusual, and would probably get picked up as a configuration error in an audit unless you document it really well - consider leaving a comment on the interface to explain to future generations why this is being done.

Also consider the "hosts can reach the router over IPv6" - this means that you should also update your loopback filter to protect your routing-engine from connections arriving on IPv6 (eg: control-plane protocols, remote access, SNMP etc.)

I'd like to know your corner case, and wonder if putting family inet6 on a loopback interface wouldn't be a better way to solve it? (a loopback IP wouldn't be exposed to any other hosts without interface routes)