IPv6 host routes are not removed from frr after configuring BGP

[georgeman]

I recently configured BGP (via os-frr) in order to peer with my ISP so they can properly install an IPv6 route for my /48 prefix. I announce the /48, they announce ::/0, everything works, we are all happy.

But I notice this on Routing -> Diagnostics -> General -> IPv6 routes:



For every single IPv6 that the firewall sees, either internal or public, I get a new /128 entry that is not removed. Right now, after frr has been running for about 12 hours since the last restart, it is showing 6000+ entries. They roughly show up every 20 secs (NDP related?).

Is this normal behavior? Please note these are not currently installed in the kernel, only show up in zebra:

Code Select Expand

root@opnsense1:~ # netstat -rn6 | grep -c fdXX:XXXX:XXXX:XXXX::XXXX
0
root@opnsense1:~ # vtysh -c "show ipv6 route fdXX:XXXX:XXXX:XXXX::XXXX"
Routing entry for fdXX:XXXX:XXXX:XXXX::XXXX/128
Known via "kernel", distance 0, metric 0
Last update 00:01:03 ago
Flags: None
Status: Installed
* directly connected, vmx0_vlan788
Routing entry for fdXX:XXXX:XXXX:XXXX::XXXX/128
Known via "kernel", distance 0, metric 0
Last update 00:01:58 ago
Flags: None
Status: Installed
* directly connected, vmx0_vlan788
Routing entry for fdXX:XXXX:XXXX:XXXX::XXXX/128
Known via "kernel", distance 0, metric 0
Last update 00:02:34 ago
Flags: None
Status: Installed
* directly connected, vmx0_vlan788

...

(222 similar entries in total)(Yes, I know it's pointless to obfuscate ULAs)

Is this just cosmetic or something is leaking badly?

Thanks!