dhcpcd-discuss

On 03/04/2020 18:41, Alarig Le Lay wrote:
> On ven.  3 avr. 18:20:08 2020, Roy Marples wrote:
> > Could be a dhcpcd bug!
> > Does this patch help?
> >
> > Roy
>
> It doesn’t seem so:
>
> Apr  3 19:37:32 core01-arendal dhcpcd[2825]: enp2s0: Router Advertisement from fe80::5e5e:abff:fe43:1ece
> Apr  3 19:38:09 core01-arendal dhcpcd[2825]: enp2s0: fe80::5e5e:abff:fe43:1ece is unreachable
> Apr  3 19:38:10 core01-arendal dhcpcd[2825]: enp2s0: soliciting an IPv6 router
> Apr  3 19:38:11 core01-arendal dhcpcd[2825]: enp2s0: Router Advertisement from fe80::5e5e:abff:fe43:1ece
> Apr  3 19:38:19 core01-arendal dhcpcd[2825]: enp2s0: fe80::5e5e:abff:fe43:1ece is unreachable
> Apr  3 19:38:20 core01-arendal dhcpcd[2825]: enp2s0: soliciting an IPv6 router
> Apr  3 19:38:32 core01-arendal dhcpcd[2825]: enp2s0: no IPv6 Routers available
> Apr  3 19:38:37 core01-arendal dhcpcd[2825]: enp2s0: Router Advertisement from fe80::5e5e:abff:fe43:1ece
>
> I applied the patch against 8.1.7 sources version.

Yes,

Running ip monitor in one terminal and issuse this command in another
ip neigh del fe80::5e5e:abff:fe43:1ece dev eth0

ip monitor WILL report that FAIL is being sent for the address.
So this look like a linux kernel bug! It should not send failure on GC of NDP 
entries, only when really unreachable.
You should report that upstream - I think my patch is still valid though in the 
hope it works one day.

> So, my question is: Is it a good idea to put this command in a post-up
> hook (and a similar one in post-down), or is there a better way?

I'm not sure it would help.
One trick might to to keep the NDP entry alive by constantly pinging it via a 
cron entry.

Rpy