RE: ipv6 stateless autoconfiguration failing
> George Manousakis a écrit :
> > If I recall correctly though a trace on the Broadcom interface
> indicated
> > that RA's were received from the router.
>
> In my experience, the kernel logs a message ("<interface>: no IPv6
> routers present") when stateless autoconfiguration is enabled and no
> RA's are received, but is silent when RA's are received.
I have noticed that as well. Too bad there is no way to confirm somehow the
expected behavior.
>
> > Anyway, I did a check triggered by your comment of the prefix
> lifetime.
> >
> > The router was configured with those values:
> > ipv6 nd prefix default 180 120
>
> What are these numbers ? Valid and preferred lifetimes ?
Yes, valid = 180 sec , pref = 120 sec
>
> > ipv6 nd ra interval 10
> >
> > and that's because I use a 6to4 tunnel to "simulate" my ipv6
> connectivity
> > and I have a dynamic ipv4 from my isp. So I want the information of
> the
> > change of the ip to be propagated ASAP.
> >
> > When I removed those lines from the router the debian host took the
> ipv6
> > address and hold it.
>
> What do you mean exactly ? Did the Debian host take a global IPv6
> address without rebooting, after it initially took and lost it ?
No, without rebooting nothing is refreshed. Not ever when the interface is
down and up.
The sequence was that I removed those lines from my cisco router and
rebooted the debian. When it was on, it got the global ipv6 correctly and
till now (10 hours later) it's still there. But it's not the correct one
anymore. The dynamic ip of me isp interconnection changed and so the /48
subnet that I could use via 6to4 tunneling.
Router still sends the RA's with the correct prefix but debian just ignores
them!
>
> > Then I reinserted the commands in the router. Debian did nothing.
>
> Again, what do you mean exactly ?
>
> > But when I caused a renewal of my public ipv4 the router found out
> about the
> > change and reconfigured the general prefix but it seems that debian
> is
> > ignoring the new RA's.
>
> After or without rebooting ?
Without rebooting.
> It seems to have stuck with the old global ipv6 and
> > of course no ipv6 destination is reachable.
> >
> > I still cannot find any information on the debian regarding the
> timers or
> > logs of the autoconfigured ipv6.
>
> Stateless autoconfiguration is done entirely by the kernel (except the
> processing of the RDNSS option which requires some userland), so it is
> not Debian specific. It is controlled by a set of per-interface sysctl
> parameters, in net.ipv6.conf.<interface>, see
> Documentation/ip-sysctl.txt for information.
>
> > I don't know if there has anything to do but I am mentioning about
> this
> > since it's an ipv6 operation: on my debian host I run quagga with
> zebra,
> > bgpd, ospfd and ospf6d
>
> Is IPv6 forwarding enabled ? If so, this disables stateless
> autoconfiguration. (And by the way IPv6-wise the box is a router, not a
> host.)
Well, maybe I should answer only this one.
First of all you are correct about the routing functionality of the box.
Since I have enabled ospfv3 you may call it a router even if I have only one
interface enabled.
Let's go on:
linux:~# cat /proc/sys/net/ipv6/conf/all/forwarding
1
And then I changed the value to "0"
The interface took immediately the new ipv6 address. Ok, it hasn't released
the old one but at least it took the new one and was able to ping
ipv6.google.com again.
It's just odd because I remembered checking that value previously and find
it "0". I didn't change it because the box only has one ipv6 enabled
interface so ipv6 forwarding is not an issue. I have no idea how the value
was set to "1" and of course I had no idea that by setting this value to "1"
autoconfiguration is disabled.
Thank you for your help!
>
> --
> To UNSUBSCRIBE, email to debian-ipv6-REQUEST@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact
> listmaster@lists.debian.org
> Archive: 4C66569D.5080200@plouf.fr.eu.org">http://lists.debian.org/4C66569D.5080200@plouf.fr.eu.org
Reply to: