Re: IPv6 networking problems in 14.3

Chris Ross <[email protected]>
Newsgroups gmane.os.freebsd.devel.net
Message-ID <[email protected]>

> On Sep 20, 2025, at 10:58, Tom Pusateri <[email protected]> wrote:
> 
> Have you considered simplifying everything to the most basic setup first. It’s hard to keep up with what you’re doing. If it were me, I would use a different box with a new install of a release version of 14.3 and test a direct connection to the ONT (no switch, just a cable). Knowing whether that works will dictate your next steps. Surely you can find an old piece of hardware to put a clean install on that you can locate right next to the ONT. If it doesn’t work, then try a new install of the release version of 14.2. If that doesn’t work, try 14.1. Get a standard and simple setup working and then work your way backwards.
> 
> Simplifying is key. Removing as many unknowns is key.

Thank you.  That’s a good suggestion.  I do have a tendency both to
make things to complicated, and also to change as few variables as
possible.  But, there are too many variables here already, I now see.

I don’t know that removing the switches is possible, but everything else
should be.  I’ll try to prep a system for that so that I can experiment
when I am able to next deny internet to the house.  ;-)

- Chris

>> On Sep 20, 2025, at 10:39 AM, Chris Ross <[email protected]> wrote:
>> 
>> 
>> 
>>> On Sep 18, 2025, at 16:36, Chris Ross <[email protected]> wrote:
>>> 
>>>> On 17 Sep 2025, at 14:20, Chris Ross <[email protected]> wrote:
>>>> So, on the idea of trying to back-date the whole machine, I have ZFS
>>>> snapshots of the whole root from just before the first upgrade,
>>>> Aug 7/8.  […]
>>> 
>>> [...]  I’ve set bootfs to the older 14.1 system and gotten that running.
>>> 
>>> It [...] out SOLICIT6 messages and [..] if I waited long enough (2ish
>>> hours in my case), it eventually did get an answer!
>> 
>> And, this is where things went odd again.  The above attempt in older
>> filesystem, I had during boot menu chosen my ROUTER custom kernel.  Just
>> a config I traditionally use that has fewer devices in it occupying memory.
>> So, thinking it shouldn’t make a difference I rebooted into the default,
>> GENERIC, kernel on the system.  The default kernel had problems [1], so I
>> chose the boot/kernel.generic on the fs.
>> 
>> Sadly, it did not work.  I get the same state where the router LL is
>> unroutable, ndp -an shows no MAC for it, and I have no IPv6.  So,
>> despite thinking the kernel config wasn’t key detail, I booted back
>> into kernel.router.  Now I have no IPv6 there either.  I tried just
>> shutting down dhcpcd overnight to see if it ISP was confused, but no help.
>> 
>> So, it certainly could be the ISP, but the fact that I’m getting RA’s
>> (and dhcp6 responses) which dhcpcd is processing, but then the MAC for
>> the routers LL isn’t available to the system, makes me think it’s _not_
>> the ISP.  I can’t imagine why a working rootfs and kernel could
>> show a system level problem intermittently, though.
>> 
>> Thoughts again invited.  Not able to reproduce success now, I don’t
>> know that I can actually _test_ anything, but I’d love to hear thoughts
>> about any options/possibilities.
>> 
>>                  - Chris
>> 
>> 
>> 
>> 
>> [1] Sadly there is something wrong with /boot/kernel in that fs.  When
>> I boot it I get a number of:
>> 
>> KLD uhid.ko: depends on kernel - not available or version mismatch
>> KLD ums.ko: depends on kernel - not available or version mismatch
>> KLD usbhid.ko: depends on kernel - not available or version mismatch
>> KLD uhid.ko: depends on kernel - not available or version mismatch
>> 
>> errors late in boot.  That kernel is 14.1-RELEASE-p7 (despite the fs
>> freebsd-update made named zroot/ROOT/14.1-RELEASE-p8_2025-08-08_074410).
>> The rest of the kernels I have on the fs are 14.1p5. something must be
>> off with that one.  So I booted /boot/kernel.generic, 14.1p5, and it
>> boots fine.
>> 
>> Current output log of dhcpcd: note "fe80::3e8a:b0ff:fe3e:4dce is unreachable”.
>> That’s a kernel routing issue, and I think the primary problem.
>> 
>> Sep 20 10:20:32 [30273]: vlan0: REPLY6 received from fe80::3e8a:b0ff:fe3e:4dce
>> Sep 20 10:20:32 [30273]: vlan0: renew in 3600, rebind in 5760, expire in 7200 seconds
>> Sep 20 10:20:32 [30273]: lo0: adding reject route to 2600:4040:2c9c:2d00::/56 via ::1
>> Sep 20 10:20:32 [30273]: vlan0: writing lease: /var/db/dhcpcd/vlan0.lease6
>> Sep 20 10:20:32 [30273]: vlan0: delegated prefix 2600:4040:2c9c:2d00::/56
>> Sep 20 10:20:32 [30273]: vlan0: executing: /usr/local/libexec/dhcpcd-run-hooks BOUND6
>> Sep 20 10:29:08 [30273]: vlan0: Router Advertisement from fe80::3e8a:b0ff:fe3e:4dce
>> Sep 20 10:29:08 [30273]: vlan0: no global addresses for default route
>> Sep 20 10:29:08 [30273]: vlan0: executing: /usr/local/libexec/dhcpcd-run-hooks ROUTERADVERT
>> Sep 20 10:30:15 [30273]: vlan0: fe80::3e8a:b0ff:fe3e:4dce is unreachable
>> Sep 20 10:30:15 [30273]: vlan0: delaying IPv6 router solicitation for 0.7 seconds
>> Sep 20 10:30:16 [30273]: vlan0: soliciting an IPv6 router
>> Sep 20 10:30:16 [30273]: vlan0: sending Router Solicitation
>> Sep 20 10:30:16 [30273]: vlan0: Router Advertisement from fe80::3e8a:b0ff:fe3e:4dce
>> Sep 20 10:30:16 [30273]: vlan0: executing: /usr/local/libexec/dhcpcd-run-hooks ROUTERADVERT
>> 
>> 
>> 
>
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.