Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: nsp: ipv6

How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix

 

 

nsp ipv6 RSS feed   Index | Next | Previous | View Threaded


msmith at internode

Apr 6, 2011, 11:04 PM

Post #1 of 18 (6119 views)
Permalink
How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix

Hi,

A number of our customers are having trouble with IPv6 under Windows
Vista and 7 on our IPv6 trial.

Currently we are giving customers a static delegated prefix via
DHCPv6-PD (a /60 currently). When they have an ADSL drop out, some IPv6
CPE is deprecating the assigned LAN interface global /64 prefix by
sending an RA with a zero second preferred lifetime for the prefix.
Windows Vista and 7 are quite correctly labelling the prefix as
deprecated, as it still has a non-zero valid lifetime. Deprecating the
prefix at this time makes sense as there is a possibility that the
delegated prefix may not be re-assigned to the CPE after the ADSL link
is restored, and may be re-assigned by the ISP to another customer. This
is also the behavior specified in the recently published RFC6204, "Basic
Requirements for IPv6 Customer Edge Routers".

When the ADSL link comes up again, and the CPE re-acquires the same
delegated prefix, it re-announces the same /64 onto the LAN with a
non-zero preferred lifetime. Windows Vista and 7 receive and update the
non-zero prefix lifetimes for the addresses, but do not change the
addresses back to preferred. Windows XP does, as do other operating
systems. Windows Vista and 7 not setting the addresses back to preferred
seems to then cause IPv6 connectivity to then fail, causing Vista and 7
to revert back to using IPv4 only. Windows Vista and 7 have no trouble
at this time with new prefixes, so the issue seems to be specifically
related to changing a previously deprecated prefix back to a preferred one.

Does anybody know how I can report this bug with Microsoft on behalf of
our customers?

(If anybody wants to have a go reproducing this, the latest CVS version
of radvd now has a "DeprecatePrefix" option for prefixes. To view the
IPv6 address status and lifetimes under Windows, use the "netsh
interface ipv6 show address" "cmd" command.)

Thanks,
Mark.


swmike at swm

Apr 6, 2011, 11:44 PM

Post #2 of 18 (5946 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

On Thu, 7 Apr 2011, Mark Smith wrote:

> Does anybody know how I can report this bug with Microsoft on behalf of
> our customers?

There are a few @microsoft.com people on the IETF v6ops list that have
been discussing 6to4 there the last few days. I would recommend to contact
them, they seem to have influence.

I have tried to report IPv6 behaviour through normal Microsoft channels
and failed miserably, I wouldn't recommend the experience to my worst
enemy. Very frustrating.

--
Mikael Abrahamsson email: swmike [at] swm


tore.anderson at redpill-linpro

Apr 6, 2011, 11:46 PM

Post #3 of 18 (5956 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

Hi,

* Mark Smith

> Does anybody know how I can report this bug with Microsoft on behalf of
> our customers?

Get in touch with Chris Palmer <Christopher.Palmer [at] microsoft>, he's
the IPv6 Program Manager for MS Windows.

That said, I believe this particular bug has been reported to him before
by Johannes Endres. But I guess it doesn't hurt to ask for a fresh
status update.

Best regards,
--
Tore Anderson
Redpill Linpro AS - http://www.redpill-linpro.com
Tel: +47 21 54 41 27


otroan at employees

Apr 6, 2011, 11:53 PM

Post #4 of 18 (5956 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

Mark,

> A number of our customers are having trouble with IPv6 under Windows Vista and 7 on our IPv6 trial.
>
> Currently we are giving customers a static delegated prefix via DHCPv6-PD (a /60 currently). When they have an ADSL drop out, some IPv6 CPE is deprecating the assigned LAN interface global /64 prefix by sending an RA with a zero second preferred lifetime for the prefix. Windows Vista and 7 are quite correctly labelling the prefix as deprecated, as it still has a non-zero valid lifetime. Deprecating the prefix at this time makes sense as there is a possibility that the delegated prefix may not be re-assigned to the CPE after the ADSL link is restored, and may be re-assigned by the ISP to another customer. This is also the behavior specified in the recently published RFC6204, "Basic Requirements for IPv6 Customer Edge Routers".

no, it isn't.
this behaviour breaks local communication within the home.

> When the ADSL link comes up again, and the CPE re-acquires the same delegated prefix, it re-announces the same /64 onto the LAN with a non-zero preferred lifetime. Windows Vista and 7 receive and update the non-zero prefix lifetimes for the addresses, but do not change the addresses back to preferred. Windows XP does, as do other operating systems. Windows Vista and 7 not setting the addresses back to preferred seems to then cause IPv6 connectivity to then fail, causing Vista and 7 to revert back to using IPv4 only. Windows Vista and 7 have no trouble at this time with new prefixes, so the issue seems to be specifically related to changing a previously deprecated prefix back to a preferred one.
>
> Does anybody know how I can report this bug with Microsoft on behalf of our customers?
>
> (If anybody wants to have a go reproducing this, the latest CVS version of radvd now has a "DeprecatePrefix" option for prefixes. To view the IPv6 address status and lifetimes under Windows, use the "netsh interface ipv6 show address" "cmd" command.)

cheers,
Ole


hahn at berkom

Apr 7, 2011, 12:20 AM

Post #5 of 18 (5944 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Mark,

> When the ADSL link comes up again, and the CPE re-acquires the same delegated
> prefix, it re-announces the same /64 onto the LAN with a non-zero preferred
> lifetime. Windows Vista and 7 receive and update the non-zero prefix lifetimes
> for the addresses, but do not change the addresses back to preferred. Windows XP
> does, as do other operating systems. Windows Vista and 7 not setting the
> addresses back to preferred seems to then cause IPv6 connectivity to then fail,
> causing Vista and 7 to revert back to using IPv4 only. Windows Vista and 7 have
> no trouble at this time with new prefixes, so the issue seems to be specifically
> related to changing a previously deprecated prefix back to a preferred one.
We discovered this bug during some lab testing by coincidence back in 2009,
please see conversation on this list in the archives starting with message:
http://lists.cluenet.de/pipermail/ipv6-ops/2009-December/002718.html

I was privately contacted by Sean Siler of Microsoft (who should be still on the
list), but after some mail exchanges there was nothing but silence :(
We tried the "official" way of bug reporting but where discouraged by the
procedures which included also money.
In the end I gave up since this isn't my main business. I last checked if the
bug exists in December 2010 and now you give me the confirmation that it still
exists.

Some side informations:
1) We discovered that the generated privacy addresses will not be renewed in the
described scenario.
2) You can reproduce this behavior without sending PIO with PrefLT=0 if you send
the Vista/Win7 system in standby and wake it up after time X.
Where X is defined as ValidLifetime > X > PreferredLifetime.

good luck,
Christian

>
> Does anybody know how I can report this bug with Microsoft on behalf of our
> customers?
>
> (If anybody wants to have a go reproducing this, the latest CVS version of radvd
> now has a "DeprecatePrefix" option for prefixes. To view the IPv6 address status
> and lifetimes under Windows, use the "netsh interface ipv6 show address" "cmd"
> command.)
>
> Thanks,
> Mark.
>
>
>
>
>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.15 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/

iEYEARECAAYFAk2dZakACgkQ6kMW7HW86235NACglY2chaVGjTC9QgAabzCA9j4D
LbUAniSD1uPRChEWE25ePqVWQtFvW4GX
=fmDK
-----END PGP SIGNATURE-----


msmith at internode

Apr 7, 2011, 12:50 AM

Post #6 of 18 (5943 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

Hi Ole,

On 7/04/2011 4:23 PM, Ole Troan wrote:
> Mark,
>
>> A number of our customers are having trouble with IPv6 under Windows Vista and 7 on our IPv6 trial.
>>
>> Currently we are giving customers a static delegated prefix via DHCPv6-PD (a /60 currently). When they have an ADSL drop out, some IPv6 CPE is deprecating the assigned LAN interface global /64 prefix by sending an RA with a zero second preferred lifetime for the prefix. Windows Vista and 7 are quite correctly labelling the prefix as deprecated, as it still has a non-zero valid lifetime. Deprecating the prefix at this time makes sense as there is a possibility that the delegated prefix may not be re-assigned to the CPE after the ADSL link is restored, and may be re-assigned by the ISP to another customer. This is also the behavior specified in the recently published RFC6204, "Basic Requirements for IPv6 Customer Edge Routers".
>
> no, it isn't.
> this behaviour breaks local communication within the home.
>

Hmm. Well that's the behavior this CPE seems to be exhibiting. I think
I've been fooled a bit by misremembering this clause from the RFC -


L-13: If the delegated prefix changes, i.e., the current prefix is
replaced with a new prefix without any overlapping time
period, then the IPv6 CE router MUST immediately advertise the
old prefix with a Preferred Lifetime of zero and a Valid
Lifetime of the lower of the current Valid Lifetime and 2
hours (which must be decremented in real time) in a Router
Advertisement message as described in Section 5.5.3, (e) of
[RFC4862].


However, I don't think that means that Vista and Windows 7 doesn't have
a bug though. Surely it is valid for a prefix to be "undeprecated"?


>> When the ADSL link comes up again, and the CPE re-acquires the same delegated prefix, it re-announces the same /64 onto the LAN with a non-zero preferred lifetime. Windows Vista and 7 receive and update the non-zero prefix lifetimes for the addresses, but do not change the addresses back to preferred. Windows XP does, as do other operating systems. Windows Vista and 7 not setting the addresses back to preferred seems to then cause IPv6 connectivity to then fail, causing Vista and 7 to revert back to using IPv4 only. Windows Vista and 7 have no trouble at this time with new prefixes, so the issue seems to be specifically related to changing a previously deprecated prefix back to a preferred one.
>>
>> Does anybody know how I can report this bug with Microsoft on behalf of our customers?
>>
>> (If anybody wants to have a go reproducing this, the latest CVS version of radvd now has a "DeprecatePrefix" option for prefixes. To view the IPv6 address status and lifetimes under Windows, use the "netsh interface ipv6 show address" "cmd" command.)
>
> cheers,
> Ole
>

Thanks,
Mark.


otroan at employees

Apr 7, 2011, 12:59 AM

Post #7 of 18 (5942 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

Mark,

>>> A number of our customers are having trouble with IPv6 under Windows Vista and 7 on our IPv6 trial.
>>>
>>> Currently we are giving customers a static delegated prefix via DHCPv6-PD (a /60 currently). When they have an ADSL drop out, some IPv6 CPE is deprecating the assigned LAN interface global /64 prefix by sending an RA with a zero second preferred lifetime for the prefix. Windows Vista and 7 are quite correctly labelling the prefix as deprecated, as it still has a non-zero valid lifetime. Deprecating the prefix at this time makes sense as there is a possibility that the delegated prefix may not be re-assigned to the CPE after the ADSL link is restored, and may be re-assigned by the ISP to another customer. This is also the behavior specified in the recently published RFC6204, "Basic Requirements for IPv6 Customer Edge Routers".
>>
>> no, it isn't.
>> this behaviour breaks local communication within the home.
>>
>
> Hmm. Well that's the behavior this CPE seems to be exhibiting. I think I've been fooled a bit by misremembering this clause from the RFC -

not one of ours? if so let me know.

>
> L-13: If the delegated prefix changes, i.e., the current prefix is
> replaced with a new prefix without any overlapping time
> period, then the IPv6 CE router MUST immediately advertise the
> old prefix with a Preferred Lifetime of zero and a Valid
> Lifetime of the lower of the current Valid Lifetime and 2
> hours (which must be decremented in real time) in a Router
> Advertisement message as described in Section 5.5.3, (e) of
> [RFC4862].
>
>
> However, I don't think that means that Vista and Windows 7 doesn't have a bug though. Surely it is valid for a prefix to be "undeprecated"?

absolutely. didn't mean to create any doubt that that is a bug.

cheers,
Ole


berni at birkenwald

Apr 7, 2011, 7:59 AM

Post #8 of 18 (5921 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

On Thu, Apr 07, 2011 at 03:34:36PM +0930, Mark Smith wrote:

Hello,

> A number of our customers are having trouble with IPv6 under Windows
> Vista and 7 on our IPv6 trial.

I cannot help you with your problem, but I also have some issues with
Windows 7 and IPv6. Maybe someone has an idea how to fix this:

We have an internal network with about 200 clients. Mostly Dell Desktops
from the past three years with different operating systems (various
Linux distributions, Windows XP, Windows 7). A few MacOS boxes as well.
The router is a Cisco 6500 with SXI5, the switching infrastructure are
non-IPv6-aware HP Procurve. We run non-standard timers here (RA
interval 10s, RA lifetime 30s, Prefix valid for 86400s, preferred for
1800s).

Every once in a while a Windows 7 box gets broken IPv6. Symptoms are a
deprecated address and no default route. It pretty much acts like it
does not receive the RAs anymore. But the RAs are there (verified with
Wireshark on the affected client and from other clients which work just
fine). Most of the times the issue can be fixed by disabling/reenabling
the network interface on the client.

Mostly we have seen this issue with clients that have been suspended a
few times, but today I had a client that had just rebooted. Basically
the expiry timers started running when the box was booted up and never
got refreshed.

Anyone got similar experiences?

Bernhard


cholzhauer at sscorp

Apr 7, 2011, 10:02 AM

Post #9 of 18 (5934 views)
Permalink
RE: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

It almost sounds like I'm in the minority.

I'm doing RA from a Cisco ASA running 8.3 to roughly fifty Windows 7 and Windows Vista machines and I've never had a problem.

I've turned off privacy extensions, Teredo, 6to4 and ISATAP.

We are running Vista SP2 and Windows 7 SP1, but never had a problem before applying the service packs.


-----Original Message-----
From: ipv6-ops-bounces+cholzhauer=sscorp.com [at] lists [mailto:ipv6-ops-bounces+cholzhauer=sscorp.com [at] lists] On Behalf Of Bernhard Schmidt
Sent: Thursday, April 07, 2011 11:00 AM
To: ipv6-ops [at] lists
Subject: Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix

On Thu, Apr 07, 2011 at 03:34:36PM +0930, Mark Smith wrote:

Hello,

> A number of our customers are having trouble with IPv6 under Windows
> Vista and 7 on our IPv6 trial.

I cannot help you with your problem, but I also have some issues with Windows 7 and IPv6. Maybe someone has an idea how to fix this:

We have an internal network with about 200 clients. Mostly Dell Desktops from the past three years with different operating systems (various Linux distributions, Windows XP, Windows 7). A few MacOS boxes as well.
The router is a Cisco 6500 with SXI5, the switching infrastructure are non-IPv6-aware HP Procurve. We run non-standard timers here (RA interval 10s, RA lifetime 30s, Prefix valid for 86400s, preferred for 1800s).

Every once in a while a Windows 7 box gets broken IPv6. Symptoms are a deprecated address and no default route. It pretty much acts like it does not receive the RAs anymore. But the RAs are there (verified with Wireshark on the affected client and from other clients which work just fine). Most of the times the issue can be fixed by disabling/reenabling the network interface on the client.

Mostly we have seen this issue with clients that have been suspended a few times, but today I had a client that had just rebooted. Basically the expiry timers started running when the box was booted up and never got refreshed.

Anyone got similar experiences?

Bernhard


berni at birkenwald

Apr 7, 2011, 12:15 PM

Post #10 of 18 (5935 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

On 07.04.2011 19:02, Carl Holzhauer wrote:

Hi,

> It almost sounds like I'm in the minority.
> I'm doing RA from a Cisco ASA running 8.3 to roughly fifty Windows 7 and Windows Vista machines and I've never had a problem.

Or you haven't noticed it :-)

We have been doing native IPv6 in that 200-odd devices network for about
six years. I have only very recently seen the first of those issues, and
I have no idea at all how frequent they are. At least in our case, it
fails completely graceful and just switches over to ISATAP or IPv4-only.
So far I have only seen two users complaining, one who had disabled
ISATAP and could not access IPv6-only resources anymore (we have very
few of those), and one who switched to ISATAP and was blocked by his own
over-eager ACL on a host.

Other than that, I had always wondered about the tiny fraction of ISATAP
from networks with native IPv6, but I had never bothered. I'm going to
have a closer look.

Bernhard


hahn at berkom

Apr 8, 2011, 1:40 AM

Post #11 of 18 (5917 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello Bernhard,

> Every once in a while a Windows 7 box gets broken IPv6. Symptoms are a
> deprecated address and no default route. It pretty much acts like it
> does not receive the RAs anymore. But the RAs are there (verified with
> Wireshark on the affected client and from other clients which work just
> fine). Most of the times the issue can be fixed by disabling/reenabling
> the network interface on the client.
>
> Mostly we have seen this issue with clients that have been suspended a
> few times, but today I had a client that had just rebooted. Basically
> the expiry timers started running when the box was booted up and never
> got refreshed.
I assume it's similar to what we discovered. If you send a client in standby and
wake it up after time X (where X is defined as ValidLifetime > X >
PreferredLifetime), IPv6 is broken.

cheers,
Christian

>
> Anyone got similar experiences?
>
> Bernhard

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.15 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/

iEYEARECAAYFAk2eyeEACgkQ6kMW7HW86200WACglkF9n4mBdGgZLx+8iF7kzB9G
9NUAn3uHT6tjxqTDePHuzn/8WECJGTW1
=M6xT
-----END PGP SIGNATURE-----


msmith at internode

Apr 11, 2011, 8:00 PM

Post #12 of 18 (5877 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

Hi Ole,

On 7/04/2011 5:29 PM, Ole Troan wrote:
> Mark,
>
>>>> A number of our customers are having trouble with IPv6 under Windows Vista and 7 on our IPv6 trial.
>>>>
>>>> Currently we are giving customers a static delegated prefix via DHCPv6-PD (a /60 currently). When they have an ADSL drop out, some IPv6 CPE is deprecating the assigned LAN interface global /64 prefix by sending an RA with a zero second preferred lifetime for the prefix. Windows Vista and 7 are quite correctly labelling the prefix as deprecated, as it still has a non-zero valid lifetime. Deprecating the prefix at this time makes sense as there is a possibility that the delegated prefix may not be re-assigned to the CPE after the ADSL link is restored, and may be re-assigned by the ISP to another customer. This is also the behavior specified in the recently published RFC6204, "Basic Requirements for IPv6 Customer Edge Routers".
>>>
>>> no, it isn't.
>>> this behaviour breaks local communication within the home.
>>>
>>
>> Hmm. Well that's the behavior this CPE seems to be exhibiting. I think I've been fooled a bit by misremembering this clause from the RFC -
>
> not one of ours? if so let me know.
>

No it isn't one of yours.

I've just had a closer look at what it is doing on it's LAN interface -

1. While the ADSL link is up, it advertises both it's global prefix and
a "ULA" prefix in it's LAN interface RAs. The global prefix preferred
lifetime is 3600, while the valid lifetime is 7200, regardless of what
we are setting on the delegated prefix. The "ULA" prefix is announced
with a 0 second preferred and valid lifetimes.

(I'm putting "ULA" in quotes because the random ID component is all
zeros i.e. they've recreated the site-local problems in the ULA space.)

2. When the ADSL link fails, the prefix lifetime values are reversed -
the global gets all zeros, while the "ULA"'s become 3600/7200.

3. When the ADSL link recovers, the prefix lifetime values are reversed
again. This is where the Windows Vista and 7 bug bites when the same
global prefix is re-announced with non-zero lifetimes.


>>
>> L-13: If the delegated prefix changes, i.e., the current prefix is
>> replaced with a new prefix without any overlapping time
>> period, then the IPv6 CE router MUST immediately advertise the
>> old prefix with a Preferred Lifetime of zero and a Valid
>> Lifetime of the lower of the current Valid Lifetime and 2
>> hours (which must be decremented in real time) in a Router
>> Advertisement message as described in Section 5.5.3, (e) of
>> [RFC4862].
>>
>>
>> However, I don't think that means that Vista and Windows 7 doesn't have a bug though. Surely it is valid for a prefix to be "undeprecated"?
>
> absolutely. didn't mean to create any doubt that that is a bug.
>
> cheers,
> Ole
>

Regards,
Mark.


tore.anderson at redpill-linpro

Apr 12, 2011, 12:45 AM

Post #13 of 18 (5860 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

Hi Mark,

* Mark Smith

> No it isn't one of yours.

AVM FRITZ!Box?

http://getipv6.info/index.php/Customer_problems_that_could_occur#AVM

> I've just had a closer look at what it is doing on it's LAN interface -
>
> 1. While the ADSL link is up, it advertises both it's global prefix and
> a "ULA" prefix in it's LAN interface RAs. The global prefix preferred
> lifetime is 3600, while the valid lifetime is 7200, regardless of what
> we are setting on the delegated prefix. The "ULA" prefix is announced
> with a 0 second preferred and valid lifetimes.

What happens if you connect the WAN port to an IPv4-only network; ULA
PIO with normal lifetimes? What about the RA lifetime in this case?

> 2. When the ADSL link fails, the prefix lifetime values are reversed -
> the global gets all zeros, while the "ULA"'s become 3600/7200.

I'm curious to hear if something happens with the lifetime of the RA
itself in this case. Does it get set to 0 so that the hosts on the LAN
remove the default route?

What happens if only the IPv6 part of the WAN connection fails (e.g. if
a DHCPv6 lease expires)?

The reason I'm asking is that this device might cause end-user
brokenness if it can be made to announce a ULA prefix with a router
lifetime of >0 that is *not* accompanied by a global prefix. This would
cause end-user hosts to attempt to use the ULA addresses when connecting
to dual-stacked destinations, as ULAs have global scope and are not
assigned a distinct label in the RFC3484 policy table in any recent
operating system. If that is the case I would very much like to document
this device as a source of brokenness in the ARIN Wiki.

Best regards,
--
Tore Anderson
Redpill Linpro AS - http://www.redpill-linpro.com/
Tel: +47 21 54 41 27


dr at cluenet

Apr 14, 2011, 11:54 AM

Post #14 of 18 (5811 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

On Tue, Apr 12, 2011 at 12:30:17PM +0930, Mark Smith wrote:
> 3. When the ADSL link recovers, the prefix lifetime values are reversed
> again. This is where the Windows Vista and 7 bug bites when the same global
> prefix is re-announced with non-zero lifetimes.

Arghhh... I've spent half an hour today trying to debug a Vista PC not
using IPv6 preferred to IPv4... all IPv6 addresses it had were
deprecated, although the RAs from the CPE router indeed had proper
lifetimes. And yes, I played with the IPv6 uplink of the CPE before - so
actually ran into above problem too.

Microsoft, are you listening? Pretty pretty please fix that ASAP! :-)

Best regards,
Daniel

--
CLUE-RIPE -- Jabber: dr [at] cluenet -- dr [at] IRCne -- PGP: 0xA85C8AA0


msmith at internode

Apr 14, 2011, 7:01 PM

Post #15 of 18 (5827 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

Hi Tore,

On 12/04/2011 5:15 PM, Tore Anderson wrote:
> Hi Mark,
>
> * Mark Smith
>
>> No it isn't one of yours.
>
> AVM FRITZ!Box?
>

Yes it is, it's a 7390 running firmware 84.04.88, which according to the
device web gui is the latest available production firmware.

> http://getipv6.info/index.php/Customer_problems_that_could_occur#AVM
>
>> I've just had a closer look at what it is doing on it's LAN interface -
>>
>> 1. While the ADSL link is up, it advertises both it's global prefix and
>> a "ULA" prefix in it's LAN interface RAs. The global prefix preferred
>> lifetime is 3600, while the valid lifetime is 7200, regardless of what
>> we are setting on the delegated prefix. The "ULA" prefix is announced
>> with a 0 second preferred and valid lifetimes.
>
> What happens if you connect the WAN port to an IPv4-only network; ULA
> PIO with normal lifetimes? What about the RA lifetime in this case?
>

In this situation, with the IPv6 recommended defaults ("IPv6
Connectivity - Detect settings automatically (recommended)", "Assign
unique local addresses (ULA) as long as no IPv6 connection exists
(recommended)"), the 7390 brings up a 6to4 tunnel, and then announces
the 2002:..XX../64 prefix onto the LAN in the same manner as before,
swapping the ULA for the 6to4 prefix when the ADSL interface goes down
and vice versa. The router lifetime stays at 1800 seconds. With the ADSL
link down, a ULA and end-nodes with a default gateway, the 7390 should
be generating ICMPv6 destination unreachables, but it doesn't. When I've
had a router that does generate destination unreachables in this
situation, my linux end-node immediately switches to trying to use IPv4
and then successfully accesses the IPv4 version of the website.


>> 2. When the ADSL link fails, the prefix lifetime values are reversed -
>> the global gets all zeros, while the "ULA"'s become 3600/7200.
>
> I'm curious to hear if something happens with the lifetime of the RA
> itself in this case. Does it get set to 0 so that the hosts on the LAN
> remove the default route?
>

No the default route doesn't get withdrawn by the router.

> What happens if only the IPv6 part of the WAN connection fails (e.g. if
> a DHCPv6 lease expires)?
>

We're using SLAAC on the WAN side. I'll have a think to see if I can
work out a way to simulate an IPv6 WAN failure to see what happens.

> The reason I'm asking is that this device might cause end-user
> brokenness if it can be made to announce a ULA prefix with a router
> lifetime of>0 that is *not* accompanied by a global prefix. This would
> cause end-user hosts to attempt to use the ULA addresses when connecting
> to dual-stacked destinations, as ULAs have global scope and are not
> assigned a distinct label in the RFC3484 policy table in any recent
> operating system. If that is the case I would very much like to document
> this device as a source of brokenness in the ARIN Wiki.
>

Regards,
Mark.


dr at cluenet

Jun 16, 2011, 8:12 AM

Post #16 of 18 (5405 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

On Fri, Apr 08, 2011 at 10:40:02AM +0200, Christian Hahn wrote:
> I assume it's similar to what we discovered. If you send a client in standby and
> wake it up after time X (where X is defined as ValidLifetime > X >
> PreferredLifetime), IPv6 is broken.

Seeing that here almost daily with my IPv6 testing netbook on my desk...
Very annoying.

Could Tore add that to
http://getipv6.info/index.php/Customer_problems_that_could_occur#Microsoft_Windows_2
please, so it gets onto the "bigger radar" of problems?

Best regards,
Daniel

--
CLUE-RIPE -- Jabber: dr [at] cluenet -- dr [at] IRCne -- PGP: 0xA85C8AA0


tore.anderson at redpill-linpro

Jun 28, 2011, 5:58 AM

Post #17 of 18 (5260 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

* Daniel Roesen

> Could Tore add that to
> http://getipv6.info/index.php/Customer_problems_that_could_occur#Microsoft_Windows_2
> please, so it gets onto the "bigger radar" of problems?

Okay, done.

(It is a Wiki, though, so do feel free to add stuff yourself too...)

--
Tore Anderson
Redpill Linpro AS - http://www.redpill-linpro.com
Tel: +47 21 54 41 27


dr at cluenet

Jun 28, 2011, 9:17 AM

Post #18 of 18 (5254 views)
Permalink
Re: How to report IPv6 bug to Microsoft - Vista and 7 won't "undeprecate" a prefix [In reply to]

On Tue, Jun 28, 2011 at 02:58:57PM +0200, Tore Anderson wrote:
> > Could Tore add that to
> > http://getipv6.info/index.php/Customer_problems_that_could_occur#Microsoft_Windows_2
> > please, so it gets onto the "bigger radar" of problems?
>
> Okay, done.

Thanks.

This issue is now being taken care of:
http://forums.whirlpool.net.au/forum-replies.cfm?t=1648260&p=28#r542

Word is that Vista is queued to get fixed, too.

Best regards,
Daniel

--
CLUE-RIPE -- Jabber: dr [at] cluenet -- dr [at] IRCne -- PGP: 0xA85C8AA0

nsp ipv6 RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.