Mailing List Archive

Yesterday's Windows update causes IPv4 to be default
Hi

I looks like yesterday's batch of updates causes Windows 7 to prefer
IPv4 over IPv6.
After (automatically) installing the whole lot, one of my users
complained about connection issues with WinSCP to our web server.
This runs SSH on IPv6-only. His WinSCP for some reason tried to
connect over IPv4, without falling back to IPv6.
I tried "ping www.terena.org" and that also used IPv4.
"ping -6 www.terena.org" did also work, but I'm pretty sure it used to
default to IPv6.
After clearing the cache in WinSCP things started to work again.

I didn't install the updates (about 20, including the Office ones) on
my own computer, so I had a look at them first.
One of them is called "No network connectivity on Windows 7-based or
Windows Server 2008 R2-based client computers when a DHCPv6 message is
sent that has a duplicated DUID",
http://support.microsoft.com/kb/2763523, which is the only one that
mentions IPv6.

Before installing anything, I tried to "ping www.terena.org" and that
used IPv6 all-right.
I then installed all the updates except the suspicious
http://support.microsoft.com/kb/2763523.
After rebooting, "ping www.terena.org" uses IPv4...

So I guess one of the other updates caused this; I'm now booting up a
VM to investigate which one exactly.
So far the no other major stuff broke, and it looks like only the
default has changed, but I'd thought I mentioned it here anyway, just
in case...

Anyone else has seen IPv6 issues on Windows since yesterday?



--
Dick Visser
System & Networking Engineer
TERENA Secretariat
Singel 468 D, 1017 AW Amsterdam
The Netherlands
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
http://support.microsoft.com/kb/2750841

/Assume that a computer is configured to use an IPv6 connection as the default connection.
Additionally, assume that the computer does not have a connection to an IPv6 network. In
this situation, it takes a long time for the computer to connect to an IPv6 site. //
//
//This issue occurs because Windows tries the IPv6 connection first. After the connection
fails because of a time-out error, Windows tries the IPv4 connection.//
//
//After this update is installed, Windows uses the NCSI functionality to examine the Ipv6
connection. If the connection is broken, Windows uses IPv4 instead of IPv6./

--
Tassos

Dick Visser wrote on 14/11/2012 18:46:
> Hi
>
> I looks like yesterday's batch of updates causes Windows 7 to prefer
> IPv4 over IPv6.
> After (automatically) installing the whole lot, one of my users
> complained about connection issues with WinSCP to our web server.
> This runs SSH on IPv6-only. His WinSCP for some reason tried to
> connect over IPv4, without falling back to IPv6.
> I tried "ping www.terena.org" and that also used IPv4.
> "ping -6 www.terena.org" did also work, but I'm pretty sure it used to
> default to IPv6.
> After clearing the cache in WinSCP things started to work again.
>
> I didn't install the updates (about 20, including the Office ones) on
> my own computer, so I had a look at them first.
> One of them is called "No network connectivity on Windows 7-based or
> Windows Server 2008 R2-based client computers when a DHCPv6 message is
> sent that has a duplicated DUID",
> http://support.microsoft.com/kb/2763523, which is the only one that
> mentions IPv6.
>
> Before installing anything, I tried to "ping www.terena.org" and that
> used IPv6 all-right.
> I then installed all the updates except the suspicious
> http://support.microsoft.com/kb/2763523.
> After rebooting, "ping www.terena.org" uses IPv4...
>
> So I guess one of the other updates caused this; I'm now booting up a
> VM to investigate which one exactly.
> So far the no other major stuff broke, and it looks like only the
> default has changed, but I'd thought I mentioned it here anyway, just
> in case...
>
> Anyone else has seen IPv6 issues on Windows since yesterday?
>
>
>
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On Wed, 14 Nov 2012, Dick Visser wrote:

> Anyone else has seen IPv6 issues on Windows since yesterday?

"An IPv6 readiness update is available for Windows 7 and for Windows
Server 2008 R2".

http://support.microsoft.com/kb/2750841

It seems to address 6to4 mainly, but it has some "broken IPv6
connectivity" as well.

Anyhow, after installing it on my 64bit Win7 Home Premium, I don't see any
of your issues. As far as I can tell, it thinks my IPv6 works fine and
uses it just like before.

I have a /48 tunneled to my home out of our regular PA IPv6 block, and I
set IPv6 MTU to 1400 on my home lan (to avoid PMTUD RTT delays).

--
Mikael Abrahamsson email: swmike@swm.pp.se
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On 14 November 2012 18:18, Mikael Abrahamsson <swmike@swm.pp.se> wrote:

> http://support.microsoft.com/kb/2750841
>
> It seems to address 6to4 mainly, but it has some "broken IPv6 connectivity"
> as well.

I guess this all means that my network must be broken somehow...
Weird thing is that we never had any problems until this update came along ;-)

Will do some more digging in the mean time.


--
Dick Visser
System & Networking Engineer
TERENA Secretariat
Singel 468 D, 1017 AW Amsterdam
The Netherlands
RE: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
> -----Original Message-----
> From: ipv6-ops-bounces+dwing=cisco.com@lists.cluenet.de [mailto:ipv6-
> ops-bounces+dwing=cisco.com@lists.cluenet.de] On Behalf Of Tassos
> Chatzithomaoglou
> Sent: Wednesday, November 14, 2012 9:00 AM
> To: ipv6-ops@lists.cluenet.de
> Subject: Re: Yesterday's Windows update causes IPv4 to be default
>
> http://support.microsoft.com/kb/2750841
>
> /Assume that a computer is configured to use an IPv6 connection as the
> default connection.
> Additionally, assume that the computer does not have a connection to an
> IPv6 network. In this situation, it takes a long time for the computer
> to connect to an IPv6 site. // // //This issue occurs because Windows
> tries the IPv6 connection first. After the connection fails because of a
> time-out error, Windows tries the IPv4 connection.// // //After this
> update is installed, Windows uses the NCSI functionality to examine the
> Ipv6 connection. If the connection is broken, Windows uses IPv4 instead
> of IPv6./

Yep, Windows 8 has been doing that since Windows 8 shipped. Full
details are in my Cisco Live slide deck (I'll have to find a pointer)
and a pretty good summary is
http://blogs.msdn.com/b/b8/archive/2012/06/05/connecting-with-ipv6-in-windows-8.aspx

-d


> --
> Tassos
>
> Dick Visser wrote on 14/11/2012 18:46:
> > Hi
> >
> > I looks like yesterday's batch of updates causes Windows 7 to prefer
> > IPv4 over IPv6.
> > After (automatically) installing the whole lot, one of my users
> > complained about connection issues with WinSCP to our web server.
> > This runs SSH on IPv6-only. His WinSCP for some reason tried to
> > connect over IPv4, without falling back to IPv6.
> > I tried "ping www.terena.org" and that also used IPv4.
> > "ping -6 www.terena.org" did also work, but I'm pretty sure it used to
> > default to IPv6.
> > After clearing the cache in WinSCP things started to work again.
> >
> > I didn't install the updates (about 20, including the Office ones) on
> > my own computer, so I had a look at them first.
> > One of them is called "No network connectivity on Windows 7-based or
> > Windows Server 2008 R2-based client computers when a DHCPv6 message is
> > sent that has a duplicated DUID",
> > http://support.microsoft.com/kb/2763523, which is the only one that
> > mentions IPv6.
> >
> > Before installing anything, I tried to "ping www.terena.org" and that
> > used IPv6 all-right.
> > I then installed all the updates except the suspicious
> > http://support.microsoft.com/kb/2763523.
> > After rebooting, "ping www.terena.org" uses IPv4...
> >
> > So I guess one of the other updates caused this; I'm now booting up a
> > VM to investigate which one exactly.
> > So far the no other major stuff broke, and it looks like only the
> > default has changed, but I'd thought I mentioned it here anyway, just
> > in case...
> >
> > Anyone else has seen IPv6 issues on Windows since yesterday?
> >
> >
> >
RE: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
> -----Original Message-----
> From: ipv6-ops-bounces+dwing=cisco.com@lists.cluenet.de [mailto:ipv6-
> ops-bounces+dwing=cisco.com@lists.cluenet.de] On Behalf Of Dick Visser
> Sent: Wednesday, November 14, 2012 9:26 AM
> To: Mikael Abrahamsson
> Cc: ipv6-ops@lists.cluenet.de
> Subject: Re: Yesterday's Windows update causes IPv4 to be default
>
> On 14 November 2012 18:18, Mikael Abrahamsson <swmike@swm.pp.se> wrote:
>
> > http://support.microsoft.com/kb/2750841
> >
> > It seems to address 6to4 mainly, but it has some "broken IPv6
> connectivity"
> > as well.
>
> I guess this all means that my network must be broken somehow...
> Weird thing is that we never had any problems until this update came
> along ;-)

What web browser do you normally use?

-d


> Will do some more digging in the mean time.
>
>
> --
> Dick Visser
> System & Networking Engineer
> TERENA Secretariat
> Singel 468 D, 1017 AW Amsterdam
> The Netherlands
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
If this is MS's new way of doing HE, then i'm little bit worried about choosing a protocol
based solely on connectivity to an external entity.

--
Tassos

Dan Wing wrote on 14/11/2012 20:58:
>> -----Original Message-----
>> From: ipv6-ops-bounces+dwing=cisco.com@lists.cluenet.de [mailto:ipv6-
>> ops-bounces+dwing=cisco.com@lists.cluenet.de] On Behalf Of Tassos
>> Chatzithomaoglou
>> Sent: Wednesday, November 14, 2012 9:00 AM
>> To: ipv6-ops@lists.cluenet.de
>> Subject: Re: Yesterday's Windows update causes IPv4 to be default
>>
>> http://support.microsoft.com/kb/2750841
>>
>> /Assume that a computer is configured to use an IPv6 connection as the
>> default connection.
>> Additionally, assume that the computer does not have a connection to an
>> IPv6 network. In this situation, it takes a long time for the computer
>> to connect to an IPv6 site. // // //This issue occurs because Windows
>> tries the IPv6 connection first. After the connection fails because of a
>> time-out error, Windows tries the IPv4 connection.// // //After this
>> update is installed, Windows uses the NCSI functionality to examine the
>> Ipv6 connection. If the connection is broken, Windows uses IPv4 instead
>> of IPv6./
> Yep, Windows 8 has been doing that since Windows 8 shipped. Full
> details are in my Cisco Live slide deck (I'll have to find a pointer)
> and a pretty good summary is
> http://blogs.msdn.com/b/b8/archive/2012/06/05/connecting-with-ipv6-in-windows-8.aspx
>
> -d
>
>
>> --
>> Tassos
>>
>> Dick Visser wrote on 14/11/2012 18:46:
>>> Hi
>>>
>>> I looks like yesterday's batch of updates causes Windows 7 to prefer
>>> IPv4 over IPv6.
>>> After (automatically) installing the whole lot, one of my users
>>> complained about connection issues with WinSCP to our web server.
>>> This runs SSH on IPv6-only. His WinSCP for some reason tried to
>>> connect over IPv4, without falling back to IPv6.
>>> I tried "ping www.terena.org" and that also used IPv4.
>>> "ping -6 www.terena.org" did also work, but I'm pretty sure it used to
>>> default to IPv6.
>>> After clearing the cache in WinSCP things started to work again.
>>>
>>> I didn't install the updates (about 20, including the Office ones) on
>>> my own computer, so I had a look at them first.
>>> One of them is called "No network connectivity on Windows 7-based or
>>> Windows Server 2008 R2-based client computers when a DHCPv6 message is
>>> sent that has a duplicated DUID",
>>> http://support.microsoft.com/kb/2763523, which is the only one that
>>> mentions IPv6.
>>>
>>> Before installing anything, I tried to "ping www.terena.org" and that
>>> used IPv6 all-right.
>>> I then installed all the updates except the suspicious
>>> http://support.microsoft.com/kb/2763523.
>>> After rebooting, "ping www.terena.org" uses IPv4...
>>>
>>> So I guess one of the other updates caused this; I'm now booting up a
>>> VM to investigate which one exactly.
>>> So far the no other major stuff broke, and it looks like only the
>>> default has changed, but I'd thought I mentioned it here anyway, just
>>> in case...
>>>
>>> Anyone else has seen IPv6 issues on Windows since yesterday?
>>>
>>>
>>>
>
>
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On 14 November 2012 19:16, Tassos Chatzithomaoglou
<achatz@forthnetgroup.gr> wrote:
> If this is MS's new way of doing HE, then i'm little bit worried about choosing a protocol
> based solely on connectivity to an external entity.

The only real detail I could find about this testing was the following:

http://blogs.msdn.com/b/b8/archive/2012/06/05/connecting-with-ipv6-in-windows-8.aspx
Windows 8 performs the network connectivity test when you first
connect to a new network; it caches this information and repeats the
test every 30 days. The actual test for connectivity is a simple HTTP
GET to an IPv6-only server that is hosted by Microsoft. (For standards
buffs, this is implemented between rules 5 and 6 of destination
address sorting in our implementation of RFC 3484.) Windows performs a
similar network connectivity test for IPv4 connectivity. If both IPv4
and IPv6 are functioning, IPv6 will be preferred.

Anyone know of a more detailed write-up on this "functionality"?

- Mike
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On 14/11/2012 20:12, Mike Jones wrote:
> Windows 8 performs the network connectivity test when you first
> connect to a new network; it caches this information and repeats the
> test every 30 days. The actual test for connectivity is a simple HTTP
> GET to an IPv6-only server that is hosted by Microsoft.

not so much "happy eyeballs" as "happy eye balls-up". I'm struggling to
understand how and why this is a better approach to dual stacking than rfc
6555.

Nick
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
> not so much "happy eyeballs" as "happy eye balls-up". I'm struggling to
> understand how and why this is a better approach to dual stacking than rfc
> 6555.

Probably it seemed simpler to implement:

bool try_v6 = connect6succeeded() ? true : false;
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On 14/11/2012 20:31, Erik Kline wrote:
> Probably it seemed simpler to implement:
>
> bool try_v6 = connect6succeeded() ? true : false;

programming is littered with the detritus of poor algorithm choices.

I respect Apple's position on HE much more - at least they did their
homework and made a smart technical decision instead of implementing a very
naive algorithm with clear drawbacks like this.

Nick
RE: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
> -----Original Message-----
> From: Tassos Chatzithomaoglou [mailto:achatz@forthnetgroup.gr]
> Sent: Wednesday, November 14, 2012 11:16 AM
> To: Dan Wing
> Cc: ipv6-ops@lists.cluenet.de
> Subject: Re: Yesterday's Windows update causes IPv4 to be default
>
> If this is MS's new way of doing HE, then i'm little bit worried about
> choosing a protocol based solely on connectivity to an external entity.

I concur. *shrug*. I would have liked Apple to prefer IPv6 slightly,
but Apple and Microsoft both have good reasons for their implementation
decisions -- I have talked with the engineers from both companies about
how this is implemented on their respective OSs.

Windows 8 does have an exception for RAs for specific IPv6 networks, but
that is only useful for intra-network traffic (e.g., IPv6 within an
enterprise or within a home). I don't know if that exists in the
enhancement to Windows 7, but I expect it does.

-d


> --
> Tassos
>
> Dan Wing wrote on 14/11/2012 20:58:
> >> -----Original Message-----
> >> From: ipv6-ops-bounces+dwing=cisco.com@lists.cluenet.de [mailto:ipv6-
> >> ops-bounces+dwing=cisco.com@lists.cluenet.de] On Behalf Of Tassos
> >> Chatzithomaoglou
> >> Sent: Wednesday, November 14, 2012 9:00 AM
> >> To: ipv6-ops@lists.cluenet.de
> >> Subject: Re: Yesterday's Windows update causes IPv4 to be default
> >>
> >> http://support.microsoft.com/kb/2750841
> >>
> >> /Assume that a computer is configured to use an IPv6 connection as
> >> the default connection.
> >> Additionally, assume that the computer does not have a connection to
> >> an
> >> IPv6 network. In this situation, it takes a long time for the
> >> computer to connect to an IPv6 site. // // //This issue occurs
> >> because Windows tries the IPv6 connection first. After the connection
> >> fails because of a time-out error, Windows tries the IPv4
> >> connection.// // //After this update is installed, Windows uses the
> >> NCSI functionality to examine the
> >> Ipv6 connection. If the connection is broken, Windows uses IPv4
> >> instead of IPv6./
> > Yep, Windows 8 has been doing that since Windows 8 shipped. Full
> > details are in my Cisco Live slide deck (I'll have to find a pointer)
> > and a pretty good summary is
> > http://blogs.msdn.com/b/b8/archive/2012/06/05/connecting-with-ipv6-in-
> > windows-8.aspx
> >
> > -d
> >
> >
> >> --
> >> Tassos
> >>
> >> Dick Visser wrote on 14/11/2012 18:46:
> >>> Hi
> >>>
> >>> I looks like yesterday's batch of updates causes Windows 7 to prefer
> >>> IPv4 over IPv6.
> >>> After (automatically) installing the whole lot, one of my users
> >>> complained about connection issues with WinSCP to our web server.
> >>> This runs SSH on IPv6-only. His WinSCP for some reason tried to
> >>> connect over IPv4, without falling back to IPv6.
> >>> I tried "ping www.terena.org" and that also used IPv4.
> >>> "ping -6 www.terena.org" did also work, but I'm pretty sure it used
> >>> to default to IPv6.
> >>> After clearing the cache in WinSCP things started to work again.
> >>>
> >>> I didn't install the updates (about 20, including the Office ones)
> >>> on my own computer, so I had a look at them first.
> >>> One of them is called "No network connectivity on Windows 7-based or
> >>> Windows Server 2008 R2-based client computers when a DHCPv6 message
> >>> is sent that has a duplicated DUID",
> >>> http://support.microsoft.com/kb/2763523, which is the only one that
> >>> mentions IPv6.
> >>>
> >>> Before installing anything, I tried to "ping www.terena.org" and
> >>> that used IPv6 all-right.
> >>> I then installed all the updates except the suspicious
> >>> http://support.microsoft.com/kb/2763523.
> >>> After rebooting, "ping www.terena.org" uses IPv4...
> >>>
> >>> So I guess one of the other updates caused this; I'm now booting up
> >>> a VM to investigate which one exactly.
> >>> So far the no other major stuff broke, and it looks like only the
> >>> default has changed, but I'd thought I mentioned it here anyway,
> >>> just in case...
> >>>
> >>> Anyone else has seen IPv6 issues on Windows since yesterday?
> >>>
> >>>
> >>>
> >
> >
RE: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
> -----Original Message-----
> From: Mike Jones [mailto:mike@mikejones.in]
> Sent: Wednesday, November 14, 2012 12:12 PM
> To: Tassos Chatzithomaoglou
> Cc: Dan Wing; ipv6-ops@lists.cluenet.de
> Subject: Re: Yesterday's Windows update causes IPv4 to be default
>
> On 14 November 2012 19:16, Tassos Chatzithomaoglou
> <achatz@forthnetgroup.gr> wrote:
> > If this is MS's new way of doing HE, then i'm little bit worried about
> > choosing a protocol based solely on connectivity to an external
> entity.
>
> The only real detail I could find about this testing was the following:
>
> http://blogs.msdn.com/b/b8/archive/2012/06/05/connecting-with-ipv6-in-
> windows-8.aspx
> Windows 8 performs the network connectivity test when you first connect
> to a new network; it caches this information and repeats the test every
> 30 days. The actual test for connectivity is a simple HTTP GET to an
> IPv6-only server that is hosted by Microsoft. (For standards buffs, this
> is implemented between rules 5 and 6 of destination address sorting in
> our implementation of RFC 3484.) Windows performs a similar network
> connectivity test for IPv4 connectivity. If both IPv4 and IPv6 are
> functioning, IPv6 will be preferred.
>
> Anyone know of a more detailed write-up on this "functionality"?

The URL it tries to visit is http://ipv6.msftncsi.com/ncsi.txt, and
searching the Internet for that FQDN yields some details of how it
works. If it can't retrieve the expected text at that URL, Windows
will order IPv6 to the bottom of its address preference table (by
tweaking its internal RFC3484 rules). The success (or failure) to
get to that IPv6 site is remembered for that network for 30 days,
and then re-tested. I don't know how to encourage it to try a
fresh test, but there must be a registry setting to force that to
occur.

-d


> - Mike
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On Nov 14, 2012, at 4:20 PM, Dan Wing wrote:
>>
>> http://blogs.msdn.com/b/b8/archive/2012/06/05/connecting-with-ipv6-in-
>> windows-8.aspx
>> Windows 8 performs the network connectivity test when you first connect
>> to a new network; it caches this information and repeats the test every
>> 30 days. The actual test for connectivity is a simple HTTP GET to an
>> IPv6-only server that is hosted by Microsoft. (For standards buffs, this
>> is implemented between rules 5 and 6 of destination address sorting in
>> our implementation of RFC 3484.) Windows performs a similar network
>> connectivity test for IPv4 connectivity. If both IPv4 and IPv6 are
>> functioning, IPv6 will be preferred.
>>
>> Anyone know of a more detailed write-up on this "functionality"?
>
> The URL it tries to visit is http://ipv6.msftncsi.com/ncsi.txt, and
> searching the Internet for that FQDN yields some details of how it
> works. If it can't retrieve the expected text at that URL, Windows
> will order IPv6 to the bottom of its address preference table (by
> tweaking its internal RFC3484 rules). The success (or failure) to
> get to that IPv6 site is remembered for that network for 30 days,
> and then re-tested. I don't know how to encourage it to try a
> fresh test, but there must be a registry setting to force that to
> occur.

So, what happens when there is a captive portal to force daily
registration such as at hotels or for corporate guests? With this
algorithm it would appear that ipv6 would be effectively permanently
disabled even though the network is dual stacked.

Or am I missing something?

Dale
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On 11/14/12 21:52 , Dale W. Carder wrote:

> So, what happens when there is a captive portal to force daily
> registration such as at hotels or for corporate guests? With this
> algorithm it would appear that ipv6 would be effectively permanently
> disabled even though the network is dual stacked.
>
> Or am I missing something?
>

Checkout, http://blog.superuser.com/2011/05/16/windows-7-network-awareness/

------
How does it work?
Windows does indeed check a Microsoft site for connectivity, using the
Network Connectivity Status Indicator site. There are a few variations
of the connection checking process:

1. NCSI performs a DNS lookup on www.msftncsi.com, then requests
http://www.msftncsi.com/ncsi.txt. This file is a plain-text file and
contains only the text Microsoft NCSI.

2. NCSI sends a DNS lookup request for dns.msftncsi.com. This DNS
address should resolve to 131.107.255.255. If the address does not
match, then it is assumed that the internet connection is not
functioning correctly.

The exact sequence of when which test is run is not documented; however,
a little bit of digging around with a packet sniffing tool like
Wireshark reveals some info. It appears that on any connection, the
first thing NCSI does is requests the text file (step 1 above). NCSI
expects a 200 OK response header with the proper text returned. If the
response is never received, or if there is a redirect, then a DNS
request for dns.msftncsi.com is made. If DNS resolves properly but the
page is inaccessible, then it is assumed that there is a working
internet connection, but an in-browser authentication page is blocking
access to the file. This results in the pop-up balloon above. If DNS
resolution fails or returns the wrong address, then it is assumed that
the internet connection is completely unsuccessful, and the “no internet
access” error is shown.
-----

So, it looks like there is a test for a captive portal in the IPv4
portion of NCSI. I seems logical that this happens before it gets to
the ipv6.msftncsi.com test, but I don't know for sure.


--
================================================
David Farmer Email: farmer@umn.edu
Office of Information Technology
University of Minnesota
2218 University Ave SE Phone: 1-612-626-0815
Minneapolis, MN 55414-3029 Cell: 1-612-812-9952
================================================
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
Hi,

> The URL it tries to visit is http://ipv6.msftncsi.com/ncsi.txt, and
> searching the Internet for that FQDN yields some details of how it
> works. If it can't retrieve the expected text at that URL, Windows
> will order IPv6 to the bottom of its address preference table (by
> tweaking its internal RFC3484 rules). The success (or failure) to
> get to that IPv6 site is remembered for that network for 30 days,
> and then re-tested. I don't know how to encourage it to try a
> fresh test, but there must be a registry setting to force that to
> occur.


WTF? So one failure (either locally or of a Microsoft webserver) causes an IPv6 outage of one month... It will be 'interesting' to see internet traffic patterns shift for a month when ipv6.msftncsi.com has an outage...

I think this is a *very* bad design decision.
Sander
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On Thu, Nov 15, 2012 at 10:48:58AM +0100, Sander Steffann wrote:
> I think this is a *very* bad design decision.

Indeed. It also means (as far as I understand it) that in non
Internet-connected networks, IPv4 will be preferred over IPv6,
which is really a bad thing.

Sigh. Another "great idea" to deal with. :-(

Best regards,
Daniel

--
CLUE-RIPE -- Jabber: dr@cluenet.de -- dr@IRCnet -- PGP: 0xA85C8AA0
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
Create an internal quad-a record/zone and host it yourself in an internal
only situation would be the go.

I'm surprised there is no registry setting to alter this behaviour in
windows. Surely not?
On Nov 15, 2012 9:07 PM, "Daniel Roesen" <dr@cluenet.de> wrote:

> On Thu, Nov 15, 2012 at 10:48:58AM +0100, Sander Steffann wrote:
> > I think this is a *very* bad design decision.
>
> Indeed. It also means (as far as I understand it) that in non
> Internet-connected networks, IPv4 will be preferred over IPv6,
> which is really a bad thing.
>
> Sigh. Another "great idea" to deal with. :-(
>
> Best regards,
> Daniel
>
> --
> CLUE-RIPE -- Jabber: dr@cluenet.de -- dr@IRCnet -- PGP: 0xA85C8AA0
>
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On 14/11/2012 22:20, Dan Wing wrote:
> tweaking its internal RFC3484 rules). The success (or failure) to
> get to that IPv6 site is remembered for that network for 30 days,
> and then re-tested. I don't know how to encourage it to try a
> fresh test, but there must be a registry setting to force that to
> occur.

sounds bizarre that windows is assuming that protocol stability operates on
a 30 day basis. What if ipv6 is tweaked up on day 1, then ipv6 goes down
on day 2 because the corporate lan IT people doesn't know and doesn't care
about ipv6, and the W8 end users find themselves experiencing 29 days of
irritating timeouts?

I'm sure that the Microsoft people have thought about this, but the blog
posting doesn't give enough details.

Nick
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On 14 November 2012 23:20, Dan Wing <dwing@cisco.com> wrote:

> The URL it tries to visit is http://ipv6.msftncsi.com/ncsi.txt, and
> searching the Internet for that FQDN yields some details of how it
> works. If it can't retrieve the expected text at that URL, Windows
> will order IPv6 to the bottom of its address preference table (by
> tweaking its internal RFC3484 rules). The success (or failure) to
> get to that IPv6 site is remembered for that network for 30 days,
> and then re-tested. I don't know how to encourage it to try a
> fresh test, but there must be a registry setting to force that to
> occur.


It looks like there is, see attached regedit screendump.
It differs from the one in the superuser.com blog post, in that there
are v6 versions of some parameters.
>From the looks of it, it allows you to specify which host names,
addresses, and web server file paths should be used for the probe.
So while the defaults might be sucky, you are able to change them.


--
Dick Visser
System & Networking Engineer
TERENA Secretariat
Singel 468 D, 1017 AW Amsterdam
The Netherlands
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
Hi,

On Wed, Nov 14, 2012 at 09:42:40PM +0000, Nick Hilliard wrote:
> I respect Apple's position on HE much more - at least they did their
> homework and made a smart technical decision instead of implementing a very
> naive algorithm with clear drawbacks like this.

Uh, so what exactly is "smart" about "your client web connects randomly
flip-flop between IPv4 and IPv6 when talking to the very same server"?

(I've recently been bitten by that - due to a mishap on the server,
IPv4 failed 404, while IPv6 worked - got lots of very inconclusive
results from users about this, like "I occasionally get 'access denied'
and when I press reload, it's working again". Talk about deterministic
behaviour in networks...)

Gert Doering
-- NetMaster
--
have you enabled IPv6 on something today...?

SpaceNet AG Vorstand: Sebastian v. Bomhard
Joseph-Dollinger-Bogen 14 Aufsichtsratsvors.: A. Grundner-Culemann
D-80807 Muenchen HRB: 136055 (AG Muenchen)
Tel: +49 (89) 32356-444 USt-IdNr.: DE813185279
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On 15/11/2012 12:00, Gert Doering wrote:
> (I've recently been bitten by that - due to a mishap on the server,
> IPv4 failed 404, while IPv6 worked - got lots of very inconclusive
> results from users about this, like "I occasionally get 'access denied'
> and when I press reload, it's working again". Talk about deterministic
> behaviour in networks...)

urgh, messy. I don't know how that could be handled from the end user
point of view - although I do agree that it is a valid problem. :-(

Nick
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On Wed, Nov 14, 2012 at 4:42 PM, Nick Hilliard <nick@foobar.org> wrote:

> I respect Apple's position on HE much more - at least they did their
> homework and made a smart technical decision instead of implementing a very
> naive algorithm with clear drawbacks like this.
>

The Apple algorithm has drawbacks too. To wit:

1. It biases in favour of IPv4 by trying the A record first.
2. By always preferring the fastest protocol, even on a perfect dual-stack
network it will use IPv6 only ~50% of the time (unless IPv4 is degraded).
3. It imposes twice the connection load on server operators.
4. It's non-deterministic, which some websites don't like as they tie your
cookies to your IP address.
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On 15/11/2012 12:08, Lorenzo Colitti wrote:
> 1. It biases in favour of IPv4 by trying the A record first.
> 2. By always preferring the fastest protocol, even on a perfect dual-stack
> network it will use IPv6 only ~50% of the time (unless IPv4 is degraded).
> 3. It imposes twice the connection load on server operators.

The connection load problem is not good, no. I don't know how this is
going to affect operators in the real world, but I could see that the
Googles of the world would be none too happy about it.

> 4. It's non-deterministic, which some websites don't like as they tie your
> cookies to your IP address.

This is a problem which already happens with privacy addresses. I'm
constantly getting hit by it. HE will exacerbate it.

Nick
Re: Yesterday's Windows update causes IPv4 to be default [ In reply to ]
On 15/11/2012 12:38, Nick Hilliard wrote:
> sounds bizarre that windows is assuming that protocol stability operates on
> a 30 day basis. What if ipv6 is tweaked up on day 1, then ipv6 goes down
> on day 2 because the corporate lan IT people doesn't know and doesn't care
> about ipv6, and the W8 end users find themselves experiencing 29 days of
> irritating timeouts?

I am both concerned and confused about this. What about people like me
who use a device called a *laptop* and travel around, connecting to all
sorts of networks in conferences, some being single stack and some dual
stack?
Baffled,

Olivier

--
Olivier MJ Crépin-Leblond, PhD
http://www.gih.com/ocl.html

1 2 3 4  View All