falling-sky / source Goto Github PK

From [email protected] on January 28, 2011 01:53:38

When I intentionally broke my IPv6 connectivity to see how the site reports it, the test page didn't start loading at all. Reason was because it hung on the requests to ajax.googleapis.com, which for all Google-whitelisted networks (incl. mine) is dual-stacked.

You should probably host your own copy of jquery and mootols...

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=19

From jfesler on January 17, 2011 08:43:52

opera 11 has unresolvable quirks making this site code all but impossible.

We can at least detect them - and on failure, point them to an FAQ on what we know of the problems, as well as a simpler test method.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=16

From jfesler on January 13, 2011 08:36:35

The point of aaaa.v6ns.test-ipv6.com is to test the DNS server.

It is remotely possible that the DNS server has IPV6 and the client does not. Instead create a ds.v6ns.test-ipv6.com that maximizes the chances of the client connecting.

Also update the detection logic to recognize that this test passed, if the host appears to be IPv4 only. It is worth noting!

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=10

From [email protected] on January 28, 2011 02:44:12

Current Firefox doesn't set AI_ADDRCONFIG so it always requests AAAA records. OS X doesn't do proper source address selection and prefers IPv6 (as long as there's no 6to4 address present). When there's a default IPv6 route present too, this leads to long timeout as the browser attempts to connect to the global destination using the OS X's link-local IPv6 address for the source.

See: http://getipv6.info/index.php/Customer_problems_that_could_occur#Apple_Mac_OS_X_2 Suggestion: Disable AAAA lookups by setting network.dns.disableIPv6 in the about:config preference editor.

Come to think of it, this might be a prudent suggestion to anyone with broken IPv6 using Firefox.

Firefox 4 will BTW set AI_ADDRCONFIG, fortunately - the patch is already in their developement repo and will likely be part of the next beta iteration.

Tore

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=22

From jfesler on May 12, 2011 17:08:21

To: [email protected]
Date: Thu, 12 May 2011 17:00:38
Subject: test-ipv6.com feedback [aaron@bigg....]

purpose : questions
tokens : dualstack:unsafe,dualstack:ipv6_preferred,v6ns:ok
score_transition: 10
score_strict : 10
a : ok,846
aaaa : ok,687
ds : ok,655 (via ipv6)
ipv4 : ok,471
ipv6 : ok,429
v6mtu : ok,449

v6ns : ok,891

dsmtu :

...

cookie : 7e21dc286e3fc9140dce4083e61fa741
user-agent : Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:2.0.1) Gecko/20100101 Firefox/4.0.1
referer : http://test-ipv6.com/ subdomain : server1.test-ipv6.com
...

Notes

The test results are... a little odd. I got this error:
Test for Dual Stack DNS and large packet timeout (15.004s)

However, the end result showed 10/10 for both v4 and v6. Also, the Technical Info didn't indicate any
suggested fixes or even what this actually means. Is this some kind of MTU issue?

Comments

If you need any of this just ask. My MTUs are set to 1500 everywhere as far as I know. I use a tunnel
provided by tunnelbroker.net for IPv6, I don't think my ISP offers native v6 yet. Korea proxies all
DNS, as there is a small amount of net censorship (mostly just blocking North Korean sites, as it is
illegal for regular citizens to communicate with North Korea).

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=46

Skipping this issue number to maintain synchronization with Google Code issue IDs.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=32

From jfesler on January 27, 2011 19:26:41

(from tore anderson)

Opera < 10.5 will prefer 6to4, Teredo over IPv4.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=18

From [email protected] on February 21, 2011 05:34:13

Missing table column "language" on mirrors subpage

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=39

From jfesler on January 17, 2011 22:49:31

Particularly on Opera and NoScript

Instead of reporting "You have no IPv4 address" and "You have no IPv6 address", those messages should be hidden, and a suitable replacement put up.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=17

From jfesler on January 10, 2011 21:44:03

1. firefox 3
2. ipv6 test succeeds but aaaa test fails

From "Ronnie_USA" on dslreports.com http://www.dslreports.com/forum/r25229656-IPv6-on-a-Linksys-E3000-w-Tomato-Firmware-RAF1.28.8691 1. Key in “about:config” into address bar, this will take you to the config page.

1. Key in “network.dns.disableIPv6″ into the filter bar and hit enter.
2. Right click on the line that read “network.dns.disableIPv6″ and select the “Toggle” option. This will switch the value from “true” to “false”
3. Restart the browser and enjoy the improvement.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=3

From jfesler on February 19, 2011 14:24:08

Disabling the survey through config.inc will be needed; some mirrors we can't (or don't want) to save customer stats.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=36

From jfesler on January 13, 2011 09:24:56

jgoodwin@[censored] reported.

6to4 was in the tokens; and the scores reflected it. jgoodwin reports
that the message warning 6to4 was not visibly present.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=12

From jfesler on January 12, 2011 12:37:56

Klaus Darilion [email protected]

Another test you could add to your website would be to test HTTPS connections
over IPv6. There is a bug report for Firefox - this bug is somehow triggered
only if certain conditions are met:

https://bugzilla.mozilla.org/show_bug.cgi?id=513659

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=7

From jfesler on January 30, 2011 11:48:06

This is principally for Opera compatibility.

Now that I have a mechanism for "timing out" image loads, go back to using images for the primary testing mechanism. Images should be served with cache busting techniques (other than dynamically altering the url).

Still use jQuery.jsonp - but only to try and identify the IP address of the end user. These queries should go to ipv4,ds,ipv6 to identify of the preference is ipv4 or ivp6 for dualstack.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=24

From jfesler on January 30, 2011 14:31:41

The progress bar is only good if the number of tests you plan on running is known in advance.

We're reaching the point where we might conditionally want to add additional tests, wait for those followup tests to finish, then resume scoring.

Change the progress bar to a throbber. Indicate # of tests ran, # of tests total. This number may increase.

First place this could benefit: where we fallback to images.

Second place we can benefit: where we add additional tests (such as the HTTPS check recommended in issue 7 .

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=26

From jfesler on May 05, 2011 09:13:28

What steps will reproduce the problem? 1. 2. 3. What is the expected output? What do you see instead? Please use labels and text to provide additional information.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=44

From jfesler on February 27, 2011 16:01:09

After a random period of time, mod_ip starts causing core dumps. As it walks the prefix list, it gets a pointer to inaccessible memory - which means, our memory is being stomped on by something.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=40

From jfesler on February 19, 2011 14:24:40

The "Stats" tab doesn't make sense for all sites. Allow config.inc to disable this.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=37

From jfesler on January 17, 2011 08:42:42

Generate a simple test that can be used for human observation of pass/fail.
3 images: v4, ds, v6.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=15

From [email protected] on February 17, 2011 08:50:46

Some little improvements made. There are still some things to improve, which I'll do later...

Attachment: text.de.patch

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=34

From jfesler on January 11, 2011 09:59:12

On Tue, 11 Jan 2011, Ince, Kamil wrote:

Actually both columns for v6mtu tests where missing,
Added with similar features as the others columns:

alter table survey add column status_v6mtu enum('ok','slow','bad','timeout') NOT NULL;
alter table survey add key status_v6mtu (status_v6mtu);
alter table survey add column time_v6mtu mediumint(8) unsigned;

And survey stats are now being collected nicely.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=4

From jfesler on February 07, 2011 06:45:29

http://code.google.com/p/falling-sky/wiki/LocalizationPlan Need to adjust the framework to:
1: make it easier to translate
2: make it easier to share the translation
3: stay up to date with updates from the falling-sky project

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=29

From jfesler on February 12, 2011 16:51:05

Sent anonymously

I take issue with the fact that if I have working teredo (as opposed
to teredo-minimum) but also have dualstack:ipv4_preferred, I end up
with a final score of v4:7/10 v6:7/10.

The ipv4_preferred moots the potential impact of Teredo on dual stack
machines, so under such a configuration I should be v4:10/10 v6:7/10.

Patch (sorry it is not in diff format, but it is only three new
lines). In symptoms.js check for a IPv4 preference in the dual stack.
Add an extra message in Messages.js (oh and according to the comment,
you must update MySQL) I did not change the message from the teredo
message, but it might be wort mentioning that this would only be a
concern for IPv6-only site.

[symptoms.js]
   if (teredo) {
       if (aaaa === "bad") {
           res.push("teredo-minimum");

•        } else if (mini_primary.match(/^..[os]b/)) {
•            res.unshift("teredo-ipv4preferred");
         } else {
             res.push("teredo");
         }
     }

[messages.js]
+GIGO.messages["teredo-ipv4preferred"] = [10, 7, "BLUE", "Your IPv6
connection appears to be using Teredo, a type of IPv4/IPv6 translation
using a public gateway.  The quality for this may suffer, as
you are using a public gateway to reach IPv6 based sites."];

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=30

From jfesler on January 30, 2011 14:40:21

Today the queue of tests includes only the url; we branch (later) based on url type whether to do an image test or a json test.

This should be updated to include the image function itself into the queue, so that we can push arbitrary tests to be scheduled. They can either be ran in series or parallel, depending on the browser. This combined with the status bar changes, will make it easier for us to mix and match test types.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=27

From jfesler on January 12, 2011 12:36:26

[I've done tis on the live site, but need to document it for others]

From: Andreas Papst [email protected]
To: [email protected]
Date: Wed, 12 Jan 2011 10:45:40
Subject: [test-ipv6.com] Missing AAAA-RR for v6ns1.test-ipv6.com.

Hello,

first of all thank you for providing this IPv6 online test!

May I do a suggestion: Please enter into your v6ns1.test-ipv6.com. zone
a quad-A record for the hostname of its nameserver:

v6ns1.test-ipv6.com. 360 IN AAAA 2001:470:1:18::1000

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=6

From jfesler on February 15, 2011 07:04:54

See email for full context. Publishable bits are:

From: A. Keller - Network Design GmbH

Hi Jason

we'd like to host a local copy of this test for our (mainly dsl) customers.
We provide native IPv6 (PPP IP6CP) to these customers.

The mirror can be official, when we see to much traffic (a few
Tbyte/Month) we might reconsider that...

we would like you to manage this service as stated in the wiki.

The "service" domain for the service should be: test-ipv6.ch

Additional serveraliases:
v6test.usterweb.com
v6test.usterweb.ch
v6test.usterweb.net
v6test.ndnet.ch
v6test.ndm.ch
v6test.as44706.net
sixte.st
www.sixte.st
www.test-ipv6.ch

Translation-wise german with fallback to english (browser language)
would be great. If not possible only english is ok also. (If you need a
german translation send me a .po file or some other sort of table and I
will do it...)

Server location is Zurich, Switzerland.

Zonefile test-ipv6.ch (for reference and error checking):
$ORIGIN test-ipv6.ch.
@ IN SOA ns1.ndnet.ch. hostmaster.ndnet.ch. (
2011021500
3h
30m
4w
1h
)

@ 3600 IN NS ns1.ndnet.ch.
@ 3600 IN NS ns2.ndnet.ch.
@ 3600 IN NS ns3.ndnet.ch.

@ 3600 IN A 95.143.49.6
www 3600 IN A 95.143.49.6
ipv4 3600 IN A 95.143.49.6
ipv6 3600 IN AAAA 2a00:10c0:1::6

ds 3600 IN A 95.143.49.6
ds 3600 IN AAAA 2a00:10c0:1::6

ns1 3600 IN AAAA 2a00:10c0:1::6
v6ns 3600 IN NS ns1

If you don't have time I will install the stuff and try to keep up with
your updates, but as you offer the service to manage the server, I'm
happy to let you do it :-)

If you need any further information just ask...

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=33

Imported from https://code.google.com/p/falling-sky/issues/detail?id=74

From jfesler on April 30, 2011 10:16:07

remaker @ cisco reported:

host had IPv6 address
host had IPv6 route
host was not using teredo, but instead a proper global routable address

host did NOT do AAAA lookups
host DID connect by raw IP

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=43

From jfesler on April 29, 2011 18:17:41

Need to..
when detecting "aaaa" connected but got ipv4 instead, put up a big red flag that says DNS is lying.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=42

From [email protected] on April 01, 2011 14:30:53

What steps will reproduce the problem? 1.browse to the page http://test-ipv6.com/faq_whyipv6.html 2.scroll to very bottom, last line. (line 106)

What is the expected output?
a working click-able link

What do you see instead?
see my onclick="help_page('faq_6to4.html','6to4')">6to4 comments.

What version of the product are you using? r385 - its the same on all mirrors

paul

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=41

Imported from https://code.google.com/p/falling-sky/issues/detail?id=92

If someone is viewing /index.html.de then we should link to and show /faq_whatever.html.de .

From jfesler on December 25, 2010 00:28:37

INSTALL needed.
Callback information needed for people who want their own UI.
License graffiti across all files needed.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=1

From jfesler on February 13, 2011 09:10:18

Detect proxies by the presence of XFF. When found, present a warning that the test may be of their proxy instead of their browser.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=31

From [email protected] on February 02, 2011 00:43:22

Both on Firefox 4 nightly and through curl, I get two nice Ñs in the second paragraph (first question). It looks like they're probably supposed to be em-dashes.

In fact, looking at faq_ipv6day.html in the source tree, I see Ñ, which is, in fact, quite not what you want; try —.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=28

From [email protected] on May 09, 2011 20:45:50

Unfortunately you have chosen not to list our local 6to4 tunnel offerings. We have had both commercial ( http://ipv6now.com.au/tunnels.php ) and free ( http://ipv6now.com.au/free.php ) tunnels for over 3 years. Our service has over 10,000 registered users from all over the world. Our commercial tunnel service is one of the very few that supplies a commercial grade Service Level Agreement as part of its feature set.

My request is for you to consider mentioning our service on your tunnel information page.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=45

From jfesler on February 19, 2011 14:23:25

We have locale for language.

Add a second variant that allows fetching mirror-specific versions of files or includes. This will allow for logos or any footer text such as disclaimers.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=35

From jfesler on February 20, 2011 17:51:59

To assist datamining efforts, store the "type" of IPv6 as a record with the survey data.
sixxs (which one)
tunnelbroker
teredo

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=38

From jfesler on January 28, 2011 08:23:46

tokens : dualstack:safe,ipv4_only,v6ns:okscore_transition: 7
score_strict : 0
a : ok,361
aaaa : bad,248
ds : ok,312 (via ipv4)
ipv4 : ok,253
ipv6 : ok,335
v6mtu : bad,222

ipv4_only is an error; ipv6 did work. just aaaa did not.

This is acting sort of like Teredo - no AAAA lookups (or successes).

Tunnel was via sixxs.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=23

Imported from https://code.google.com/p/falling-sky/issues/detail?id=85

From kasperd

The section about problematic 6to4 addresses should also include 2002:6440::/26. I don't know how widespread the problem is, but it is likely to become a problem on any network using RFC 6598 addresses without also having native IPv6.

From jfesler on January 12, 2011 19:42:33

Some folks have cookies already set for other purposes, assigned to cookie name "apache".

Update example apache configs, docs/INSTALL.html,
and source/php/survey.php to use a new name instead of the default "apache" name.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=8

From [email protected] on January 28, 2011 02:22:47

Maybe the site already does this but I couldn't make it warn me about broken IPv6 connectivity for some reason (I submitted a bug using the feedback form about this).

Anyway. If the user is broken or using 6to4/Teredo:

1. If he's using Mac OS X 10.6.x where x < 5, suggest upgrading - 10.6.5 added de-preferring of 6to4.

2. If he's using Mac OS X 10.y.x where y < 6 and he's on an Intel platform, suggest upgrading, but since this is a paid upgrade, maybe also suggest disabling IPv6 in OS (System Preferences -> Network -> Advanced -> TCP/IP -> Configure IPv6 -> Off)

3. If he's on a PowerPC platform, suggest disabling IPv6 in OS.

Tore

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=20

From jfesler on January 16, 2011 12:24:08

norman@[withheld] suggests:

I noticed that SixXS subnets are not detected properly though :-)This is mostly because the network addresses can't be expressed as regular expressions.For the apache module you might be able to use apr_ipsubnet_create and apr_ipsubnet_test, but I'm not sure how you'd get the apr_sockaddr_t to pass to test.

For appengine, you could use a library like http://code.google.com/p/ipaddr-py/ This would allow you to use the addresses with netmask size, and use "addr in IPv6Network".

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=14

From [email protected] on January 28, 2011 02:32:20

OK, you'll have fun implementing this one if you decide to do it. :-)

It's this bug: http://getipv6.info/index.php/Customer_problems_that_could_occur#D-Link I think you stand a chance of detecting it if you do the following. Set up a test hostname something like this and try to load an object from it:

d-link-test.test-ipv6.com IN AAAA d8da:e472::

d8da:e472 corresponds to 216.218.228.114. Now, this test is of course expected to fail, since the address is utterly bogus. However, if the D-Link bug is triggered, the host will get in return a "IN A 216.218.228.114" reply from the router and the request will succeed. If that happens, warn the user that he has a broken router and that he should either replace it or work around it by disabling IPv6 in his operating system or web browser.

Tore

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=21

From [email protected] on May 21, 2011 11:49:00

What steps will reproduce the problem? 1. Set ja (or ja-JP) preferred on your web browser.
2. Access http://test-ipv6.com 3. View source What is the expected output? What do you see instead? 1) Use appropriate language. Otherwise, browser cannot find appropriate font.

For ja-JP,
GOOD:
BAD:

2. Put appropriate charset before the first non-ascii character.
   GOOD: <meta http-equiv="Content-Type" ....> <title>...</title>

BAD: <title>...</title>
<meta http-equiv="Content-Type" ....> What version of the product are you using? On what operating system? Please provide any additional information below.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=47

From jfesler on January 13, 2011 18:35:51

from imre gergely:

I should have thought of that... A is the default when not specifying a query type to dig.
And I managed to set up my recursor, too. I'm using PowerDNS recursor (version 3.3), and I had to set outgoing query source for
IPv6 which enables outgoing IPv6 support.

From the manual: http://doc.powerdns.com/built-in-recursor.html "query-local-address6
Send out local IPv6 queries from this address or addresses Disabled by default, which also disables outgoing IPv6 support. Since
version 3.2, multiple addresses can be specified, separated by a comma."

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=13

From [email protected] on December 31, 2010 23:45:08

got revision 54 and am getting:
./build.pl -v
Processing: js/index.js
Processing: js/without-ui.js
Processing: html/index.html
crap!file error - tab_debug.inc: not found at ./build.pl line 110.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=2

From jfesler on January 13, 2011 08:32:56

6to4
Teredo
v6ns

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=9

From jfesler on January 13, 2011 09:12:45

The comment form is a bit too easy to fill out.

If the detected user is IPV4 only, and they are not in danger on flag day, introduce new text and perhaps a link to a help page explaining more of what they might do. (Bug ISP, use tunnels, etc).

Many of the comments coming in this week revolve around people that are simply IPv4 only.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=11

From jfesler on January 30, 2011 13:39:25

http://getipv6.info/index.php/Customer_problems_that_could_occur#Apple_Mac_OS_X_2 "Broken" users (ie, can't reach dual stack)" should be given hints.

Above URL is a gold mine. We should separately give that URL out in addition to other advice, due to the level of maintenance it gets.

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=25

From jfesler on January 11, 2011 12:43:23

Thanks to pieter@[censored].

tokens : ipv6:no_address,dualstack:unsafe,confused:ASK,confused:ottt:ottt
score_transition: 0
score_strict : 0
a : ok,889
aaaa : timeout,15039
ds : timeout,15039 (via ipv4)
ds : timeout,15039 (via ipv6)
v4ip : ok,437
v6ip : timeout,15038
v6mtu : timeout,15038
v6ns : timeout,8533
v6mtu : timeout,15038

v6ns : timeout,8533

[rest snipped for privacy]

Pasted in the comments:

Publish Type Met Prefix Idx Gateway/Interface Name

No Manual 256 ::/0 19 2001:470:XXX:XXX::1

Original issue: http://code.google.com/p/falling-sky/issues/detail?id=5