Emacs can't reach dx.doi.org?

Emacs can't reach dx.doi.org?

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 1538 bytes --]

Hi all,

I'm having a surprising issue, whereby Emacs -Q sometimes doesn't seem to be able to reach dx.doi.org; other websites work fine, but for that particular one the network process stays in "connect" state for a while, then goes to "failed" state. I'm pretty sure it worked a few days ago, and it has worked intermittently today. I'm on the latest pretest.

Before opening a bug report, I'd like to make sure that I'm not doing anything silly: I'm trying to open http://dx.doi.org/10.1016/s0262-4079(08)62292-7 ; with eww, I get "Loading http://dx.doi.org/10.1016/s0262-4079(08)62292-7...". With 

  (url-retrieve
   "http://dx.doi.org/10.1016/s0262-4079(08)62292-7"
   (lambda (&rest args) (message "args: %S" args)))

I get 'Contacting host: dx.doi.org:80', and nothing. After a few calls list-processes shows this:

dx.doi.org      failed   *http dx.do... --           (network connection to dx.doi.org)
dx.doi.org<1>   failed   *http dx.do... --           (network connection to dx.doi.org)
dx.doi.org<2>   connect  *http dx.do... --           (network connection to dx.doi.org)

Setting url-debug doesn't seem to make any difference.

Using curl to get the same page works fine. I tried to have a look in wireshark, but it seems that Emacs never opens a connection; at least, I can't see anything.

Does this ring a bell for anyone? Can anyone reproduce it? I can't always reproduce it myself; it happens for a while, then it goes away. In all cases the website is still reachable with curl.

Clément.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Emacs can't reach dx.doi.org?

[Colin Baxter]

>
> I'm having a surprising issue, whereby Emacs -Q sometimes doesn't seem to be able to reach dx.doi.org; other websites work fine, but for that particular one the network process stays in "connect" state for a while, then goes to "failed" state. I'm pretty sure it worked a few days ago, and it has worked intermittently today. I'm on the latest pretest.
>
> Before opening a bug report, I'd like to make sure that I'm not doing anything silly: I'm trying to open http://dx.doi.org/10.1016/s0262-4079(08)62292-7 ; with eww, I get "Loading http://dx.doi.org/10.1016/s0262-4079(08)62292-7...". With 
>
>   (url-retrieve
>    "http://dx.doi.org/10.1016/s0262-4079(08)62292-7"
>    (lambda (&rest args) (message "args: %S" args)))
>

------ snip ------

>
> Clément.

Hi Clément,

I get http://dx.doi.org/10.1016/s0262-4079(08)62292-7 with no problems,
using eww directly on GNU Emacs 25.1.50.5. If I \C-j after

(url-retrieve
"http://dx.doi.org/10.1016/s0262-4079(08)62292-7"
(lambda (&rest args) (message "args: %S" args)))

in the scratch buffer, I get the message

Contacting host: dx.doi.org:80
args: ((:redirect
"http://www.sciencedirect.com/science/article/pii/S0262407908622927"
:redirect
"http://www.sciencedirect.com/science/article/pii/S0262407908622927?via%3Dihub"
:redirect
"http://linkinghub.elsevier.com/retrieve/articleSelectSinglePerm?Redirect=http%3A%2F%2Fwww.sciencedirect.com%2Fscience%2Farticle%2Fpii%2FS0262407908622927%3Fvia%253Dihub&key=f431f9792c30720d33c588d7004396ef0c8972a9"
:redirect
"http://linkinghub.elsevier.com/retrieve/pii/S0262407908622927"))

but no web-site in an eww buffer.

Hope this is of some help.

Colin.

Re: Emacs can't reach dx.doi.org?

[Andreas Schwab]

Clément Pit--Claudel <clement.pit@gmail.com> writes:

>   (url-retrieve
>    "http://dx.doi.org/10.1016/s0262-4079(08)62292-7"
>    (lambda (&rest args) (message "args: %S" args)))
>
> I get 'Contacting host: dx.doi.org:80', and nothing. After a few calls list-processes shows this:

Do you have IPv6 connectivity problems?  What does `telnet dx.doi.org 80'
do?

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Emacs can't reach dx.doi.org?

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 238 bytes --]

On 03/19/2016 05:23 AM, Andreas Schwab wrote:
> Do you have IPv6 connectivity problems?  What does `telnet dx.doi.org 80'
> do?

It works fine:

> Trying 38.100.138.163...
> Connected to dx.doi.org.
> Escape character is '^]'.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Emacs can't reach dx.doi.org?

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 679 bytes --]

On 03/19/2016 05:23 AM, Andreas Schwab wrote:
> Clément Pit--Claudel <clement.pit@gmail.com> writes:
> 
>>   (url-retrieve
>>    "http://dx.doi.org/10.1016/s0262-4079(08)62292-7"
>>    (lambda (&rest args) (message "args: %S" args)))
>>
>> I get 'Contacting host: dx.doi.org:80', and nothing. After a few calls list-processes shows this:
> 
> Do you have IPv6 connectivity problems?  What does `telnet dx.doi.org 80'
> do?

I'm running into this again today, and it seems that you were right! Thanks:

    $ telnet dx.doi.org 80
    Trying 2001:550:100:6::138:162...

(hangs for a while)
Is there anything I can do on my side to diagnose this?

Clément.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Emacs can't reach dx.doi.org?

[Andreas Schwab]

Clément Pit--Claudel <clement.pit@gmail.com> writes:

> I'm running into this again today, and it seems that you were right! Thanks:
>
>     $ telnet dx.doi.org 80
>     Trying 2001:550:100:6::138:162...
>
> (hangs for a while)
> Is there anything I can do on my side to diagnose this?

Useful diagnostic tools: traceroute6, mtr

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Emacs can't reach dx.doi.org?

[Simon Leinen]

On Mon, Mar 21, 2016 at 8:58 PM, Clément Pit--Claudel
<clement.pit@gmail.com> wrote:
> On 03/19/2016 05:23 AM, Andreas Schwab wrote:
>> Clément Pit--Claudel <clement.pit@gmail.com> writes:
>>
>>>   (url-retrieve
>>>    "http://dx.doi.org/10.1016/s0262-4079(08)62292-7"
>>>    (lambda (&rest args) (message "args: %S" args)))
>>>
>>> I get 'Contacting host: dx.doi.org:80', and nothing. After a few calls list-processes shows this:
>>
>> Do you have IPv6 connectivity problems?  What does `telnet dx.doi.org 80'
>> do?
>
> I'm running into this again today, and it seems that you were right! Thanks:
>
>     $ telnet dx.doi.org 80
>     Trying 2001:550:100:6::138:162...
>
> (hangs for a while)

Works from here:

  : leinen@macsl[foo]; telnet dx.doi.org 80
  Trying 2001:550:100:6::138:163...
  Connected to dx.doi.org.
  Escape character is '^]'.
  ^]
  telnet> close
  Connection closed.

> Is there anything I can do on my side to diagnose this?

traceroute/traceroute6?

You seem to have a public IPv6 address and/or an IPv6 default route,
because otherwise telnet and other programs wouldn't try to connect
over IPv6.  But maybe your IPv6 connectivity is broken/blocked by
firewall/nonexistent.  Or, if IPv6 generally works for you, there may
be a routing problem further "upstream".  As I said, traceroute{,6}
might help.  What is your IPv6 address? There are some IPv6 peering
issues/disputes that may cause disconnectivity between some parts of
the IPv6 Internet, maybe you are being hit by this.  I see dx.doi.org
behind Cogent, who are involved in these disputes.

While the routing issue certainly isn't an Emacs problem, it would be
nice if Emacs would notice the issue and quickly fall back to IPv4, as
per the recommendations in RFC 6555.  I assume IPv4 works for you?
i.e. telnet -4 dx.doi.org 80
-- 
Simon.

: 130leinen@macsl[foo]; traceroute6 -I dx.doi.org
traceroute6: Warning: dx.doi.org has multiple addresses; using
2001:550:100:6::138:162
traceroute6 to dx.doi.org (2001:550:100:6::138:162) from
2001:620:0:69::101, 64 hops max, 16 byte packets
 1  vpngw-b.switch.ch  15.884 ms  11.575 ms  10.809 ms
 2  swiofw1-v15.switch.ch  10.861 ms  9.493 ms  14.564 ms
 3  swiws2-v910.switch.ch  11.518 ms  11.757 ms  15.242 ms
 4  swiws1-v640.switch.ch  13.000 ms  16.052 ms  10.730 ms
 5  swiez1-10ge-4-15.switch.ch  11.239 ms  11.567 ms  22.094 ms
 6  swiba2-10ge-1-4.switch.ch  21.159 ms  10.761 ms  14.005 ms
 7  2001:978:2:3e::2:1  12.614 ms  15.255 ms  26.509 ms
 8  te0-0-0-1.agr12.zrh01.atlas.cogentco.com  17.263 ms  16.212 ms  21.287 ms
 9  *
    te0-7-0-35.ccr21.zrh01.atlas.cogentco.com  15.346 ms  15.956 ms
10  be2295.ccr41.par01.atlas.cogentco.com  33.689 ms  43.170 ms  30.533 ms
11  be2746.ccr41.jfk02.atlas.cogentco.com  103.163 ms  96.974 ms  100.221 ms
12  be2806.ccr41.dca01.atlas.cogentco.com  105.293 ms *  103.563 ms
13  te0-0-1-0.rcr21.iad03.atlas.cogentco.com  112.268 ms  108.030 ms  105.387 ms
14  te0-0-2-3.agr11.iad03.atlas.cogentco.com  110.695 ms  159.057 ms  106.802 ms
15  2001:550:100:6::138:162  104.891 ms  113.060 ms  122.931 ms

Re: Emacs can't reach dx.doi.org?

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 1339 bytes --]

On 03/21/2016 04:24 PM, Simon Leinen wrote:
> Works from here:

Thanks for testing!

>> Is there anything I can do on my side to diagnose this?
> 
> traceroute/traceroute6?
> 
> You seem to have a public IPv6 address and/or an IPv6 default route,
> because otherwise telnet and other programs wouldn't try to connect
> over IPv6. 

Indeed, I'm on MIT CSAIL's network.

> But maybe your IPv6 connectivity is broken/blocked by
> firewall/nonexistent.  Or, if IPv6 generally works for you, there may
> be a routing problem further "upstream".  As I said, traceroute{,6}
> might help.  What is your IPv6 address? There are some IPv6 peering
> issues/disputes that may cause disconnectivity between some parts of
> the IPv6 Internet, maybe you are being hit by this.  I see dx.doi.org
> behind Cogent, who are involved in these disputes.

Interesting, thanks for the background info! I'm raising the issue with MIT's network admins.

> While the routing issue certainly isn't an Emacs problem, it would be
> nice if Emacs would notice the issue and quickly fall back to IPv4, as
> per the recommendations in RFC 6555.  I assume IPv4 works for you?
> i.e. telnet -4 dx.doi.org 80

telnet -4 works fine indeed. It also works fine in Firefox. Could Emacs use the same tricks as Firefox and fall back?

Clément.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Emacs can't reach dx.doi.org?

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 394 bytes --]

On 03/21/2016 04:24 PM, Simon Leinen wrote:
> There are some IPv6 peering
> issues/disputes that may cause disconnectivity between some parts of
> the IPv6 Internet, maybe you are being hit by this.  I see dx.doi.org
> behind Cogent, who are involved in these disputes.

This was spot on; doi.org has three IPv6 IPs, two of which (behind Cogent) are unreachable from MIT.

Clément.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Emacs can't reach dx.doi.org?

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 371 bytes --]

On 03/21/2016 04:24 PM, Simon Leinen wrote:
> While the routing issue certainly isn't an Emacs problem, it would be
> nice if Emacs would notice the issue and quickly fall back to IPv4, as
> per the recommendations in RFC 6555.

Do you know how firefox does it? If they just do what's in RFC 6555, then I'll open a feature request on Emacs' tracker.

Clément.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]