You are not logged in.
I'm trying to switch back to dwb (webkit browser) from firefox. Unfortunately, I have been unable to access these forums from dwb. At first I thought it was an isolated incident, but this has now persisted for at least 12 hours. The same pages are accessed fine from firefox on the same computer.
I just installed luakit (another webkit browser). Luakit is also unable to load any pages at bbs.archlinux.org.
Both webkit browsers are able to access archlinux.org, but just not the bbs. Both browsers are perfectly able to view a local copy of a page from bbs.archlinux.org (downloaded via firefox). In both cases, any means of navigating to a bbs.archlinux.org page progresses to only 10% completion and hangs - none of the page is displayed.
I have cleared the cookies, and removed ~/.local/share/webkit and ~/.cache/webkit, but this lead to no change.
I'm not sure how to gather more diagnostic information - any suggestions would be appreciated. Also, are other webkit users having any issues with the forums? I've found a couple superficially related posts on these forums, but they were from several years ago and seemed specific to javascript-ladden pages - I am unable to load *any* page from the specific domain/server bbs.archlinux.org, but only in webkit browsers.
---- updates:
Both browsers tested use webkitgtk2, so I also tried a gtk3 version using webkitgtk - this fails in the same way.
I also tried surf which I ran from the command line (`surf bbs.archlinux.org`). This produce several lines of output relating to the pipelight plugin then stalled without loading the page. Thinking this might be a piece of the puzzle, I disabled all pipelight plugins (just silverlight5.1 in my case) and tried again. Surf now produces no output and still does not load the page.
Running dwb from a terminal also produces no error output. If silverlight is activated it produces various output upon starting, then I get a line for the "cookies" plugin, then nothing until I kill the browser.
I've also confirmed the same results with vimprobable (vimprobable-git from AUR).
EDIT: Thread title revisions
ORIGINAL: Webkit browsers cannot load bbs.archlinux.org
PREVIOUS: Only firefox can load bbs.archlinux.org
Last edited by Trilby (2014-05-01 03:35:42)
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
Vimprobable using webkitgtk2 2.4.1-1 working with no issues. What version have you got?
Offline
Same version here webkitgtk2 2.4.1-1
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
This is strange. I use dwb and I have no problem here on Arch's forum. Do you use dwb-git from the AUR? surf works for me too.
Offline
I was using dwb from the main repos first, but I've installed dwb-git from the AUR in the process of trying to track this down - same results.
So this seems to be just me ... that narrows the problem down a bit - but not in a useful way
I've just cleared my pacman cache and reinstalled webkitgtk2 - no change. For good measure, I also reinstalled all webkitgtk dependencies (e.g. libsoup) - no change.
I can also rule out any network * browser interactions as I first saw this issue earlier today at work, and it persists at home - it is not specific to a given network connection.
Using the IP address of bbs.archlinux.org instead of the url produces the same results.
Last edited by Trilby (2014-05-01 01:20:47)
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
Weird. You don't have a http_proxy variable? w3m/links etc all work OK?
Offline
WHOA ... well there was a useful question with a surprising answer. Both w3m and lynx show the exact same problem: they can display archlinux.org, but hang indefinitely trying to connect to bbs.archlinux.org.
I have never set up a proxy, and I know next to nothing about them. Conceptually I get it, and I checked out the wikipedia and archwiki pages on them to be sure, but in terms of practical application I'm clueless. `env | grep -i proxy` returns no matches.
As w3m + lynx both failed to connect I also just tried wget. As seen below, I can wget "archlinux.org", but not "bbs.archlinux.org" (I also get the same results with `wget alderaan.archlinux.org`)
$ wget archlinux.org
--2014-04-30 22:07:12-- http://archlinux.org/
Resolving archlinux.org (archlinux.org)... 66.211.214.131
Connecting to archlinux.org (archlinux.org)|66.211.214.131|:80... connected.
HTTP request sent, awaiting response... 301 Moved Permanently
Location: https://www.archlinux.org/ [following]
--2014-04-30 22:07:12-- https://www.archlinux.org/
Resolving www.archlinux.org (www.archlinux.org)... 66.211.214.131
Connecting to www.archlinux.org (www.archlinux.org)|66.211.214.131|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 34457 (34K) [text/html]
Saving to: ‘index.html’
100%[=====================================================================================================================>] 34,457 --.-K/s in 0.1s
2014-04-30 22:07:13 (293 KB/s) - ‘index.html’ saved [34457/34457]
$ wget bbs.archlinux.org
--2014-04-30 22:04:41-- http://bbs.archlinux.org/
Resolving bbs.archlinux.org (bbs.archlinux.org)... 2a01:4f8:120:34c2::2, 78.46.78.247
Connecting to bbs.archlinux.org (bbs.archlinux.org)|2a01:4f8:120:34c2::2|:80... ^C
The latter wget command hangs and I have to break with Ctrl-C.
EDIT: I see the bbs wget command gets both an IPv6 and v4 address, while the working command only gets a IPv4 address. I'm about to test after disabling IPv6. disabling IPv6 works!.
Is there any reason why I'd need to disable IPv6 to reach bbs.archlinux.org, but not any other site?
Last edited by Trilby (2014-05-01 02:23:06)
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
wget gives you the --inet{4,6}-only switch: both also work on my custom and vanilla kernels.
Are you in one of those parts of the State with a monopoly ISP? This is (to me) looking increasingly less like an issue with the BBS or your setup. If your are reliant on $bigisp at home and work, I'd be inclined to blame them.
If not, I'd blame them anyway...
Offline
I suspect it is an ISP issue. At home it is one of the BigISP, and work is a university with a dreadful IT department.
But assuming this is an ISP issue, firefox gets around it just fine - I'm curious how. But more on point, I wonder whether disabling IPv6 is a good long term solution.
EDIT: it seems firefox likely worked as there is an option in about:config "network.http.fast-fallback-to-ipv4" which is set to true. I'd prefer to get webkit to fallback to IPv4 over entirely disabling IPv6 - if for no other reason than if my ISP every gets their act together, I will not have to remember to reenable it.
Last edited by Trilby (2014-05-01 03:07:35)
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
You'd need to wait for fukawi2 or one of the other networking specialists to show up to get a definitive answer on why Firefox gets around it.
As for disabling IPv6, there is no future in that.
# edit: Missed your edit...
Offline
Thanks for the help figuring this out. I never would have thought to check w3m/lynx which led to pinpointing the problem - even if the solution is ugly for now.
I've disabled ipv6 in the bootloader for now, and hope this is a temporary issue with the ISP. They were slow to get on the ipv6 bandwaggon back when it was first rolling out, but I definitely have had working ipv6 connections previously.
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
You'd need to wait for fukawi2 or one of the other networking specialists to show up to get a definitive answer on why Firefox gets around it.
Happy Eyeballs. And I'm not a specialist, just don't have a life
See how far the IPv6 traffic is getting:
mtr -6 bbs.archlinux.org
Are you familiar with our Forum Rules, and How To Ask Questions The Smart Way?
BlueHackers // fscanary // resticctl
Offline
Apparently that traffic is a nonstarter:
Host Loss% Snt Last Avg Best Wrst StDev
It just hangs there, there is not a single entry added to the table (and this is after re-enabling ipv6 in the bootloader and rebooting).
Also, from http://test-ipv6.comcast.net I get the following results
Your IPv4 address on the public internet appears to be XXXXXX
No IPv6 address detected, which is fine. Comcast will deploy IPv6 soon. [tech details]
World IPv6 day is June 8th, 2011. No problems are anticipated for you with this browser, at this location. [tech details]
When a publisher offers both IPv4 and IPv6, your browser appears to be happy to take the IPv4 site without delay.
Connections to IPv6-only sites are timing out. Any web site that is IPv6 only, will appear to be down to you.
Your DNS server (possibly run by your ISP) appears to have IPv6 internet access.
The fifth bullet point has a big red X. And it notes that the browser (firefox in this case) falls back to IPv4, but webkit it seems did not.
The bbs.archlinux.org vs archlinux.org seems to be partial coincidence. The bbs page seems to offer both IPv6 and IPv4, firefox would fallback on IPv4 while webkit would choke on the v6. But at archlinux.org only IPv4 is offered, so both browsers work. I've confirmed this pattern on other sites that work and/or fail.
Last edited by Trilby (2014-05-01 03:35:09)
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
So you're directly connected to the Comcast modem?
If you re-enabled IPv6, what is the value of /proc/sys/net/ipv6/conf/all/accept_ra? And what is the output of `ip -6 r s`?
Try stopping your kernel accepting Router Advertisements by changing it to 0.
Are you familiar with our Forum Rules, and How To Ask Questions The Smart Way?
BlueHackers // fscanary // resticctl
Offline
Yes, I'm connected via ethernet cable directly to the router. Though this problem was initially noticed at work where I connect wirelessly to a different network - however all information in this thread has been gathered from the home Comcast connection.
$ cat /proc/sys/net/ipv6/conf/all/accept_ra
1
$ ip -6 r s
2001:558:6017:50:b5e5:3770:48eb:6b1f dev enp0s25 proto kernel metric 256 expires 345539sec
fe80::/64 dev enp0s25 proto kernel metric 256
default via fe80::21d:45ff:fe70:4ee2 dev enp0s25 proto ra metric 1024 expires 1797sec
There was no change in behavior after changing accept_ra to 0, and the second command above provided identical output.
Last edited by Trilby (2014-05-01 11:30:24)
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
Interesting -- it seems something on your network has started advertising IPv6. Try installing ndisc6 and looking for the mac address of that router in the default route:
ndisc6 fe80::21d:45ff:fe70:4ee2 enp0s25
Are you familiar with our Forum Rules, and How To Ask Questions The Smart Way?
BlueHackers // fscanary // resticctl
Offline
$ ndisc6 fe80::21d:45ff:fe70:4ee2 enp0s25
Soliciting fe80::21d:45ff:fe70:4ee2 (fe80::21d:45ff:fe70:4ee2) on enp0s25...
Target link-layer address: 00:1D:45:70:4E:E2
from fe80::21d:45ff:fe70:4ee2
I'm not sure how to interpret that output, but there should be nothing else on this network (on my end at least). I have a comcast co-ax cable line come into this room and no other room of my home. This cable goes to the modem which is connected via ethernet cable to my computer. There are no other computers in my home.
From skimming the ndisc6 man page, I'm under the impression that 00:1D:45:70:4E:32 is the mac address of a discovered "neighbor". I'm checking my modem and ethernet card now, but neither of them seem to match that.
Last edited by Trilby (2014-05-01 12:09:23)
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
I'm currently out and on a public wifi with the same computer and can confirm that all sites are perfectly reachable with all browsers. So this is specific to my ISP (and perhaps my university's network).
So I'm perfectly content considering this a fully solved and closed issue - many thanks to JWR and Fukawi. Fukawi, if there are other productive ways to dig a little deeper, I'd definitely be interested in learning more - either for curiosity's sake or to get details to report to my ISP if that seems to be the right path. But again, I'm content with having a workaround that consistently works (disabling ipv6 at home).
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
But again, I'm content with having a workaround that consistently works (disabling ipv6 at home).
Understood, I was only digging deeper for curiosity's sake
I'm guessing 00:1D:45:70:4E:32 is your Comcast modem/router then? I don't know how it works in the states; do Comcast manage that device for you? It seems strange that it would start advertising IPv6 if you didn't tell it to. FWIW, after changing accept_ra earlier, than will only stop a new address being assigned via RA's; the existing one would stay until it expired. Setting accept_ra to 0 in sysctl.conf might be less drastic than disabling IPv6 altogether and allow you to easily enable IPv6 if (when) you need it
Are you familiar with our Forum Rules, and How To Ask Questions The Smart Way?
BlueHackers // fscanary // resticctl
Offline
To answer some of the questions: yes comcast manages the device. It doesn't surprise me that they'd change something like that. The "grr" thread is filled with stories of what the couple big ISPs around here regularly do that is absolutely insane. But the service is pretty reliable, as long as we tolerate their whims.
I've actually opted for leaving ipv6 on, but I have a shell function to toggle it as needed. The function case a case statement for "on, off" and other handy mnemonics, but essentially it just runs `sudo sysctl net.ipv6.conf.all.disable_ipv6=1` or with the 1 as 0.
This is a laptop I carry around, so any time I connect to my home network I have to physically plug in the ethernet cord on the side - so running a single simle command is no inconvenience. In fact, I have another shell function called "ether" as I so rarely use ethernet - so when I connect at home "ether" runs dhcpcd, starts up offlineimap, and some other little bits. So I can just add the sysctl command to that function and it will be seamless - this way ipv6 is only down when I connect to this one network.
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
Ah I see, when you said you disabled it in the kernel I thought you meant at boot time in your bootloader
So strange to hear the ISP manages your CPE -- over here you buy whatever modem you want (usually a "supported" one from the ISP) but configuring and managing it is generally your responsibility. Even our business connection at work I had to ask nicely to get them to configure the Cisco 887 before they sent it out.
I guess if they're playing funny buggers with it, there's not much you can do except act in self-defence and wait for them to figure out WTF they're doing.
Are you familiar with our Forum Rules, and How To Ask Questions The Smart Way?
BlueHackers // fscanary // resticctl
Offline
I guess if they're playing funny buggers with it, there's not much you can do except act in self-defence and wait for them to figure out WTF they're doing.
Gotta love that sunny optimism!
Offline
... I thought you meant at boot time in your bootloader
Sorry for the confusion - I *did* breifly put this in my bootloader. But then I remembered the sysctl method which I had to use on *all* networks around here when IPv6 first "went live." So I removed it from my bootloader and just used the sysctl toggle.
... wait for them to figure out WTF they're doing.
I may be waiting for a very long time.
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
Trilby, only for register effect i would to say i have same problem here.
My IPS and my ADSL router/modem don´t accept ipv6 and make same problems to me, like you have reported.
Offline