Some websites have stopped loading

[#1]

Since yesterday, suddenly some websites started to fail to load, whilst other websites load OK.
In Safari the blue progress bar starts to move and then halts. Nothing is ever displayed.
In Firefox, the “Connecting…” message appears, but nothing ever happens.
This is happening in Firefox and Safari on all our computers.
Clearing browser caches makes no difference.
Rebooting the router makes no difference.
If I connect to a VPN, then all websites load OK.
Overriding the default Plusnet DNS servers to be GoogleDNS or OpenDNS (on a computer locally, or on the router) makes no difference.

Example websites that fail to load:
http://www.rightmove.co.uk
http://www.thompson-morgan.com

Does anybody have any ideas as to what to try next to get it to work, or what the cause might be?
Given that changing the DNS server doesn’t fix it, but going via a VPN does fix it, does that imply there is some problem with PlusNet’s connections to those websites? If so, then there’s nothing I can do to fix it, except report it to PlusNet.

[#2]

Both those sites load ok for me using a PN connection. Not much help but it least it shows it isn't a Plusnet-wide issue.

I'm using Google DNS servers:

8.8.8.8

8.8.4.4

[#3]

Both of those links opened for me. Using Google and Hub1, so not a PN fault.

[#4]

Are you experiencing any other issues with your connection? I ask as you seem to have disconnected quite frequently this afternoon/evening:

[#5]

I've rebooted the router 2 times today. I've also changed various settings on the router (DNS etc) so maybe that caused a reconnection (but I don't think those setting do cause a reconnection).

I did change the MTU setting about 5 times (to try and get traceroute to give more info) - I wonder if this causes a reconnection? The MTU is back at the original setting now: 1458.

[#6]

Which router is this? What OS is on your PC? If when this happens if you try to ping the failing URL what address is returned?

[#7]

Router: NetGear DG834pn

OS: OS X 10.7.5, OS X 10.11.6, iPhone with IOS 10.1.1 using wifi. The sites don't load on all these.

Sites that don't load:  rightmove.co.uk www.thompson-morgan.com taxresearch.org.uk

taxresearch.org.uk is probably the best example for testing, because it is a simple site/server with static IP/domain. rightmove.co.uk has some kind of dynamic server allocation (the domain DNS seems to be updated every few minutes!) and www.thompson-morgan.com seems to have 3 IP addresses associated with it.

If I connect via a VPN (using the same DNS server at the VPN), or using mobile data on iPhone, all sites work OK.

Ping:

    rightmove.co.uk and www.thompson-morgan.com always timeout. (Probably the servers don't respond to ping?)

    taxresearch.org.uk pings correctly. 

Traceroute:

traceroute to rightmove.co.uk (146.101.19.102), 64 hops max, 72 byte packets

 1  192.168.0.1 (192.168.0.1)  1.300 ms  0.778 ms  0.733 ms

 2  lo0.central10.psb-bng02.plus.net (195.166.130.253)  37.278 ms  27.188 ms  26.097 ms

 3  411.be6.psb-ir01.plus.net (84.93.253.107)  26.692 ms  26.390 ms  26.774 ms

 4  195.99.125.134 (195.99.125.134)  26.527 ms  26.345 ms  26.272 ms

 5  peer1-et-2-1-0.telehouse.ukcore.bt.net (62.172.103.25)  27.128 ms  26.875 ms  26.829 ms

 6  * * *

 7  146.101.1.61 (146.101.1.61)  27.537 ms  27.040 ms  26.682 ms

 8  146.101.1.62 (146.101.1.62)  30.538 ms  28.795 ms  27.375 ms

 9  * * *

10  * * *

11  * * *

12  * * *

13  * * *

Never completed after 10 minutes

traceroute: Warning: www.thompson-morgan.com has multiple addresses; using 34.250.52.249

traceroute to prod-t-and-m-lb-37216428.eu-west-1.elb.amazonaws.com (34.250.52.249), 64 hops max, 72 byte packets

 1  192.168.0.1 (192.168.0.1)  1.229 ms  0.810 ms  0.758 ms

 2  lo0.central10.psb-bng02.plus.net (195.166.130.253)  26.916 ms  26.505 ms  25.768 ms

 3  411.be6.psb-ir02.plus.net (84.93.253.111)  25.933 ms  27.049 ms  26.156 ms

 4  195.99.125.138 (195.99.125.138)  26.117 ms  26.627 ms  26.286 ms

 5  peer1-hu0-9-0-2.slough.ukcore.bt.net (62.172.103.153)  28.935 ms  28.093 ms  28.362 ms

 6  109.159.253.105 (109.159.253.105)  27.940 ms  28.186 ms  28.325 ms

 7  * * *

 8  * * *

 9  * * *

10  * * *

11  * * *

12  * * *

13  * * *

14  178.236.0.109 (178.236.0.109)  38.949 ms  36.920 ms  37.483 ms

15  * * *

16  * * *

Never completed

traceroute to taxresearch.org.uk (217.160.223.242), 64 hops max, 72 byte packets

 1  192.168.0.1 (192.168.0.1)  1.320 ms  0.833 ms  0.743 ms

 2  lo0.central10.psb-bng02.plus.net (195.166.130.253)  26.071 ms  32.482 ms  27.408 ms

 3  411.be6.psb-ir01.plus.net (84.93.253.107)  26.257 ms  26.257 ms  26.571 ms

 4  195.99.125.134 (195.99.125.134)  26.462 ms  26.536 ms  26.269 ms

 5  peer1-xe0-1-1.faraday.ukcore.bt.net (213.121.193.205)  26.735 ms  27.126 ms  26.475 ms

 6  * * *

 7  ae-0-0.bb-a.ba.slo.gb.oneandone.net (212.227.120.105)  28.180 ms  27.328 ms  27.204 ms

 8  ae-11-0.bb-a.fra3.fra.de.oneandone.net (212.227.120.154)  40.440 ms  39.490 ms  39.815 ms

 9  ae-1.bb-c.act.fra.de.oneandone.net (212.227.120.90)  40.273 ms  77.636 ms  40.835 ms

10  ae-11.bb-c.bs.kae.de.oneandone.net (212.227.120.18)  51.230 ms  49.082 ms  49.943 ms

11  kundenserver.de (217.160.223.242)  42.301 ms  42.309 ms  42.039 ms

It arrived at the IP of the domain!

[#8]

OK, router and ping results rule out one known issue which gives the symptoms described.

[#9]

Testing your line I'm seeing the signal to noise ratio at 3db which can cause a variety of problems.

Once our suppliers finish the maintenance they are currently performing on their systems, I'll process a DLM change to move the target to 6db to see if that helps. Should be around 5pm.

[#10]

It doesn't seem likely that the target SNRM, or anything else affecting the DSL, will affect some websites but not others.

Perhaps check the router's log to see if there are any log entries corresponding to the times you attempt to access the websites that don't load.

[#11]

@ejs,

Do the incomplete tracert results tell us anything ... such as problems with routing / transit peers?

[#12]

Not really, because I can access the sites and the traceroutes look much the same.

[#13]

I have found out that the problem is something weird on the router, and managed to get things working, but not sure why yet!

The router settings hadn't been changed for years before the problems started, but I went through all of the settings on the router to see if I could find anything that would change the problem. 

There were 2 outbound firewall rules for http and https, which were set up to ALLOW all LAN addresses to connect to all WAN addresses, (i.e. they should ALLOW all web traffic out, as if the rules weren't defined). If I disable these rules, everything works OK.

These rules were created years ago and have been turned on for years, so no idea as to why they suddenly started causing a problem.

I tried deleting the rules and recreating them, but that still causes the problem.

These rules are not required (I originally created them years ago, when I sometimes set them to DENY outbound web access, but since then they have been left enabled but set to ALLOW outbound web access). As they are currently set to ALLOW outbound traffic, they are redundant, so disabling them is an OK solution.

Why this only recently started to cause a problem and why only on some sites is a mystery.

Maybe there's been some difference to the infrastructure on the routes to the different websites that has recently changed, that causes a problem with this router (IPv6 or IPv4 handling?).

Or maybe the router is starting to fail or some internal settings/software has been corrupted somehow. I guess I could try to reset the router to the factory settings and recreate everything...

Anyway, it seems I have a fix for the problem, so thanks for all the help!

If I do find out anymore info that explains it better, I'll make another post.

[#14]

Glad it's sorted