Intermittent connection issues loading/changing site- secure connection failed/site can't be reached

[EDIflyer]

Is anyone else occasionally getting a 'connection down' error from Chrome (or 'Secure Connection Failed' from Firefox) when changing page - if I hit refresh it works fine. At first I wondered if it was the site I was using but have noticed it on multiple different major websites over the past couple of months and not noticed it from other locations where I don't use Three Broadband so does seem to be related to the router/connection (this is via wired Ethernet too, so not a WiFi problem).  I've tried changing DNS server in case that helps but no difference.

It mainly seems to happen when trying to first load a site or (annoyingly) at checkout when a different site is being loaded as part of the checkout process.  It certainly doesn't happen everytime but does happen with reasonable frequency.  I've also noticed I often get it when trying to pull/push from/to Github too and have to do so a number of times for it to work (browsing the Github website works fine).

I'm using a NR5103E with Firmware Version V1.00(ACBJ.0)b14 - I tried rebooting it but to no effect.

[applepieman]

This really is quite disappointing I have to say and generic technical guidance really isn't the answer for customers. 

[BrummyGit]

I got fed up with the problems as I need a reliable and fast connection, therefore decided to try an expensive fix. I bought myself a Zyxel FWA710 5G outdoor router and mounted it in my loft next to the Three supplied NR5103E, then swapped my SIM over.

For the last week running with the FWA710 using the latest OEM firmware and IP Passthrough mode I haven't experienced any issues other than a loss of service overnight at the weekend that I suspect was mast maintenance as it started at exactly 2am and lasted an hour.

[JonathanB]

Hey everyone,

Sorry for the delay in updates. The support teams on our side haven't been reproducing the issue, so it does seem to be quite elusive. We're still digging deeper into what could be the common link for those of you who are affected.

There's a few potential causes in terms of cached data, DNS settings, Firewalls, etc, here's a couple of guides/suggestions for related errors:

https://kinsta.com/knowledgebase/err_connection_closed/

https://kinsta.com/knowledgebase/pr-end-of-file-error/

Please feedback if any of the suggestions in the above articles help with this issue.

I'll let you all know as soon as we have any more to share on our side.

Thanks,
Jonathan

[AlSeek]

Thanks for the update @JonathanB. I've tried all these suggestions but no difference for me. Also tried the DNS flush suggested by @EDIflyer but soon had the usual error on affected websites. TBH I don't think it makes sense to be a local device issue since, as mentioned in the past, switching my router to use 4G only pretty much resolves the issue besides a very rare instance but 5G is extremely badly affected, if it were a local device issue of DNS cache or otherwise I'd expect similar incidence on 4G and 5G. (I have primarily been using 4G this last month besides occasionally going to 5G to check if it's still an issue, which it severely is).

[EDIflyer]

Thanks for the update, @JonathanB .  It's quite disappointing that the engineers haven't been able to replicate it given how many of us has posted re issues.  I had a look at the links and most of them seem very generic and not necessarily applicable as they are referring to Chrome/Firefox when a lot of us have also confirmed the issue on mobile phones (on WiFi), smart TVs, headless Linux devices, etc.

However one of the later solutions was more interesting - re flushing the DNS cache. I followed the more detailed guidance at https://devconnected.com/how-to-flush-dns-cache-on-linux/ as well as the Windows guidance at https://kinsta.com/knowledgebase/flush-dns/ and when running the code from @Tatablack from upthread (slightly adjusted below) I've noted a marked improvement with no errors after 100 curl requests:

for i in {1..100}; do curl -I --no-progress-meter -o /dev/null https://github.com || true; done;

I'm not certain it's 100% sorted yet but I'm going to try running things without the VPN for a bit and see how I get on (although for shop check-outs I might still pop it back on for a bit just in case!)

[tucu]

Tried this, made no difference. It was better for a few days when I switched APNs, but the problem is still there. Interestingly, the error is different for me:

curl: (35) LibreSSL SSL_connect: SSL_ERROR_SYSCALL in connection to github.com:443

The fact that a VPN fixes the issue AND that @BrummyGit still had issues with another router makes me wonder what kind of networking monitoring does Three do 🤔 (I did try disabling the adult filter as suggested in another thread - made no difference either).

[EDIflyer]

Just to update - still seeing it occasionally, albeit much less than before...

for i in {1..100}; do curl -I --no-progress-meter -o /dev/null https://github.com || true; done;
curl: (35) OpenSSL SSL_connect: Connection reset by peer in connection to github.com:443

[EDIflyer]

@JonathanB although improved definitely still not fixed - still seeing intermittent issues when I'm not on VPN.

23-11-29 22:17:04 ERROR (MainThread) [supervisor.supervisor] Update of Supervisor failed: Can't install ghcr.io/home-assistant/armv7-hassio-supervisor:2023.11.6: 500 Server Error for http+docker://localhost/v1.43/images/create?tag=2023.11.6&fromImage=ghcr.io%2Fhome-assistant%2Farmv7-hassio-supervisor&platform=linux%2Farm%2Fv7: Internal Server Error ("Get "https://ghcr.io/v2/": EOF")

[BrummyGit]

I can't see my earlier reply on this topic, but I also had the same issues updating Home Assistant this evening whilst connected to Three Broadband using my FWA710 not NR5103e. I tried repeatedly before swapping to my BT FTTC service where it worked first attempt.

I am not seeing intermittent page loading issues on this new router, but the problem getting content from GitHub was exactly the same for me tonight.

[JonathanB]

Hey @BrummyGit,

Sorry about that, our automated Spam filter must have not liked something about the post. I think it may have seen the text and URLs in the log extract as a bit suspicious. I've restored the post now though.

Thanks,
JonathanB