~5s disconnects ruining online games, seems to not be a fault on our end

[Bucket3456]

For months we have been experiencing random disconnects when gaming. It does not matter the game, nor the PC, they happen seemingly at random and last 4 to 5 seconds each time, then service resumes as normal. This is increasingly frustrating as it has not improved and continues to happen quite frequently. This wouldn't be a problem for e.g. watching TV streaming services that maintain a buffer of upcoming video, but for real time gaming this completely ruins any game it happens during.

To investigate whether this was the fault of my PC, my router, or of somewhere in the TalkTalk connection to us, I ran continual pings on two PCs, each pinging both the router and a Google DNS server. This is in the attached image, for clearer understanding. Both of these PCs are hardwired with ethernet to the router.

You can clearly see on the ping graphs where I have marked with red lines, this is where the game disconnection was experienced. The graphs are stretched differently as one PC had them full-screened, the other had them beside a game window, but both are the same 1 minute duration. During this time, neither PC experienced any noticeable increase in ping to the router itself, suggesting the router was keeping up just fine and not at fault. However, during this time both PCs experienced a large increasing ping spike to the Google DNS server 8.8.8.8. After a few seconds (always somewhere between 4 and 5 seconds) that ping spike dropped back to normal, at the same time the game manages to reconnect. I am unsure if this ping spike increasing over time rather than being a solid block of disconnect was because of the `gping` tool I used doing some short term averaging of ping, or because it actually was slowly increasing in latency, but regardless the experience in games is a hard disconnect every time.

From this experimentation I'm relatively confident it isn't a fault on our end, but if anyone has any suggestions as to what might be causing this (or if TalkTalk could take a look into this as it has been happening for months now) I would be very grateful!

[Bucket3456]

(3rd reply to same comment)

With a suspicion it was somehow a scheduled event, I sat in wait with the ping graphs and a clock on screen. Sure enough, 5:53pm rolls around, and the disconnect happens with the same behaviour again. As I've noticed it at both 5:53pm and 11:53pm, I can't be sure there aren't other times too that I just haven't needed a real time connection during, but at least these two times seem to be consistently a problem.

Attaching screenshot from this most recent occurrence.

Edit: and waited for it this evening too, 11:53pm, it happens again.

[Bucket3456]

Second reply to same comment - this is now having changed DNS settings on my network adaptor to 1.1.1.1 and re-trying to ping graphs

Got another disconnect at ~5:53pm and again, router ping was unaffected during the spike (did increase a few ms as the spike ended but still sub-5ms while the DNS pings are up to 145ms during the spike according to the 'max' readout at the top, and also didn't even do this during the other disconnects, so is not the cause), while both DNS ping graphs showed the same large spike for 5s as they did before.

I was a little later screenshotting this one but it still clearly shows the same behaviour after changing my DNS setting over to the other option.

What's the next step from here?

Thanks

EDIT: It's just happened again at 11:53pm. There seems to be a pattern to the times it happens, at least for the ones I've checked the time for (only the last few). They've each been ~53 minutes past, one past 5pm and the other two past 11pm. Will make a note if I notice more about the time pattern. Still looking for further advice before then!

[Bucket3456]

Have just had it happen at 11:53pm GMT while pinging the router, AND 1.1.1.1, AND 8.8.8.8, with DNS settings still set to 8.8.8.8.

BOTH 1.1.1.1 and 8.8.8.8 experienced the huge spike, and again the game disconnected for 5 seconds, and the router ping experienced no dip. Please see attached screenshot (router, 8.8.8.8, 1.1.1.1 top to bottom, as shown in the top left of each CMD window)

Will changing my DNS settings to 1.1.1.1 and observing latency during another disconnect still be beneficial given both DNS IPs got the spike?

[Karl-TalkTalk]

Hi

I would try a ping to 1.1.1.1 first and see if the results are different, if still the same, then try changing to 1.1.1.1 and ping this again to see if they stay the same.  I do use google DNS myself, but some say that cloudfare is slightly quicker.

Karl.

[Bucket3456]

I think my question was missed in the middle of my reply

Would you suggest keeping my DNS set to 8.8.8.8 but pinging 1.1.1.1 instead, or changing DNS to 1.1.1.1 and still pinging 8.8.8.8?

[Michelle-TalkTalk]

Morning,

Ok, please let us know how you get on.

Thanks

Michelle

[Bucket3456]

Hi Karl, thanks for the response!

I can do some experimenting but due to the random occurrence of the spikes I can't reproduce it immediately and will get back when I've had the time to try!

I believe the devices are set to use 8.8.8.8 as their actual DNS currently - would you suggest keeping that but pinging 1.1.1.1 instead, or changing DNS to 1.1.1.1 and still pinging 8.8.8.8? I suppose both might provide some more information but would have to get at least two random spikes while recording the ping graphs

Thanks

[Karl-TalkTalk]

Hi

With you mentioning the google DNS server, are you only directing your ping traffic to this, or do you have google DNS set on your devices.  Do you see the same results when using say 1.1.1.1 (cloudfare DNS) etc.

Thanks  

Karl.