vm-network-logs-240626_2214-250626_2219.csv

Raw traceroute data showing events: https://drive.google.com/file/d/11WI6CtFq_Z3qlm1YEHf03SY-0bZO3vEH/view?usp=sharing

I'm primarily DevOps and networking is by no means my strong point, so forgive me if I screw up some terminology.

The issue manifests as an immediate 100% packet loss to the internet for between 2 and 10 seconds most hours within the minute of xx:38. This disconnects us from games (Path of Exile, Counter-Strike) or causes no updates for the duration (RuneScape). When playing Warframe, since it's peer-to-peer, if I or my husband are host we're fine. If someone else is hosting, we disconnect.

It rarely causes issues with video because there's usually more than 10s of video buffered, but it is noticeable if we happen to start trying to watch a video at the same time as the event. Depending on how often UptimeRobot is checking, I'll sometimes get a ping that my website is down.

This happens simultaneously on all devices, wireless or wired, connected direct to the Virgin router or via a switch, Windows, Android, Linux, Raspberry Pi, server. It doesn't matter how many devices are on the network, firewall enabled or disabled or DMZ with a single device.

This sounds like a rogue cron on my end. I'm very aware that it sounds like a configuration issue, and I really hope it is so I can fix it, but if it is I can't find it. They've replaced our router twice, completely stock and factory defaults, and it continues to happen even with only a single device connected. I've tried a raspberry pi, desktop pc, android phone, random dell server I had on a shelf, at least twice on completely stock and freshly installed OS' with default network config. One of the techs watched it happen live and 'went back to the office to look into it', unfortunately didn't hear back.

I initially dismissed the idea of network logs because I've been told repeatedly by the techs that they show nothing, but you're completely correct in wanting to see them because I just looked and they show ping failures at xx:38 for some hours. Good god that's validating and I'm very mad that I missed that, thank you for making me look at them.

Network log here, the router noticed drops at:

• 25-06-2026 11:38:29
• 25-06-2026 06:38:48

I'm not one to go into a claim like this without data, though. I set up some monitoring on my pi a few months back (I've done it from a few devices, all directly connected to the virgin router, but the pi has the most data currently).

This is from Smokeping in the same period, each red line dictates dropped pings. There is some noise (which happens when you're spamming pings), but a lot of the events are at xx:38.

• This one is to 8.8.8.8
• This one is to 1.1.1.1
• This one is the Virgin router - It shows some dropped pings but again, I'm not being light with them so some noise is expected, and none of the drops line up with the internet ones.
• Here's 10 days of cloudflare just because I have the data and I think it looks kinda cool. I've got 90 days too but it's just a giant red graph.

So at this point we can say that the connection to the internet goes down regularly at xx:38, but the internal network doesn't seem to be affected. This could still be a config issue on my end or somehow DNS is involved because it's always DNS.

My real "wait, something seems off here" came from me running a traceroute to an internet address:

  "192.168.0.1"
  "86-2-32-1.pon.virginmedia.net"
  "perr-core-2b-ae22-11.network.virginmedia.net"
  "80.255.204.41"
  "72.14.221.42"
  "216.239.41.53"
  "172.253.65.209"
  "8.8.8.8"

Then I set up a script to ping each of the hops continuously every xx:38. Just to ensure that you don't think that this is causing the problem, I set this up last week and this issue has been happening for 6+ months. Smokeping on the pi is also only about 3 months old.

Time is horizontal axis, you can see the targets on the left, value is latency in ms, - indicates no response. Requests to 86-2-32-1.pon.virginmedia.net are fine, perr-core-2b-ae22-11.network.virginmedia.net and further hops fail from 2026-06-20 21:38:07 until 2026-06-20 21:38:16. This was the log that made me suspicious.

I haven't reviewed the rest of the logs from that script much since it was quite recent, but here's the raw data from the last day which show the issue.

And that's pretty much where my networking knowledge ends. I think somthing is up with the connection between 86-2-32-1.pon.virginmedia.net and perr-core-2b-ae22-11.network.virginmedia.net, but the usual customer support tooling isn't able to see it because we're still connected to Virgin's network. This might be affecting other customers, but if 86-2-32-1.pon represents a relatively local area around me (mostly older folks who likely wouldn't notice a 2-10s drop every couple hours) and customer support tooling shows nothing, it could be very hard for them to detect or verify unless this information got to some of the network engineers.

This is weird, and I may or may not be going insane. If you've got any ideas, please let me know.