Is the Internet grinding to a halt?

Do we have new conspiracy theory to worry about?

Like most SysAdmins, I try to be aware of performance issues, and to fix them before they become a real problem. Around here — my home lab and network — my spouse usually reports the problems to me while I’m working on a project or two. So I guess I’m not especially successful at that.

In this case she began telling me that accessing web sites was taking a long time. So then I started paying attention and she was right. Don’t tell her I said that, please!

I started digging into the problem, at first thinking it was something in my home network.

Years ago it started out to be a simple network but has exploded into 1, 2, or 4 wired connections in most rooms of my house, multiple WiFi routers, multiple switches, a dozen or so computers, and of course, the server that hosts this website and my own name services service, and the system that acts as my firewall and router which connects my two internal networks to the Internet. So there are plenty of places for problems to arise and most of the ones that do tend to be self-inflicted in one way or another.

A couple nights ago, everything external came to a grinding halt as she was watching UNC basketball. That’s unacceptable so she insisted I fix it — “NOW!!”

So I started looking internally. I checked my DNS server. I found a few minor issues that needed fixing, like deleting entries for systems no longer present, or adding an entry to the reverse-lookup zone. I found nothing that would cause a problem. Lookups for my internal hostnames were only a few milliseconds at most. Yet most lookups for external domains were a few seconds or more and some would time out without returning the requested IP address. I used both nslookup and dig, the latter of which reports the time it took to receive the response, for this testing.

I also used ping and mtr directly to a known IP address for example.com so it didn’t need to do a name lookup. That worked splendidly. So its definitely a name services problem — outside my network.

As the outage continued, I called AT&T, provider of my Gb fiber-optic connection. The (so-called) AI that answered ran some tests that didn’t finish properly. It started to tell me to reboot my computer, so I started saying “Real person” and got switched to a person who was reasonably knowledgeable. Yes, that works.

After a bit of discussion, the agent started some tests, and — naturally — everything started working perfectly including DNS. So the agent told me everything was working and we parted ways.

Everything worked for the rest of the game so I was off the hook. For now.

A couple nights later, we were streaming one of our favorite British TV shows, and started experiencing short interruptions of the service. There was still one thing I have control over, and that’s the AT&T router in my house. I rebooted that and, after the wait for it, everything now works well. Much better, in fact, than it has in weeks. And it’s stable. At least until the next time.

All is currently working well. I hope I can remember to reboot the AT&T router if/when it happens again. That’s why I’m writing this, because most of the articles I post here are memory aids. You know, when you know you’ve seen something before but can’t quite put your finger on it and wish you’d written it down, especially the part about how you fixed it.

So the answer is — No — there’s not a conspiracy. It was just a router that had been running for months until I rebooted it.

But it still leaves a question. Why does a router work fine for raw IP addresses, but fail to connect to an external name server? And how do I deal with that?