Intermittent Internet access issues

[amcewen]

This morning the Internet has been down in the space a number of times. Generally only for a couple of minutes each time.

[amcewen]

• @JackiePease reported it to Baltic, but they can't see any issue at their end.
• When it happens I can still talk from one computer to another over the WiFi, so it's the connection out to the wider Internet rather than the local network.
• I've put a spare laptop directly onto the Internet, with the one spare IPv4 address we've got. That's running ping in a screen session to monitor the Internet connection. When there are outages (such as now, when I'm typing this), that is still running fine, so it seems to be something between the Internet and DoES' WiFi
• I had tried rebooting the Unifi box under the Internet router, but we've had another outage since then, so that hasn't fixed things

[johnmckerrell]

Interesting that your other laptop isn't seeing issues. Can you SSH into 10.0.0.2 and run a ping in a screen from there maybe? I'm fairly sure there's no other "things" that could be temporarily having issues that you haven't already tried rebooting. If it was the APs or the switches you wouldn't be able to ping internally. I guess you could try replacing the cable between our router and the gigabit "external" switch but I wouldn't really expect a failing cable to fail so periodically. You could also try rebooting the switches though if you haven't already.

To confirm 10.0.0.2 is running the Unified controller software, it is just a node on the network but it's at least physically closer so might be more reliable for running a ping than a computer on the Wi-Fi.

[amcewen]

Ah, I did have a session on the dashboard for 10.0.0.2 open which then timed out when the network went, so I don't know if that points to that device having issues?

I've tried the usual trick of stealing the power supply from the Ackers Bell to replace the one for the Unifi router - that did have a 12V 1A centre-positive supply; it's currently got a 12V 2A centre-positive supply. We can see if that's causing the instability.

[johnmckerrell]

I've logged into the unifi controller on 10.0.0.2 in the web, I don't see any flashing red lights or anything but don't know that anything would show up there, maybe you can try when the outage happens. I found it a little odd that our biggest media service is something called MiaoPai but I dunno, maybe someone actually is using it a lot?

There are software updates available but again wouldn't expect a sudden period of downtime to be fixed by a software update, seems more like a hardware issue.

Hm... yes odd that your session on 10.0.0.2 went down too, given that's just a node on the network pinging out in the same way as your laptop is. As I understand it that box is just controlling the others but without it they would just carry on as they are. Wouldn't hurt to reboot it/try a different power supply for it if the existing PSU swap doesn't fix things.

[johnmckerrell]

@MatthewCroughan was asking about this recently, has there been continued problems? I was just trying to look on the unifi dashboard to see if there was a device using too much network capacity but I can't get the Traffic Stats tab to load. I thought the interface had crashed but actually all the other tabs load fine. It did make me wonder if the Raspberry Pi running the controller is struggling, although I don't know that it really has much of a job to do, depends if it's collecting the traffic information I guess or if it's just managing some config. FWIW it's a Raspberry Pi 2 Model B Rev 1.1 that's been doing the job since we moved in 7 years ago so might not hurt to upgrade it (although if it's not actually broke...!)

If there's no persistent issues though we should close this issue?

[amcewen]

There haven't been any reported outages of late, so I'm closing this.