How can I investigate random outages on my internet connection?

Posted on

Problem :

I have an unlimited 10Gbps consumer broadband fibre connection in Switzerland. I have noticed random loses in connectivity to the internet, both on wifi and wired local connections. What steps can I take to diagnose the issue?

Symptoms:

  • Random losses in connections to external services like netflix, spotify, whatsapp, etc
  • loading of web pages is mostly extremely fast, but sometimes randomly takes more than 10s to connect, but then load in less than 1 sec

All symptoms occur on all clients (windows 10 machine, android phones, LG TV etc) at random times. Its not specific to any external domain.

The issue has been present for many weeks now.

My setup:

  • fibre connection to all in one router / switch / wifi access point provided by my ISP
  • Multiple wifi clients
  • Multiple ethernet clients
  • I run a pi hole DNS & DHCP server for local ad blocking off a raspberry pi 3.
  • The DHCP server on the router is disabled and most clients have reserved internal id addresses

Steps I have taken so far:

  • I have rebooted all devices
  • I have verified that the router and raspberry pi are running within operational limits (eg, cpu, temps, ram etc)

Tools:

  • I have reasonable knowledge of networking tools
  • I have only limited capability to diagnose the router as it is a locked down firmware. However, it does have some basic diagnostic functions. I can run repeating pings to external sites, and DNS lookups

Solution :

Sounds like you may have intermittent packet loss. To check this, you need a permanent monitoring solution, like SmokePing. You could target Google DNS and Cloudflare DNS with this monitoring system. They are highly available. It’ll look somewhat like this:

SmokePing graph

(As you can see, I have quite some packet loss to Google DNS every now and then.)

You can run this software on your Raspberry Pi. The system running this software should have a wired network connection.

Once you have identified packet loss, you could then pester your ISP about it or just accept that, unless you have an enterprise SLA, this just happens.

In the end I was able to pins this down to two things:

Poor wifi performance of the router – fixed by adding an additional access point.

A bad cable connecting the PC to the router – fixed by installing a much shorter CAT7 cable.

Sometimes it pays to replace hardware starting with the cheapest items first.

Leave a Reply

Your email address will not be published. Required fields are marked *