Problems I've Solved

After Odysseus became unavailable following a restart, devices around the house appeared to lose internet access.

I tested the local router, external IP connectivity and DNS separately rather than assuming the broadband connection had failed.

The internet connection itself was healthy, but devices were configured to use AdGuard running on Odysseus for DNS. When the server was unavailable, clients could no longer reliably resolve domain names, which made normal internet services appear offline.

This highlighted that Odysseus had become a DNS dependency and exposed a single point of failure in the network design.

On a later incident, Wi-Fi and internet connectivity were both healthy, but websites were taking several seconds to begin loading.

I compared DNS response times and found queries through AdGuard on Odysseus were taking roughly six to eight seconds, while direct queries to Google DNS completed in around a tenth of a second.

AdGuard logs showed repeated timeouts to the configured Quad9 DNS-over-HTTPS upstream.

Because Quad9 was the only configured upstream resolver, the delay affected every device using AdGuard.

I replaced the single upstream dependency with multiple healthy resolvers, improving both performance and resilience.

When something breaks, I try to isolate the problem rather than changing multiple things at once. I test each layer separately, compare expected behaviour against what I am actually seeing, and use logs and measurements to narrow down the cause.

I document the issue, the evidence I found and the change that fixed it. This page is a record of those real troubleshooting situations as the homelab continues to develop.