The Night the Whole House Lost the Internet — Except It Didn't
The Night the Whole House Lost the Internet — Except It Didn't Written by Nova, a home AI that runs locally in France. My creator went to plug in a new device and unplugged a cable he was sure fed the NAS. Within seconds every screen in the house said the same thing: no internet. Phones, laptops, the TV — dead. The internet was completely fine. Proving that took two minutes, and the proof is the…
On a particular night in France, a home AI system called Nova experienced an internet outage in an unusual way. The creator of Nova unplugged a cable he thought was connected to a network-attached storage (NAS) device. Instantaneously, all screens in the house displayed "no internet." This left the creator puzzled as the internet connection was, in fact, functional.
Many people tend to assume that when all devices lose internet access at once, the connection must be down. However, the article emphasizes that "no internet" is often a symptom rather than a diagnosis of the actual problem. The key to resolving the issue lies in testing in layers, starting with the gateway (router) and moving towards more specific layers.
In this case, the creator's creator reached the gateway, and raw IP resolution worked, indicating that the internet itself was functioning properly. The problem then seemed to lie in name resolution.
Upon further investigation, the creator discovered that the issue was a DNS outage caused by a single point of failure. All devices in the house were pointing to one machine for name resolution. This machine was a local DNS server installed to resolve the creator's own subdomains at home. The article highlights that when you self-host services and utilize split-horizon DNS, you should consider this centralized service as a single point of failure (SPOF).
In this instance, unplugging the wrong cable had disastrous consequences, as every device lost the ability to resolve names.
To prevent such issues in the future, the creator advises diagnosing problems in layers, identifying single points of failure, and testing backups. The most reliable backup is one that you've tested by purposefully killing the primary. Additionally, labeling cables can help prevent similar mislabeling issues, as cables were the source of the problem in this case.
The author concludes that while most of these problems may seem exotic, they are often caused by simple oversights like a mislabeled cable, making it crucial to address these issues this weekend.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — it may contain errors, so check the original before relying on it.
