Theory Says Redundancy Matters. Practice Teaches It When You Can't Walk Downstairs.

I knew redundancy mattered. On paper. I had read about it. I understood the concept. Knowing and feeling are two different things though. One went down when I wasn't home. The other taught me why I needed it in a way no article ever could.

Within about a week and a half of being away, there was a short power blip on my homelab. My laptop had no UPS yet because I was waiting for the Unifi 2U model to come in stock. That wouldn't have been such a big deal if the laptop still had its battery.

When I first built out my homelab setup, I noticed the battery was swelling enough to affect the trackpad. I pulled it out immediately rather than risk permanent damage or a fire hazard sitting in there indefinitely. That decision saved me from a worse problem later but left the laptop with zero backup power and no way to handle even a brief outage gracefully.

The server didn't recover on its own either. Unlike my repurposed gaming machine running InferDeck, the laptop's BIOS doesn't have automatic reboot built in. When the power came back, nothing started. Thirty-six Docker containers, all offline, waiting for someone to press the power button.

Thankfully I had family nearby that day and they were able to pop in and turn it on. Without that luck, it would have been a six-hour round trip just to restore things manually. I knew redundancy was important before this happened but never grasped how quickly a minor inconvenience becomes a major one when you're not there to fix it.

I've now installed the Unifi UPS 2U and those small blips are handled automatically. It won't cover longer outages, but that's fine for what I need right now. The real lesson here wasn't about buying hardware though. It was about understanding cost in a way that theory never gives you.

When I work from home and the power flickers, I pop downstairs, press the button, and everything's back in thirty seconds. That convenience shapes how you think about these problems. You figure someone will be around to deal with it when it matters. Until someone isn't.

What I'm planning next is a redundant node so services can fail over if the machine itself dies. I also want a smaller secondary machine that auto-reboots on startup to handle the laptop's limitation permanently. The laptop works fine as a server, just not when it needs physical intervention to come back online.

I always nodded along when people talked about redundancy planning. Now I get why they bother. Theory tells you something is important. Experience teaches you how much.