The phone buzzed at 2:37 AM on a Thursday in March 2024. In my line of work, that usually means someone ignored a warning label until the last possible second. This time it was a logistics manager for a small forwarding company—about 35 employees, night shift running, and a server room that sounded like a smoke detector with a grudge.
“Our UPS is screaming at us,” she said. “A repair guy came yesterday and said the batteries are bad. He can order replacements in three to five days. We don’t have three days.”
The UPS was an Eaton 9PX that had been in service for nearly six years. The battery health had been declining for months, but nobody acted because the warning signs hadn’t caused an outage. That part is probably familiar to anyone who maintains a small server room. Most Eaton UPS repair calls I go on start exactly this way: the UPS looks like a giant battery box, and by the time it actually fails, the components inside are already toast.
What made this call different was the customer’s size. A large data center would have spare modules in a warehouse. This client had one rack, a few servers, and an old storage array. They couldn’t tolerate downtime because their shipping software and billing system lived in that rack. And they were small enough that I could have pushed them to the end of the queue when a bigger contract called. I’ve handled more than 200 rush jobs over the past three years, so I know how easy it is to prioritize the big fish. But here’s the thing: small clients often need the fastest response, because they don’t have backups, spare parts, or the staff to improvise. So I told her we’d be there in 90 minutes.
When we arrived, the UPS had already switched to bypass mode. The beeping wasn’t the real problem. The battery cabinet was. Batteries were swollen, voltage readings were all over the place, and the UPS couldn’t hold a load through even a small sag. The load stayed up because someone had wired a maintenance bypass switch, but that switch was a three-position unit that I did not trust enough to keep using for the next few hours.
The better tool for this situation is a Pro Tran 2 transfer switch. It’s a small rack-mount switch that lets you move a critical load from the UPS to utility power without opening the circuit. That gives you a safe way to isolate the UPS, work on it, and bring it back online without telling the client to restart everything at 4 AM. We installed the Pro Tran 2 between the UPS and the server load, transferred traffic to utility, and then pulled the battery cabinet for replacement.
I’ll be honest: I had a post-decision doubt moment when we first energized the switch. What if the load exceeded the Pro Tran 2’s rating? What if the old input wiring couldn’t handle it? I checked the numbers twice. I didn’t fully relax until we transferred the load back to the UPS and it held a clean output. That was around 4:30 AM. So glad I had brought the switch.
After everything was stable, I noticed a burnt Lenovo battery charger hanging half off a power strip behind the rack. The plastic around the plug was melted. The client said it had died a few weeks earlier and they’d just bought another one. That’s a clue, not a coincidence. Poor power quality kills chargers, and it eventually kills much more expensive gear too.
I asked about their electrical setup. The building still had a fuse box. That’s not inherently unsafe, but from an emergency response perspective, it makes outages worse: when a fuse blows, you need the exact same rating, and you’re standing in a dark server room at midnight trying to find a spare. I’ll be honest again—this gets into electrical engineering territory, which isn’t my expertise. I’m not an electrician, so I can’t tell you how to replace a fuse panel. What I can tell you from the field is that sites with circuit breakers tend to recover faster, and there’s a reason the industry keeps making the case for circuit breaker vs fuse box decisions as a business continuity point, not just a code compliance point.
We didn’t replace their fuse panel that night, and we didn’t try to. But I flagged it for their electrician, because that burnt Lenovo charger was probably a warning label of its own.
Here’s what I hope you take from this story:
That small logistics company made it through the night. They didn’t miss a single shipment, and the owner later said a failed server day would have cost them about $14,000 in contract penalties. We charged a fraction of that for the emergency repair, and honestly, I felt we were the lucky ones. Not because we made great money, but because we got to remind ourselves why this work matters.
I still wish someone had called us months earlier, when the warning light first appeared. The job would have been a two-hour visit instead of a 5 AM finish. But it was still a good night, and it reminded me that small customer orders deserve the same responsiveness as the big ones. Small doesn’t mean unimportant. It just means there’s less room for error.
If you run a small server room, please don’t wait for the beeping to get loud. Check the UPS, test the batteries, plan for the bypass, and consider the fuses before they blow. You’ll thank yourself at 2 AM.