I kept checking my phone while family was visiting.
It felt rude because it was rude. The problem was that a large part of Shutterstock's server fleet had started behaving badly, and nobody knew why.
We found the answer about half an hour before I sent Amanda the explanation.
A leap second had been added to the world's clocks. The one-second adjustment triggered a software problem across the servers. I described it as the computer version of a solar eclipse: rare, predictable in theory, and still capable of making the day strange when it arrived.
I also sent a graph as proof that I had not been ignoring everybody for no reason.
This was the hard part of being responsible for infrastructure. The systems did not care who was visiting or what the plan had been. When a broad production problem appeared, the phone entered the room with me.
I loved the work. I did not love every place it showed up.
The leap-second bug was at least interesting once we understood it. Before that, it was only a pattern of failures with no obvious cause. The team had to compare symptoms, eliminate the usual explanations, and find the one global event that had changed underneath us.
Then we had to fix it.
The personal apology and the technical explanation ended up in the same message. That feels accurate to the period. Shutterstock was growing fast. My responsibility was growing with it. Amanda and the people around us got an unplanned education in why I sometimes stared at a phone during dinner.
One second is nothing until enough computers disagree with it.
Then it is the whole day.