$ cat post/the-rollback-succeeded-/-that-script-still-runs-somewhere-deep-/-the-secret-rotated.md
the rollback succeeded / that script still runs somewhere deep / the secret rotated
Title: Kubernetes Christmas Carol
December 25, 2017 was a quiet day in the tech world. The trees were lit up outside my window as I sat down to write this post. The container wars seemed like ancient history now, and everyone had settled on Kubernetes as their standard for orchestration. Helm made Kubernetes just a little bit easier to manage, and Istio promised better service meshes. Outside, reindeer probably shuffled around the roof of our building, but inside my office, I was grappling with some real issues.
It’s funny how technology trends can make you feel so old. Just a few months ago, everyone was still trying out Docker Swarm or Mesos. Now it seems like Kubernetes is everywhere and we’re all moving on to the next big thing—whether that’s serverless or whatever flavor of microservices du jour happens to be trendy.
Today, I spent most of my day wrestling with some pesky issues around our Kubernetes cluster. We were hitting a strange deadlock where pods wouldn’t start after they got scheduled by the master node. I’d see the API requests coming in, and then just… nothing. It was like the Kubernetes scheduler had taken over a pod and was holding it hostage.
I spent hours digging through logs, trying different debugging tools. Prometheus was great for monitoring metrics, but Grafana still felt clunky to work with. Sometimes I wished we were back on Nagios; at least that was straightforward. But then again, having everything visualized right in front of me in Grafana made things easier to diagnose.
On top of the technical challenges, there was a lot going on around me. The F.C.C. had just repealed net neutrality rules, and everyone was talking about how this would affect internet access and freedom. Meanwhile, Firefox was facing some rough times, with its market share sliding. I found myself wondering if any of these decisions would even matter in the grand scheme of things.
I also noticed a few articles about machine learning popping up on Hacker News. One particularly interesting slidedeck detailed the journey of getting into machine learning over two years. It felt like the industry was still in its early stages, with everyone scrambling to figure out what worked and what didn’t. Maybe that’s why so many companies were using Facebook for job ads—perhaps they thought it would help them target younger candidates.
As I wrapped up my day, I couldn’t help but think about how much had changed over the past few years. From containerization wars to Kubernetes winning, from serverless hype to people arguing whether microservices are the right path forward. And yet, despite all this change, some things remained constant: debugging, arguing, and learning.
Tomorrow would bring new challenges, but for now, I could sit back and watch the snowflakes dance outside my window. As we say in my family, “Merry Christmas to us and all the world!” Happy holidays, everyone—let’s hope 2018 brings more stability than these pods have been bringing me.
I think that captures a bit of the spirit of the time, with some personal reflections on work and broader tech trends.