What did I do in November

Well, I am going to be honest. I do not remember much right now, so I will try to recall things as I write.

I did a lot of pair programming to fix production issues, and I changed a webhook implementation that receives a lot of data periodically from a third party. It is a lot of data. I was also reminded of how important retries are, and how you have to adjust for deployment gaps. Some events might get missed, so they need to be replayed.

I handled some DevOps work around deploying a new service on our Kubernetes cluster. We use ArgoCD. Infrastructure as YAML still scares me a little, but it is manageable. The world is changing. I spent most of my time building scripts to automate sending data exports to clients and fixing production bugs.

We had made many changes directly on production, so we merged the production branch back into the develop branch. There were a lot of conflicts and I resolved them. We made many logical changes across the codebase to make it more stable. I also changed the logic behind one of our sync mechanisms. We sync data from a source of truth table, aggregate it, and display it to the user.

I need to improve modularity and code reuse. When two or three people work on similar tasks it becomes very difficult. When you are creating a new project, you should first add some basic functions that handle the core work of the project. I also want to solve more difficult technical problems. But I am afraid of issues that happen in distributed systems because they are so difficult to trace, and the fixes are usually trade offs. Maybe I should try working on developer tooling or even start my own company (ehehe just kidding). I do not have the technical expertise to get employed at companies that build databases or compilers or similar systems. I would love to write code that powers a satellite someday.

I have become the old developer who writes specification documents for everything. I list out the classes, the functions, and the logic flow. I think it also helps the LLM understand the problem. I have always been too eager to start writing code. I think I need to slow down and think more. Writing a spec document helps with that.

What else did I do? I also updated by my website. Well, not the resume, but the website. It looks cool now.

I am not sure if any of this directly generates revenue. Building revenue generating features is fun because the impact is measurable. I think I will be working on more production issues now. Dealing with customer issues, finding root causes, fixing them, and doing hot patches.

The absence of technical metrics hurts. We should have a dashboard where we can see which API was hit, how many times, and the failure rate. We are working on metrics but they are not on production yet. Since we are moving everything to Kubernetes, we do not want to spend time adding metrics to our current production servers. The production server I ssh into is huge, with thirty two cores and sixty four gigabytes of RAM. Everything runs so fast on it. I love it.

I get bored easily if I do not have interesting things to do, but I also know that the most redundant tasks generate the most value sometimes. I would also like to add Black formatter in our codebase because the code is very unstructured. Some cleanliness would help.

I was in a lot of meetings, which I usually do not prefer. They drain my social energy. Most calls can easily be emails.

A big trade off I have observed is deciding whether information is needed urgently or whether a delay is acceptable. Adding redundancy never hurts. Losing events can be a headache. Also, not all events are important. We are creating a new environment (like staging env) and that is fun in a way. Maybe not entirely fun, but still fun. It is all YAML. I do not like unstructured DevOps. Most use cases do not need Kubernetes. Basic automation is more than enough for ninety nine percent of scenarios.

I also audited the system for data discrepancies. We found some, which were fixed by the user. Then we found a few more. It is okay. What else? Let me think. I did a lot of small things that I do not fully remember. I helped my teammates remove blockers and handled deployments. November was mostly production issues.

I’d have written this earlier but damn viral cold. As I was writing this Cloudflare had an outage lmao. I was also thinking about developing a solution to minimise the lost baggage claims, but I have to deep dive into why it happens and everything. I dunno if there can be a software fix for this. Anyway, that’s it. Bye. Tata.