Hacker News Reader

Hacker News

Comment by ingvul | Hacker News Reader

Full thread

ingvul

I do think I'm experiencing burn out, but I don't see myself quitting. It's weird.

I feel fine when I have to:

- refactor code

- create some new application code from scratch

- fix some business logic bug

I feel anxiety (and I experience burn out symptoms) when I have to:

- deal with any k8s issue

- incorporate any third-party library like oauth2-proxy

- run some migration in a big table

I don't like "infrastructure" topics, but as a "senior" software engineer I don't have a choice. I love dealing only with application code (e.g., classes, interfaces, modules, business logic, tests, etc.)

View on HN

Do you feel like you don't like the infrastructure side of things because it's harder to debug problems while being under immense pressure since this usually happens when production is on fire?

Developing code isn't easy but if it's code you've written, you at least always know the problem is with your code or maybe a library you're using. It feels like between Googling and looking at source code or examples there's always going to be an answer. Plus it's a private struggle because this usually only happens in uncommitted code on your dev box while you're working through developing the feature.

With infrastructure issues, it's like you're fully exposed because your site might be down and non-technical people (business higher ups) are being directly affected, and if you're the person responsible for this it all comes back to you.

Plus Kubernetes is Kubernetes. I'm not a veteran with it yet but on a local test cluster I still sometimes run into problems where I end up deleting the cluster and making a new one because after idling for 10 days everything stopped working. That strategy is simply not going to work in production. I'm hoping that's due to kind[0] bugging out and it won't happen on a cloud hosted provider. I'll be deploying a real production workload for a very non-trivial project in the near future.

[0]: https://kind.sigs.k8s.io/

Why would it not be an option?

I am not an SRE but I do know that we are actually able to just throw away our k8s cluster and it will be recreated from scratch.

In fact our dev environments work the same and I literally did this a couple if days ago because I had a similar problem to what you are describing. Throwing it away and letting it reinitialize from scratch fixed it.

You don't _want_ to do that in prod for sure. But it's a great thing to have in your back pocket, if it's tested well (disclaimer: I have no idea if and when they last tried this for prod ;))

> I have no idea if and when they last tried this for prod ;)

Famous last words :)

Man this resonates with me.

Reply on news.ycombinator.com