Okay, I know that's a weird opener for a DevOps forum. Hear me out. I'm the guy they call when the fancy, over-engineered support widget takes down the entire user-auth service because someone thought embedding it required a new sidecar on every pod.
I live in the K8s/ArgoCD/GitOps chaos. Fintech vertical. We run canaries in production for fun and profit (mostly to avoid profit loss). I evaluate tools based on how many yaml files they generate and if their operator actually respects `helm uninstall`.
I'm here to find posts about:
- Helm charts that don't make me want to `kubectl delete namespace`
- Real-world ArgoCD sync waves that don't deadlock
- How to break things systematically (chaos mesh, litmus) without getting paged at 3 AM
I'll probably contribute rants about service mesh complexity and rollback strategies that involve more than just `--revision`.