Sudo Show 12: GitOps for Sysadmins
Episode 12 of the Sudo Show is the one where Brandon and I dig into GitOps, and I think it’s one of our better technical episodes because we ground it in stuff that’s actually happened to us instead of just walking through a whitepaper.
The core argument we make is that GitOps isn’t really about Git and it isn’t really about operations. It’s about giving developers the ability to provision and deploy infrastructure using templates that operations built and tested, so nobody’s waiting three weeks for a VM and nobody’s throwing tickets over a wall. We walk through the four tenets: collaboration, declarative configuration, observability, and auditability. That last one is where the “git blame” joke lands, though the real point is that accountability and blame are very different things. A bad config that breaks Apache should get rolled back and fixed, not turned into a performance review.
The war stories are what make this one worth listening to. I share the time I joined a company with four generations of undocumented sysadmin work piled up, discovered Oracle UIDs mismatched across three servers, and had to schedule a weekend outage for a system that wasn’t even in production yet. Brandon talks about inheriting a startup environment where no one documented anything and MySQL passwords were basically a treasure hunt. The point we keep coming back to is that self-documenting automation solves both of those problems at once.
If the 2AM page has ever ruined a vacation for you, this one’s for you. Give it a listen.


