[asdjlkadsjklads]

We used NixOps for a while. Small company (<100, >20), with lots of client deployments on prod and various pre-prod environments as well. NixOps worked well enough, and i never heard complaints about it. Though i wasn't in the trenches with it personally.

Our problem was hiring for it. Ultimately we ended up moving away from NixOps and into more Kube/AWS solutions to provide this basic orchestration. Unlike a lot of Kube talk, it isn't entirely overkill for us - but NixOps worked just fine, and is probably better for us. We just didn't have Ops with enough knowledge in Nix to properly exploit Nix to easily achieve our goals. Nothing was easy.

With Kube/AWS though a lot of hiring decisions were dead easy and bringing people on who knew it was dead simple. As much as i'd prefer to be on Nix, reducing decision fatigue is at least nice.

[lambdaba]

Thank you, insightful. Do you recall if diving in to NixOps internals and fixing bugs yourselves was necessary at some point? Did you have issues with incomplete coverage of AWS APIs?