Having work as Ops person at second place, most of time it ends up like that because Ops become dumping ground so they throw up walls in vain attempt to stem the tide. Application throwing error logs? Page out Ops. We made a mistake in our deploy. Oh well, page out Ops so they can help us recover. Security has found a vulnerability in a Java library? Sounds like Ops problem to us.

I've observed the same and, anecdotally, how much of a pain the ops team is is a function of how much responsibility shifts from the dev team to the ops team during the project lifecycle.

Basically, is the ops team there to support the developer/development team, or the product?

In the case of the development team, the ops team will tend to be willing to provide advice and suggestions but be flexible on tooling and implementation. In the case of the product, the ops team will tend to be a lot more rigid and inflexible.

This plays out in things like:

When the PWA becomes critical for the production line and then "is not working" at 3AM, who is getting paged? If it's the developer, then ops is "supporting the developer". If it's the ops team getting called to debug and fix some project they've never laid eyes on before at 3AM, then it's the product. They are, naturally, going to start caring a lot more about how it is set up, deployed, and supported because nobody likes getting woken for work at 3AM.

When some project's dependencies start running past EOL, who is going to update it? If it's the developer, then ops is "supporting the developer". If the ops team isn't empowered to give a deadline and have _someone else_ responsible for keeping the project functioning, then they're supporting the product and by letting it be deployed effectively committed to maintaining it in perpetuity and they're going to start caring a lot more about what sort of languages, frameworks, etc are used and specifically how projects are set up because context switching to one of dozens of different projects at 3AM is hard enough as-is without having to also be trying to learn some new framework du jour.

(And before anyone says "well the updates probably aren't necessary this is just ops being a pain"--think of the case of a project relying on GCP product that's being shutdown or some kubernetes resource that's been changed. In one case inaction will cause the project to fail, in the other ops' action will cause it to fail. See the first point as to who is going to get called about that. Even in the happy case, consistency brings automation and allows the team to support a _class_ of deployments instead of individual products.)

I don't think places exist stably in the middle ground because it's a painful place to be for very long. The responsibility and the control land on separate people, and the person with the responsibility but without the control is generally going to work to wrestle control to reduce misery. In the case where ops acts as if they're supporting the developers but is in practice supporting the product, it's not going to take too many 3AM calls before they start pushing back on how the product's deployed and supported.

I've been both of those ops guys you describe. When I was the "checklists, meetings, and picking the project apart" guy it had nothing to do with me wanting to make anyone's life difficult or flexing my knowledge. It had to do with the 3AM calls waking myself, my wife, and my newborn up. If I was taking on responsibility for keeping your _product_ functional through its useful life, yeah, I wasn't going to let people dump stuff on my plate unless I had some reasonable basis to believe it wasn't going to substantially increase my workload and result in more middle of the night calls. The checklists were my way of trying to provide consistency and visibility into the process of reducing my own pain, not my way of trying to create pain for others.

> upgrade windows

Ok, that explains this and the above comment. The last time I had to restart anything to apply an OS update was when I moved to a new RHEL LTS version, the lifespan of which is about 10 years. And there are many ways to do similar GNU/Linux upgrades without a restart at all.

Does Windows Server really need to restart for updates like normal Windows? If so, that's hilariously crap and I'm glad I've never had to touch it.

Edit: not saying a single VPS is fine if it's GNU/Linux, just remaking on the "restart to update" thing they mentioned

They went from leading / pioneering horizontal scalability and database deploying and scaling and orchestration to "quiet-quitting" 15 years ago and doing almost nothing ever since - today they're barely worthy of mention in any discussion on any tech that solves these problems.