For local development I used to use Vagrant, but now I use Docker almost exclusively, primarily because of how fast it starts and stops containers. Also I find the config format easier to work with.
I still use Vagrant in those rare instances when Docker is not viable. Sometimes I really do need full OS instance, or need to run something other than Linux, or I just want normal bridged network interfaces.
IMHO Vagrant is "stable". Possibly boring. (A good thing for tools, generally.) But it works well.
Why would you ask if it's dead?
About the point of using a dumb terminal: I tried it with a cheap DO VM, and came to the conclusion that they're woefully underpowered for any serious work compared to the laptop's processor.
The cost to performance wasn't there: just using my laptop for development was cheaper and more performant than renting a VM monthly for cloud development.
Yes, I still use vagrant, on pretty much every project I work on - both my own/my companies, and client projects.
I'll give two examples of how I use it:
I (or technically, my company) produce a number of tools for managing various common aspects of a Linux web application stack.
For each tool, there's a regular Vagrant VM setup from the root of the repo, which is used for basic testing, building, running tests etc - essentially this is just to achieve a known state, and to abstract the environment from whatever OS it's running on. Some of the tools either specifically require, or support, 'cluster' mode, where some kind of state or configuration is managed across multiple machines. For these tools, there's an additional multi-machine Vagrantfile (generally just in a subdir) which sets up a mini cluster of 2 or 3 nodes, and allows testing of the sync/replication/what have you.
For a client I work with right now, we again, use it in two ways, achieving two separate goals:
The primary, day to day use, is providing a completely self-contained local dev environment for the webapp. This setup is used for anyone doing dev on the app - backend, frontend, whatever.
The less common, but still very necessary use, is similar to what you mentioned, multi-machine infra testing. The infra for this client is all provisioned using a shell script that works on a system of roles and environments. To test this, the repo with the setup has a vagrant config that matches the production environment.
Additionally, this repo also has a packer config, which builds a custom base box, for the aforementioned "day to day dev" Vagrant env, using the same setup script.
As mentioned by @ocdtrekkie, Vagrant is relatively stable. It'll be interesting to see where 3.0 takes things, but right now it's a pretty reliable tool, that just gets the job done.
Edit: typo (anything -> anyone)
I prefer Docker. I believe it’s a better and more convenient abstraction model.
But not all systems are a good fit for Docker. When I need an actual VM, then Vagrant is a godsend.
Is this something people use Nix for?
Docker
I just checked their GitHub [0], and they seem to have a release on 27th July, 2021. Also it looks like they are re-writing vagrant 3.0 using go [1]
[0]: https://github.com/hashicorp/vagrant
[1]: https://www.hashicorp.com/blog/toward-vagrant-3-0