While you can run Hydra on your own laptop, in a team it definitely makes sense to setup a dedicated machine for it. This machine will then act as the CI and as a Nix cache for all your devs.
First you need to go through the NixOS installation process.
The last steps of the process are:
nixos-generate-config --root /mnt # generate template of /etc/nixos config files
vim /mnt/etc/nixos/configuration.nix # edit the config
nixos-install # build the system
reboot # boot into it
Once you generate the NixOS config file (first of the above commands), you will want to modify your configuration.nix to reasemble
the one in ./hydra and include the files that it references.
Then run nixos-install and you're ready to reboot.
NOTE
When running any command that will use the above config file you will get a warning:
warning: unknown setting 'allowed-uris'This is a bug, but the config works just fine - you can safely ignore it.
The configuration.nix file is annotated with comments, so I recommend spending some time reading it.
You can fork the above config at https://github.com/fghibellini/hydra-nixos-server.
Now on your client machines you have to modify your /etc/nix/nix.conf file to:
binary-caches = http://10.0.0.25:5000/ http://cache.nixos.org/
require-sigs = false
Where 10.0.0.25 is the IP address of your Hydra instance.
TODO darwin builder
TODO cover nix copy --to ssh://appuipieds $(nix-build shell.nix)
Now that we have a NixOS machine up and running we can proceed to the next chapter where we configure Hydra to automatically run our builds.