we use something along the lines of

    boot.binfmt.registrations = {
      qemu-aarch64 = {
        magicOrExtension = "\\x7fELF\\x02\\x01\\x01\\x00\\x00\\x00\\x00\\x00\\x00\\x00\\x00\\x00\\x02\\x00\\xb7";
        mask = "\\xff\\xff\\xff\\xff\\xff\\xff\\xff\\x00\\xff\\xff\\xff\\xff\\xff\\xff\\xff\\xff\\xfe\\xff\\xff";
        interpreter = "${pkgs.qemu}/bin/qemu-aarch64";
      };
    };

    nix.sandboxPaths  [ "/run/binfmt" "${pkgs.qemu}" ];

In reply to @wucke13:matrix.org
even if I have only a localhost machine, I also need to enable nix.distributedBuilds ?

In reply to @wucke13:matrix.org
Is it possible to have hydra build the nixosConfigurations in a flake?

In reply to @wucke13:matrix.org
Any chance that you also know an answer to this one? 🙂

In reply to @andreas.schraegle:helsinki-systems.de
one note about the machines file, that might be useful, is that you don't have to use /etc/nix/machines, but can give hydra its own machines file, e.g. with services.hydra.buildMachinesFiles
no idea about the flakes thing, though

In reply to @tim:stratum0.org
wucke13 hydraJobs."<attr>"."<system>" = derivation; should work.

In reply to @wucke13:matrix.org
Answering my own question: I think .#nixosConfigurations.my-machine.config.system.build.toplevel will do :)

Well, in that case I have a nother convenience snippet:

hydraJobs = builtins.listToAttrs (lib.mapAttrsToList (machine: config:
  lib.nameValuePair machine {
    "x86_64-linux" = config.config.system.build.toplevel;
  }) nixosConfigurations);

Any feedback on how the system can be read from an existing nixosConfiguration would be welcome, to fix the remaining ugly code

And again I think I have an answer to my own question:

hydraJobs = builtins.listToAttrs (lib.mapAttrsToList (machine: config:
  lib.nameValuePair machine {
    "${config.config.nixpkgs.system}" = config.config.system.build.toplevel;
  }) nixosConfigurations);

The system can be read from .#nixosConfigurations.my-machine.config.nixpkgs.system