➜ colmena apply --on laptop
[INFO ] Using flake: git+file:///home/isaac/.dotfiles
[INFO ] Enumerating nodes...
[INFO ] Selected 1 out of 4 hosts.
       ❌ 6s Failed: Child process exited with error code: 1
laptop ✅ 5s Evaluated laptop
laptop ✅ 0s Built "/nix/store/9zsi0h4lbmnj9hggbg8vzxr1mhy1wqw3-nixos-system-laptop-23.05pre-git"
laptop ❌ 1s Push failed: Child process exited with error code: 1
[ERROR] Failed to complete requested operation - Last 1 lines of logs:
[ERROR]  failure) Child process exited with error code: 1
[ERROR] Failed to push system closure to laptop - Last 9 lines of logs:
[ERROR]  created)
[ERROR]    state) Running
[ERROR]   stderr) copying 10 paths...
[ERROR]   stderr) copying path '/nix/store/3ddp6lnxxi98gb43rgqx6531n94ygcpl-home-manager-path' to 'ssh://isaac@10.0.0.22'...
[ERROR]   stderr) error: cannot add path '/nix/store/3ddp6lnxxi98gb43rgqx6531n94ygcpl-home-manager-path' because it lacks a valid signature
[ERROR]   stderr) copying path '/nix/store/jdc1w4iw3dd1n5rz257awxld13hs45f5-starship-config' to 'ssh://isaac@10.0.0.22'...
[ERROR]   stderr) error (ignored): error: writing to file: Broken pipe
[ERROR]   stderr) error: unexpected end-of-file
[ERROR]  failure) Child process exited with error code: 1
[ERROR] -----
[ERROR] Operation failed with error: Child process exited with error code: 1

For some reason there is an increasing number of people asking for support who use colmena.

And it is quite common that one wants to build the machine that causes issues to inspect the system closure.

For something like that it seems as if colmena build is quite practical, though at the same time it seems as if it lacks an argument to specify a single host to build. Is this indeed missing or did I miss something in the docs?

Also, it would be nice if there were appropriately named out links after a build.

NobbZ:

For something like that it seems as if colmena build is quite practical, though at the same time it seems as if it lacks an argument to specify a single host to build. Is this indeed missing or did I miss something in the docs?

You can specify the nodes to build with colmena build --on node-name. The --on argument is a bit misleading because the build isn't (necessarily) happening "on" the node. The reason is that colmena build --on xyz is just an alias of colmena apply --on xyz build (you can pass it other goals besides build like dry-activate).

Also, it would be nice if there were appropriately named out links after a build.

This is an interesting suggestion. Currently there is --keep-result which will cause Colmena to create GC roots under the .gcroots subdirectory. When deployment.buildOnTarget is true, builds happen on the target nodes themselves, so it's not possible to create out links locally in that case. The problem with creating out links in a user-facing manner is that people will be wanting to specify the names (-o result2) or otherwise depend on the outputs for scripting purposes, a use case colmena apply UI doesn't support.

NobbZ:

For something like that it seems as if colmena build is quite practical, though at the same time it seems as if it lacks an argument to specify a single host to build. Is this indeed missing or did I miss something in the docs?

You can specify the nodes to build with colmena build --on node-name. The --on argument is a bit misleading because the build isn't (necessarily) happening "on" the node. The reason is that colmena build --on xyz is just an alias of colmena apply --on xyz build (you can pass it other goals besides build like dry-activate).

Also, it would be nice if there were appropriately named out links after a build.

This is an interesting suggestion. Currently there is --keep-result which will cause Colmena to create GC roots under the .gcroots subdirectory. When deployment.buildOnTarget is true, builds happen on the target nodes themselves, so it's not possible to create out links locally in that case. The problem with creating out links in a user-facing manner is that people will be wanting to specify the names (-o result2) or otherwise depend on the outputs for scripting purposes, a use case the colmena apply interface doesn't support (such is delegated to colmena eval).

In reply to @ronixx:matrix.org
sent an image.

{ config, pkgs, inputs, ... }:

{

  boot.isContainer = true;
  deployment = {
    targetHost = "ip"; # x86_64-linux
    targetPort = 22;
    buildOnTarget = true;
    targetUser = "nix";
    tags = [ "theDev" ];

    # keys = {
    # 	"freshrss" = "helloworld";
    # };

    keys."freshrss.secret" = {
      # Alternatively, `text` (string) or `keyFile` (path to file)
      # may be specified.
      text = "helloworld";

      # keyCommand = [ "vault" "read" "-field=env" "secret/dns01" ];

      destDir = "/run/keys"; # Default: /run/keys
      user = "nix"; # Default: root
      # group = "nginx";             # Default: root
      permissions = "0640"; # Default: 0600

      # uploadAt = "pre-activation"; # Default: pre-activation, Alternative: post-activation
      uploadAt = "post-activation";
    };
  };

  ## ---------- System Packages --------- ##
  environment.systemPackages = [
	pkgs.cowsay
  ];	

  fileSystems."/mnt/drive" = {
    device = "/dev/disk/by-id/scsi-0DO_Volume_volume-blr1-01";
    fsType = "ext4";
  };

  networking.hostName = "theDev"; # Define your hostname.

#   security.acme.acceptTerms = true;
#   security.acme.email = "email@gmail.com";

#   services.freshrss.enable = true;
#   services.freshrss.package = pkgs.freshrss;
  # services.freshrss.name = "god";
#   services.freshrss.passwordFile = "/run/keys/freshrss.secret";
#   services.freshrss.virtualHost = "freshrss";


  ## ---------- Postgresql ------------ ##
  
  services = {
	postgresql = {
		enable = true;
	};
	postgresqlBackup = {
		enable =  true;
		startAt = "*-*-* 01:15:00";
	};
  };

  ## ---------- Netdata System Monitor ------------ ##

  services.netdata = {
    enable = false;
    config = {
      global = {
        "page cache size" = 32;
        "update every" = 15;
      };
    };
  };

  ## ---------- Nginx Webserver ------------ ##

  services.nginx.enable = true;
  services.nginx.commonHttpConfig = ''
    limit_req_zone $binary_remote_addr zone=limitbyaddr:10m rate=5r/s;
    limit_req_status 429;
  '';

  services.nginx.virtualHosts = {

  };

  system.stateVersion = "22.05";

}