Nix Flakes | 866 Members | |
| 168 Servers |
| Sender | Message | Time |
|---|---|---|
| 2 Dec 2024 | ||
| Redacted or Malformed Event | 13:33:16 | |
| 6 Dec 2024 | ||
| 04:36:51 | ||
| There must be something I'm not understanding about flake inputs - why does
| 04:37:49 | |
| if I had to guess, it's using nixpkgs from my flake registry which is too far out of sync to build. How can I make it use the locked inputs from my dependency? | 04:38:32 | |
| it should be by default using the locked ones | 10:43:43 | |
you can check that with nix flake metadata | 10:43:49 | |
| it could be that the flake.lock was not updated accordingly when changing inputs. Try nix flake update | 10:44:17 | |
| 8 Dec 2024 | ||
| 19:21:34 | ||
| I use a nix flake to rebuild my NixOS configuration which includes a configuration for Nginx. When I run: sudo nixos-rebuild -v test I don't see the generated nginx configuration file (nginx.conf) in the generated result directory: lrwxrwxrwx 1 root root 85 Dec 7 08:54 result -> /nix/store/xdc1ayh50k13x2mlk9199x4k88a10cds-nixos-system-nixos-24.11.20241203.b681065 What am I missing and how can I find the version that was built? | 20:17:01 | |
Where exactly do you look for the nginx.conf? It is not placed at /etc/nginx, but inside the /nix/store directly. When you look for the systemd unit, there should be referenced as passed via -c | 20:28:02 | |
* Where exactly do you look for the nginx.conf? It is not placed at /etc/nginx, but inside the /nix/store directly. If you look for the systemd unit, there should be referenced as passed via -c | 20:29:15 | |
| I can search the nix store and do find multiple versions of the file but since all dates in the nix store are the same how do I find the latest version generated. How does the nginx server determine the correct version to use. I was assuming that there would be a symbolic link pointing to the latest version from somewhere in the result directory generated by the rebuild. | 20:38:23 | |
| the systemd unit gets rebuild when the config changes. | 20:54:33 | |
| and it contains a reload config directive when the config file changes AFAIK | 20:58:26 | |
| Is this documented somewhere along with how to use this to determine my latest version of my nginx config file | 21:17:40 | |
| i am not aware of that, to check the latest used i check the systemd unit with systemctl cat nginx | 21:19:53 | |
| That worked | 21:39:49 | |
| 9 Dec 2024 | ||
I just pgrep -fal nginx and grab the config from there. | 14:03:05 | |
| 10 Dec 2024 | ||
| Many thanks🙂 | 00:32:26 | |
| Hi Everyone I know flake-parts is the recommended way to modularize a flake, but I'm finding it a bit confusing. If the main purpose is to manage system, why can't perSystem take arbitrary options?I'm trying to set a custom output called pkgs-unstable to be able to do nix run .#pkgs-unstable.hello along with nix run .#hello that pulls from nixos-stablethis seem to only be possible through flake = {} and not perSystem = {}? Which then defeats the purpose as I'd need to manually set the system of the unstable nixpkgs input. | 19:52:37 | |
is there a way to set arbitrary options on perSystem? | 19:54:30 | |
| Here are two examples: Vanilla Flakes
Flake Utils
In the Flake Util example, I'm able to build a multi-platform flake that exposes Can anyone help me with recreating this minimal Flake Util setup in flake-parts ? | 22:07:41 | |
| * Here are two examples: Vanilla Flakes
Flake Utils
In the Flake Util example, I'm able to build a multi-platform flake that exposes Can anyone help me with recreating this minimal Flake Util setup in flake-parts ? | 22:08:07 | |
| * Here are two examples: Vanilla Flakes
Flake Utils
In the Flake Util example, I'm able to build a multi-platform flake that exposes Can anyone help me with recreating this minimal Flake Util setup in flake-parts ? | 22:34:36 | |
| 11 Dec 2024 | ||
| 14:19:07 | ||
| 14:20:33 | ||
| 12 Dec 2024 | ||
| 06:24:39 | ||
| Redacted or Malformed Event | 17:41:20 | |
| Good afternoon - we've got a large deployment (~45 hosts in each of 3 tiers, each looking ~identical but with differing hostnames and the like) we're hoping to manage with deploy-rs or colmena (we've yet to choose). We've started using flake-parts for integration with other tools (mostly treefmt and numtide/devshell). This has been working well so far and we are making some decent use of the module system to abstract everything well. We are now looking to handle system abstraction. We know our fleet is entirely x86_64-linux, so we don't require that abstracted at all, but I am struggling to handle the system definition generation. Currently we have a function that we want to import to make the api for definition a bit cleaner, but I am struggling to get it importing without infinite recursion. Is anyone abstracting their system definitions somehow that might be able to provide pointers? Is there some way to import a library of functions that aren't held to the nixos/flake-parts module schema? | 17:46:26 | |
| Unfortunately, I can't share much of the code we have now, but I can describe in more detail if that helps anyone. | 17:47:18 | |