| 17 Jun 2021 |
 srhb | With "pre" you just can't. | 12:08:48 |
| srhb | (ie. conditionfileexists) | 12:09:32 |
 @grahamc:nixos.org | https://github.com/NixOS/nixops/blob/master/nix/keys.nix#L215-L260 | 12:09:40 |
| @grahamc:nixos.org | conditionfileexists will just not start the unit, but maybe depending on a .path unit might do it? afaik the only time I've used .path units I got it wrong :D | 12:12:36 |
| @grahamc:nixos.org | cool, thanks srhb I think this works | 12:17:28 |
| @grahamc:nixos.org | to summarize, an option per key:
uploadAt = 'post-activation' | 'pre-activation' (default); | 12:18:03 |
| @grahamc:nixos.org | * to summarize, an option per key:
- uploadAt = 'post-activation' | 'pre-activation' (default);
- manually uploading secrets always uploads the secrets
| 12:18:18 |
| @grahamc:nixos.org | * to summarize, an option per key:
- uploadAt = 'post-activation' | 'pre-activation' (default);
- manually uploading secrets always uploads the secrets
| 12:18:28 |
| srhb | In reply to @grahamc:nixos.org
conditionfileexists will just not start the unit, but maybe depending on a .path unit might do it? afaik the only time I've used .path units I got it wrong :D Ah yes, you're right of course. | 12:18:29 |
| srhb | In reply to @grahamc:nixos.org
to summarize, an option per key:
- uploadAt = 'post-activation' | 'pre-activation' (default);
- manually uploading secrets always uploads the secrets
Agree with the summary. I think this is non-controversial :) | 12:19:13 |
| @grahamc:nixos.org | cool | 12:19:22 |
| @grahamc:nixos.org | thanks! | 12:19:23 |
| @grahamc:nixos.org | posted a summary: https://github.com/DBCDK/morph/issues/99#issuecomment-863193469 | 12:24:40 |
 hexa | what's the general opinion on terragrunt? | 14:59:42 |
| hexa | currently looking into how I can get a clean directory structure | 15:00:02 |
| 19 Jun 2021 |
| hexa | nixops depends on a certifi version from 2020/04 | 12:47:38 |
| hexa | * nixops transitively depends on a certifi version from 2020/04 | 12:47:49 |
| hexa | hope I don:t have to tell anybody how bad an idea that is | 12:48:07 |
| hexa | * hope I don't have to tell anybody how bad an idea that is | 12:48:14 |
 David Arnold (blaggacao) | I wonder what appeal people find in a stateful gitops tool. | 12:53:44 |
| David Arnold (blaggacao) | (looks like quite a bad idea, too) | 12:54:07 |
| David Arnold (blaggacao) | * (looks like quite a bad idea, too -- my association here) | 12:54:25 |
| David Arnold (blaggacao) | * (looks like quite a bad idea, too -- my _actual_ association here) | 12:54:57 |
| David Arnold (blaggacao) | Isn't the whole point of gitops is that the only state is the vcs one? (+ some transparent caching, eventually) | 12:57:43 |
| hexa | okay, worse. 2020/04 doesn't work, even if upstream says so | 13:00:42 |
| hexa | adisbladis: https://github.com/NixOS/nixpkgs/pull/127453 | 13:00:53 |
|  nazarii left the room. | 20:16:55 |
| 22 Jun 2021 |
 Gytis Ivaskevicius | Does deploy-rs have any advantage over nixops v2? | 07:55:14 |
| David Arnold (blaggacao) | If nixops v2 is stateless, a rebranding might be in order. Also, from looking at the readme deploy-rs does not pretend to be a terraform. | 12:11:04 |
 ryantm | In reply to @gytis-ivaskevicius:matrix.org
Does deploy-rs have any advantage over nixops v2? deploy-rs lets you deploy arbitrary nix profiles, not just the NixOS main system profile. I'm not aware of that being a feature of NixOps. | 12:37:33 |
