| 5 Apr 2023 |
 @elvishjerricco:matrix.org | I'm realizing that the networkd PR doesn't hide the boot.initrd.systemd.network.* options from the manual. I wonder if I should fix that or if we should just stop hiding boot.initrd.systemd.* options now that it's gotten pretty darn good | 02:03:53 |
| @elvishjerricco:matrix.org | To that end: https://github.com/NixOS/nixpkgs/issues/224457#issuecomment-1496903296
Hopefully that was in line for that thread | 04:45:35 |
|  R̴̨͕͇͍̞̮̐̅͆̌̀̉̐͋̈́̃̀͒́̎̅̚̚̚͠͝Ĕ̵̡̛͖͖̟̙̫̱͈̘̞̭͍͍͑̌̄͑̓̋̓̀̈̏̈́͊̇͊͆̉͂̏̀̃̚͘͝͝ͅͅD̶̡̢͔̱̖̮͙͉̘̺͓͍̩̮͈͍͗̃̀̏͌͘͜ͅŚ̸̬̭̯̬͙͇͓̬̩̳̤͚͓̤̩̺͉͖̉͛̓̿̎͊̿̆́̐͂̇͌̄̇̓͘ͅͅT̴̞̫̘̝͇͔̟̪̪̦͂̔̎̀̎ͅŎ̷̡̬̹̪͈̭̣͈̭̭͉̦̖̝̘̪͖͔̥̦̘̻̳Ṋ̶̛̫͈̳̘͚̜̔̋͆̅̈́͊̑͊̉̌̈́̾͑̈́̚ͅË̸̡̨̨̛͇̜̖͔͖̻̟̗̠̙͓̘̗̥͉͇̜͑͆͊͑͑̀̓͒͜͝͝ joined the room. | 05:51:36 |
 Arian | Agreed! | 06:04:32 |
| Arian | Lets make it visible | 06:04:37 |
| 6 Apr 2023 |
 @uep:matrix.org | Yes | 08:51:29 |
 K900 | Should probably add a release note too | 08:53:21 |
| K900 | Like "this thing is here now please test it" | 08:53:29 |
 @crtified:crtified.me | I don't know whether it's well known, as I got curious about systemd stage1 a few weeks ago: Using a specialization makes it really easy to test this out longterm without risking a non-bootable system | 08:55:15 |
| @crtified:crtified.me | So right now, I'm basically using this to have a fallback to the scripted stage1:
boot.initrd.systemd.enable = lib.mkDefault true;
specialisation.scripted-stage1.configuration.boot.initrd.systemd.enable = false;
| 09:05:23 |
| @uep:matrix.org | Good tip, maybe even worth adding to that release note. I just made the switch and figured I'd roll back if it broke, but you make a good point about resilience / recovery with ongoing changes as it stabilises. I think we're probably past that point now though | 09:10:12 |
 colemickens | Except there's an assert for fallbackToPassword that one has to confront. https://github.com/NixOS/nixpkgs/commit/843a7b63eb35e0da8a2eaf1e03803699f414424b | 12:54:34 |
| colemickens | There might be a way to nix around it but I found it easier to just drop what I think should just be ignored anyway. | 12:56:09 |
| colemickens | (and I think this might've only mattered in the specific case where systemd-stage1 was the specialization, rather than the "main" config.) | 12:57:15 |
| @crtified:crtified.me | In reply to @colemickens:matrix.org
(and I think this might've only mattered in the specific case where systemd-stage1 was the specialization, rather than the "main" config.) At least I didn't hit that, although I'm using LUKS | 14:07:40 |
| colemickens | Right it's a LUKS option - You'd only hit it if you were using a keyFile/yubikey and then wanted to fallback to password. | 18:40:17 |
| colemickens | and then if you set it in your main config, you can't "unset" it for the specialisation. Or maybe I'm unaware of a way to do that with the module system | 18:40:49 |
| 8 Apr 2023 |
|  @matthewp:matrix.org joined the room. | 22:00:55 |
| 9 Apr 2023 |
|  @fritz.otlinghaus:helsinki-systems.de joined the room. | 16:36:42 |
| 10 Apr 2023 |
 @nickcao:nichi.co | Tried netbooting on vultr with sytemd initrd, failing mounting nix store. While I failed to enter the emergency console (the console does seem to register keystrokes), I noticed that the fileSystems.<name>.depends option is not used anywhere. So the currently correct ordering of the overlayfs mount could be a pure luck? | 03:03:48 |
| @nickcao:nichi.co | * Tried netbooting on vultr with sytemd initrd, failing mounting nix store. While I failed to enter the emergency console (the console does not seem to register keystrokes), I noticed that the fileSystems.<name>.depends option is not used anywhere. So the currently correct ordering of the overlayfs mount could be a pure luck? | 03:07:38 |
| @nickcao:nichi.co | And mounting only the squashfs works: https://github.com/NickCao/nixpkgs/commit/f22f82f4e1f23ae3c0e23b6be5a6c449469398bf, confirming my theory. | 03:14:31 |
| @elvishjerricco:matrix.org | Nick Cao: Yea, depends doesn't work with systemd-based mount things yet. That's on my list of fairly easy thiings to fix. It's an issue in stage 2 as well. depends only works for stage 1 file systems with the old initrd | 03:15:23 |
| @nickcao:nichi.co | Could be as simple as setting https://www.freedesktop.org/software/systemd/man/systemd.mount.html#x-systemd.requires-mounts-for= | 03:16:31 |
| @elvishjerricco:matrix.org | yea, that's the solution | 03:16:38 |
| @elvishjerricco:matrix.org | just haven't gotten around to it yet | 03:16:42 |
| @nickcao:nichi.co | It's strange that systemd recommends toolings to write fstab instead of generating mount units directly | 03:17:10 |
| @elvishjerricco:matrix.org | nah, fstab is way better | 03:17:31 |
| @elvishjerricco:matrix.org | it has a lot of niceties | 03:17:37 |
| @nickcao:nichi.co | String concat does not look nice to me | 03:17:48 |
