| 10 Mar 2023 |
 @elvishjerricco:matrix.org | Well the existing unit also doesn't do anything :P If we're going to keep it, we should have an implementation. And if we have an implementation, that's logic we have to maintain even if no one's using it | 21:32:29 |
| 11 Mar 2023 |
 jaen | Hi, I've asked previously for help with getting systemd-stage1 to work with my LUKS setup, but didn't get any responses, so I've put it aside for a while. I'm trying again now and I think the only issue I have is getting my unit dependencies wrong, as doing the same steps manually on a running system properly shows Plymouth password prompt and decrypts what it should decrypt. | 11:52:42 |
| jaen | Is there any way to show all the units and their dependencies for the stage1 systemd before booting into it? Maybe that will make it easier for me to figure out where to put my units. | 11:52:44 |
| jaen | If it helps, what I'm trying to do at a high-level is taking advantage of the fact that if you put a domain socket as keyfile source, systemd will try to get keyfile contents from that socket. As such, I have added a keyfile.socket to the stage1 units and it spawns a unit with a simple bash script using systemd-ask-password --no-tty and cryptsetup to read the keyfile. I can provide relevant parts of config if you would need more details. | 11:58:18 |
| jaen | * If it helps, what I'm trying to do at a high-level is taking advantage of the fact that if you put a domain socket as keyfile source in crypttab, systemd will try to get keyfile contents from that socket. As such, I have added a keyfile.socket to the stage1 units and it spawns a unit with a simple bash script using systemd-ask-password --no-tty and cryptsetup to read the keyfile. I can provide relevant parts of config if you would need more details. | 11:58:33 |
| jaen | I think that's everything relevant – https://gist.github.com/jaen/f1e32d2d7810b20ba97159a19f8374db | 12:06:49 |
| 13 Mar 2023 |
 @hexa:lossy.network | Redacted or Malformed Event | 15:00:26 |
| 14 Mar 2023 |
|  ckie (they/them) changed their display name from ckie (they/them) to ckie (they/them; heavily limited keyboard usage, dictation or voice only). | 01:09:13 |
| 22 Mar 2023 |
 @linus:schreibt.jetzt | ElvishJerricco: re zfs mount generator, I wonder if in the future (when we have multiple initramfs support) we could even make the generator work in the systemd initramfs, by updating an initramfs which contains the cache in the zedlet | 13:25:55 |
| @linus:schreibt.jetzt | and thus make non-legacy mounts for /var/lib and stuff work too | 13:26:22 |
| @elvishjerricco:matrix.org | I have thought about doing that already. But it's not clear where to source that cache from. If the point is to keep the mountpoint details out of the nix config, then it has to be acquired when the generation is added to /boot, which doesn't seem ideal | 13:29:02 |
| @elvishjerricco:matrix.org | oh, I misread | 13:29:16 |
| @elvishjerricco:matrix.org | have the zedlet do it. | 13:29:21 |
| @elvishjerricco:matrix.org | huh | 13:29:22 |
| @elvishjerricco:matrix.org | that seems... risky | 13:29:28 |
| @elvishjerricco:matrix.org | and also not helpful for nixos-install | 13:29:40 |
| @linus:schreibt.jetzt | hm yeah | 13:30:36 |
| @elvishjerricco:matrix.org | btw I do use non-legacy mountpoints for initrd FSes, I just also have them in fileSystems with the zfsutil option :P | 13:30:38 |
| @linus:schreibt.jetzt | right | 13:30:50 |
| @elvishjerricco:matrix.org | it's just nice to have mount.zfs handle turning properties into mount options, and to have the fs hierarchy so you can just make new datasets and have the mountpoints be inferred | 13:31:27 |
| @linus:schreibt.jetzt | yeah | 13:31:49 |
| @linus:schreibt.jetzt | [root@sol:~]# zfs list -Ho name | wc -l
277
| 13:32:12 |
| @elvishjerricco:matrix.org | (though I also have different datasets with mountpoint=/ canmount=off... | 13:32:17 |
| @linus:schreibt.jetzt | I don't want to do that in nixos config lol | 13:32:19 |
| @elvishjerricco:matrix.org | yea... | 13:32:26 |
| @linus:schreibt.jetzt | (ok, like half of that is podman image layers actually. But still) | 13:32:52 |
