| 4 May 2023 |
 @mberndt:matrix.org | It doesn't use systemd-ask-password | 05:42:16 |
| @mberndt:matrix.org | It just runs Plymouth: https://github.com/stratis-storage/stratisd/blob/77cc6df928c554464a84e9aac7af4718196babdf/dracut/90stratis/stratis-rootfs-setup#L20
I don't know why they did it that way. I guess we could change it | 05:45:28 |
 @elvishjerricco:matrix.org | Oh wow. Yea that's unfortunate. I guess they wanted to support non-systemd-initrd | 05:47:10 |
| @elvishjerricco:matrix.org | but that very much sucks because plymouth is definitely not a given either | 05:48:20 |
| @mberndt:matrix.org | I do have an assertion to make sure it is used | 05:48:43 |
| @elvishjerricco:matrix.org | I know | 05:48:54 |
| @elvishjerricco:matrix.org | it's just a bummer | 05:49:38 |
| @elvishjerricco:matrix.org | I mean we could patch it but I dunno if it's worth it | 05:49:55 |
| @mberndt:matrix.org | I don't think their initrd support actually works without systemd, it looks like an oversight to me | 05:50:04 |
| @elvishjerricco:matrix.org | Oh | 05:50:14 |
| @mberndt:matrix.org | I can just create a PR in their repo to fix that | 05:50:36 |
| @mberndt:matrix.org | I mean they use a systemd generator. That isn't going to work without systemd | 05:51:07 |
| @elvishjerricco:matrix.org | Well do they exclusively do it that way? It's not uncommon to support non-systemd dracut and systemd dracut | 05:52:26 |
| @elvishjerricco:matrix.org | * Well do they exclusively do it that way? It's not uncommon to support both non-systemd dracut and systemd dracut | 05:52:36 |
| @mberndt:matrix.org | It doesn't look like it to me:
https://github.com/stratis-storage/stratisd/blob/77cc6df928c554464a84e9aac7af4718196babdf/dracut/90stratis/module-setup.sh | 05:54:58 |
| @mberndt:matrix.org | And in this other script they just use systemd-ask-password:
https://github.com/stratis-storage/stratisd/blob/77cc6df928c554464a84e9aac7af4718196babdf/systemd/stratis-fstab-setup#L27 | 05:56:00 |
| @elvishjerricco:matrix.org | Interesting. Well, might as well open a PR; it's a real simple change | 05:56:31 |
| @mberndt:matrix.org | Yup | 05:56:39 |
| @elvishjerricco:matrix.org | could then remove plymouth from these "dependencies" I guess https://github.com/stratis-storage/stratisd/blob/77cc6df928c554464a84e9aac7af4718196babdf/dracut/90stratis/module-setup.sh#L37 | 05:56:53 |
| @mberndt:matrix.org | Yes | 05:57:18 |
|  @aktaboot:tchncs.de left the room. | 19:44:34 |
| 5 May 2023 |
| @elvishjerricco:matrix.org | Arian: Did you end up looking into the repart thing? | 03:02:49 |
| @elvishjerricco:matrix.org | I did realize one arguably good reason to run repart in initrd even if you're not autoformatting root. Generators, such as gpt-auto-generator, will only be able to see the partitions if you run repart in initrd, or if you systemctl daemon-reload after repart finishes. | 04:37:13 |
 @nikstur:matrix.org | Could we somehow make systemd treat /nix/store similarly to /usr (without patching systemd)? Then the semantics would completely line up | 21:28:59 |
| @nikstur:matrix.org | In reply to @elvishjerricco:matrix.org
I did realize one arguably good reason to run repart in initrd even if you're not autoformatting root. Generators, such as gpt-auto-generator, will only be able to see the partitions if you run repart in initrd, or if you systemctl daemon-reload after repart finishes. gpt-auto-generators must've been the primary reason for why systemd-repart also reads definitions from /sysroot | 21:30:43 |
| @elvishjerricco:matrix.org | The problem is that /usr stores the current OS's configuration. /nix/store is not a single OS though | 21:31:37 |
| @elvishjerricco:matrix.org | so it's not clear how to make it use the correct configuration without...... activation | 21:31:56 |
| @elvishjerricco:matrix.org | and that just means mounting the whole initrd FS again | 21:32:03 |
| @nikstur:matrix.org | Mhmm that's a good point. | 21:33:07 |
| @nikstur:matrix.org | But maybe if we teach the systemd tools the locations of the files (like /etc/os-release, repart definitions etc.) it expects, don't we get close enough? | 21:35:45 |
