| 30 Oct 2022 |
|  Claes joined the room. | 04:56:20 |
| 31 Oct 2022 |
 @oxalica:matrix.org | https://github.com/NixOS/nixpkgs/blob/00bde1ac69b572f56ca461a76d8f1db59b29da0f/nixos/modules/services/hardware/udev.nix#L13
Is there a reason why not all upstream udev rules are included? Some rules seems important, like btrfs and vconsole | 16:55:16 |
| @oxalica:matrix.org | Also I found that systemd udev rules is even not in stage 2 (services.udev.packages), which is very problematic to me | 16:57:00 |
| @oxalica:matrix.org | I'm trying to enable all rules in at least stage 2. https://github.com/NixOS/nixpkgs/pull/198823
Maybe we should also include more rules in initrd? | 18:31:13 |
 @elvishjerricco:matrix.org | oxalica: I could be wrong but doesn't systemd automatically include the udev rules in its own out path? Have you confirmed that they aren't actually currently in use in stage 2? | 18:32:14 |
| @elvishjerricco:matrix.org | (stage 1 is another story of course) | 18:32:20 |
| @oxalica:matrix.org | In reply to @elvishjerricco:matrix.org
oxalica: I could be wrong but doesn't systemd automatically include the udev rules in its own out path? Have you confirmed that they aren't actually currently in use in stage 2? Yes. Because vconsole-setup doesn't restart when my GPU driver loaded | 18:32:48 |
| @elvishjerricco:matrix.org | And you've confirmed that adding all udev rules from systemd changes that behavior? | 18:33:16 |
| @oxalica:matrix.org | It used to work, and it broke about several months ago | 18:33:32 |
| @oxalica:matrix.org | In reply to @elvishjerricco:matrix.org
And you've confirmed that adding all udev rules from systemd changes that behavior? Yes it fixes for me | 18:33:42 |
| @elvishjerricco:matrix.org | Hm... I'm guessing there's some kind of reason for it being the way it is | 18:34:09 |
| @elvishjerricco:matrix.org | as for stage 1, we certainly can't support all the rules that systemd provides because we don't include all the user space tools required, so it's probably best to only have the ones we know we need? Not sure | 18:34:40 |
| @elvishjerricco:matrix.org | But hopefully there's some kind of explanation for stage 2 in git history or a PR thread somewhere... | 18:35:04 |
| @oxalica:matrix.org | In reply to @elvishjerricco:matrix.org
oxalica: I could be wrong but doesn't systemd automatically include the udev rules in its own out path? Have you confirmed that they aren't actually currently in use in stage 2? :thinking: Seems systemd-analyze cat-config udev/rules.d do contain upstream rules even without that PR. But I double checked that vconsole-setup do break iff we don't explicitly include rules... | 18:40:36 |
| @oxalica:matrix.org | * 🤔 Seems systemd-analyze cat-config udev/rules.d do contain upstream rules even without that PR. But I double checked that vconsole-setup do break iff we don't explicitly include rules... | 18:40:51 |
| @elvishjerricco:matrix.org | I'm not sure that systemd-analyze command is doing what you think it is. When I list the files in my /etc/udev/rules.d/ on my desktop, it has nothing in common with the stock systemd udev rules | 18:52:58 |
| @elvishjerricco:matrix.org | which seems quite odd indeed... | 18:53:05 |
| @elvishjerricco:matrix.org | maybe there was a commit where systemd was accidentally removed from services.udev.packages? | 18:54:42 |
| @elvishjerricco:matrix.org | for instance there's these lines that have to do with a udev rule that's in systemd but not installed on my system: https://github.com/NixOS/nixpkgs/blob/15a051a5b64c8b8bd7c402d102ec273df807e897/nixos/modules/services/hardware/udev.nix#L147-L149 | 18:55:25 |
| @elvishjerricco:matrix.org | I eval'd config.services.udev.packages in nixos versions back to 19.09 and systemd was never in there. Though interesetingly, udisks is newly missing in nixos-unstable, but was there from 22.05 and backwards | 19:07:38 |
| @oxalica:matrix.org | My vconsole-setup issue is spotted from around 2022-09-12
https://matrix.to/#/%23stage1-systemd%3Anixos.org/%24hY0xc__Lg14Fy8aUCcfYc-rq42MHNePejrnhxQHgTSY?via=matrix.org&via=tchncs.de&via=helsinki-systems.de&via=ckie.dev | 19:11:19 |
| @oxalica:matrix.org | * My vconsole-setup issue is spotted since around 2022-09-12
https://matrix.to/#/%23stage1-systemd%3Anixos.org/%24hY0xc__Lg14Fy8aUCcfYc-rq42MHNePejrnhxQHgTSY?via=matrix.org&via=tchncs.de&via=helsinki-systems.de&via=ckie.dev | 19:11:47 |
|  @tinybronca:sibnsk.net changed their display name from underpantsgnome! to underpantsgnome. | 20:29:29 |
| 1 Nov 2022 |
| @elvishjerricco:matrix.org | While working on the systemd 252 update, I noticed that the hibernate-systemd-stage-1 test is already failing on hydra. There are way too many units getting started before the hibernation resume unit does. | 01:31:14 |
| @elvishjerricco:matrix.org | When you resume from hibernation, ideally initrd has started only the units necessary to bring up the resume device | 01:32:50 |
| @elvishjerricco:matrix.org | I wonder if we're missing a unit from upstream or something | 01:33:07 |
 flokli | I think some tests with systemd-initrd are "known to be broken". We should probably at least add a comment on these, so people know what to expect. | 07:26:18 |
| @elvishjerricco:matrix.org | Uh. They weren't broken before. They should be fixed | 07:28:39 |
| flokli | Nixpkgs bisect go brr then, i guess | 08:04:11 |
| flokli | Sheesh, there needs to be some better regression tracking. | 08:04:29 |
