| 21 Oct 2025 |
 dramforever | workaround idea for the systemd static nodes thing, might be too cursed: just modprobe all the static nodes on boot | 12:22:21 |
| dramforever | wait this is technically not a staging matter anymore | 12:22:40 |
 emily | are there any channel blockers that need someone looking at them right now? | 23:58:08 |
| 22 Oct 2025 |
| emily | I see https://github.com/NixOS/nixpkgs/pull/454260 on the status page but not clear to me what it's waiting on | 00:16:37 |
 prince213 | https://github.com/NixOS/nixpkgs/pull/454422 | 02:32:56 |
| prince213 | Need some review on this CVE fix | 02:33:01 |
 Vladimír Čunát | This is unresolved so far, too
https://github.com/NixOS/nixpkgs/pull/453603#issuecomment-3426233438 | 05:20:30 |
 ElvishJerricco | oh crap, I didn't see that comment | 05:23:09 |
| dramforever | oh, yeah that's why i was asking again about whether we want to work around this... | 05:23:51 |
| Vladimír Čunát | And this one is a channel blocker:
https://hydra.nixos.org/build/309828960#tabs-buildsteps | 05:26:19 |
| Vladimír Čunát | Also nixpkgs-unstable is blocked by an unrelated merge that happened before merging staging-next
https://github.com/NixOS/nixpkgs/pull/452844#issuecomment-3425510646 | 05:28:27 |
| ElvishJerricco | I can't really think of a good way to workaround this. Just loading all modules in kmod static-nodes is really excessive, since we only need to fix the nodes that would get uaccess. And patching systemd.package seems...... not great | 05:32:56 |
| Vladimír Čunát | So do that filename replacement in those roughly 10 .nix files? | 05:35:58 |
| Vladimír Čunát | * So we do that filename replacement in those roughly 10 .nix files? | 05:36:02 |
| ElvishJerricco | right | 05:36:03 |
| ElvishJerricco | that's the thing that fixes the tests but not the bug | 05:36:14 |
| Vladimír Čunát | (and revert these on staging afterwards) | 05:36:20 |
| ElvishJerricco | but it's a pretty minor bug | 05:36:23 |
| dramforever | it is fairly easily work-around-able | 05:37:49 |
| dramforever | but that counts on users knowing it's the systemd 258 problem | 05:41:03 |
| Vladimír Čunát | Uh, amazon-ssm-agent did build for me on the first attempt, even though failing like 4-6 times on Hydra 😢 | 05:52:39 |
| Vladimír Čunát | And now it passed even there. | 06:20:08 |
| Vladimír Čunát | Annoying, but not critical. | 06:20:22 |
| Vladimír Čunát | So at least nixos-unstable might be unblocked right now. We'll see in a few hours. | 06:39:16 |
| ElvishJerricco | vcunat: and in a few hours a bot will merge that to staging so I can revert, right? | 06:43:37 |
| Vladimír Čunát | It was conflicted and didn't receive a merge for over 24h. | 06:48:15 |
| Vladimír Čunát | Merged now manually, so we can revert immediately. | 06:48:27 |
| ElvishJerricco | nice, thanks: https://github.com/NixOS/nixpkgs/pull/454459 | 06:49:41 |
 Grimmauld (any/all) | In reply to @elvishjerricco:matrix.org
I can't really think of a good way to workaround this. Just loading all modules in kmod static-nodes is really excessive, since we only need to fix the nodes that would get uaccess. And patching systemd.package seems...... not great Tbh overriding systemd.package doesn't sound that bad to me... | 07:06:15 |
| dramforever | maybe a script to run after udev rules to boop everything in /run/udev/static_node-tags/uaccess/? | 07:07:08 |
