| 22 Apr 2025 |
|  @cirnolovetech:matrix.org left the room. | 09:30:20 |
 jaredmontoya | Redacted or Malformed Event | 10:17:21 |
 aidalgol | You might want to ask in the dedicated home-manager room: #hm:rycee.net | 10:21:56 |
| jaredmontoya | thank you | 10:22:19 |
| jaredmontoya | * | 10:23:07 |
|  Thomas joined the room. | 17:31:22 |
 zowoq |
When’s the billing cycle for Hetzner stuff?
Billing is hourly and invoiced monthly.
| 23:15:29 |
| 23 Apr 2025 |
 emily | Darwin builder
13:56 up 6 days, 14:28, 2 users, load averages: 33.73 36.94 36.14
| 13:56:48 |
| emily | cc Gaétan Lepage? trying to test channel-blocking 24.11 / 25.05 stuff | 13:57:02 |
| emily | it's >3× overloaded | 13:57:10 |
 Gaétan Lepage | Killed | 13:57:20 |
| emily | ty | 14:05:44 |
| emily | load average on the Darwin builder is 55 right now. maybe the default cores/max-jobs need tweaking? | 21:23:14 |
| emily | (I am not sure how they apply to remote builders though) | 21:23:19 |
| Gaétan Lepage | I'm responsible for 5 LAV now | 21:23:29 |
| emily | looks like it is mostly C and Rust stuff right now 😅 | 21:24:04 |
| emily | Rust is particularly bad about eating up cores (it doesn't even support load limit) | 21:24:29 |
 Austin Horstman | yeah... i got a few things i build all the time that are rust projects and my computer just crawls | 23:44:35 |
| Austin Horstman | it's annoying | 23:44:41 |
| emily | I think they might have jobserver support, or at least be working on it. | 23:44:52 |
| emily | that would be the ideal solution | 23:45:07 |
| emily | https://doc.rust-lang.org/rustc/jobserver.html | 23:45:19 |
| Austin Horstman | i'm not too famliar with it, so is the jobserver shared among all the different builds you might be running? | 23:47:30 |
| emily | the Nix daemon should implement a job server based on a configured global load limit, let it through the sandbox, and pass down MAKEFLAGS pointing to it | 23:47:33 |
| emily | then GNU Make, rustc, and others would ask it for free slots before starting build jobs, essentially | 23:47:54 |
| Austin Horstman | that sounds great | 23:48:04 |
| emily | Ninja doesn't have an implementation though, I think there are long-stalled PRs trying to make it happen | 23:48:15 |
| emily | https://github.com/ninja-build/ninja/pull/2506 | 23:48:28 |
| emily | looks like this should land at some point | 23:48:34 |
| emily | anyway, it'd be better than load limits, but substantially more engineering work | 23:49:16 |
