| 26 Jun 2026 |
 emily | I didn't know where they run tbh | 15:07:09 |
| emily | but I'm not sure how adding another build to the jobset to replace some code running outside the sandbox would be a security regression | 15:07:27 |
 hexa | they run on pluto.nixos.org | 15:08:16 |
| hexa | the host that runs monitoring, too | 15:08:21 |
| hexa | not on any system that has anything else to do with the hydra stack | 15:08:40 |
| emily | right. well I am concerned about Hydra security in general but as long as stdenv gets built on there we can't really improve our security by moving things out of Hydra builds | 15:09:26 |
|  jutty joined the room. | 19:46:05 |
|  Sapphire joined the room. | 21:16:26 |
 John Ericson | Hoping we can rip out a lot of abandoned stuff | 21:34:31 |
| John Ericson | I do like drv-level scheduling, that's the core value proposition of hydra | 21:34:59 |
| hexa | it's where most other scheduling systems fail | 21:40:11 |
| 27 Jun 2026 |
 Yorusaka Miyabi | https://hydra.nixos.org/build/332738696/download/1/manual/ Yields a 500 error | 03:35:18 |
| Yorusaka Miyabi | (Build refers to develop version of Nix manual) | 03:35:43 |
 Grimmauld (any/all) | i just had some fun plotting occupancy against 1d standard deviation (blue). It seems with new queue runner, we have much more fluctuations in occupancy. Why is this? Intuitively i would expect more deviation with worse scheduling. Don't get me wrong, total utilization seems to be higher, but this looks like we may have a delay between a job being finished and a new one being queued, causing occupation to temporarily drop | 05:24:05 |
| Grimmauld (any/all) | 
Download image.png | 05:24:05 |
 Mic92 | Before we don't have https://github.com/NixOS/infra/pull/1099 I don't think those graphs are very meaningful. There are some known bottlenecks I introduced to make it stable. | 05:42:58 |
| Mic92 | hexa (signing key rotation when): I also added now also the patch to macOS builders: https://github.com/NixOS/infra/pull/1104 But my internet upload is still not great until Sunday, if you wanted to take a stab if it. | 05:52:11 |
| Mic92 | * hexa (signing key rotation when): I also added now also the patch to macOS builders: https://github.com/NixOS/infra/pull/1104 But my internet upload is still not great until Sunday, if you wanted to take a stab earlier | 05:52:17 |
| Mic92 | * hexa (signing key rotation when): I also added now also the patch to macOS builders: https://github.com/NixOS/infra/pull/1104 But my internet upload is still not great until Sunday, if you wanted to take a stab at it earlier | 05:52:30 |
 whispers [& it/fae] | out of curiosity, is it expected/known that no *-linux jobs are being scheduled at all? the only one is https://hydra.nixos.org/build/332926387, which has been running for five hours. the rest of the machines seem to be sitting fully idle: https://hydra.nixos.org/machines | 06:05:06 |
| whispers [& it/fae] | * | 06:05:29 |
| whispers [& it/fae] | * | 06:06:48 |
| whispers [& it/fae] | * | 06:09:43 |
 Vladimír Čunát | Looks like very slow ingestion right now. | 06:51:26 |
| Vladimír Čunát | (no idea why) | 06:51:32 |
| Vladimír Čunát | And it prefers nixpkgs/unstable which has no linux builds left. | 06:52:15 |
| Vladimír Čunát | Though we also have builds in nixos/*-small 🤔 That seems weird. | 06:52:55 |
| Vladimír Čunát | Ah, wrong assumptions. In the 7h old eval there are 1.2k linux builds left. | 07:00:21 |
| Vladimír Čunát | Let me try restarting the queue runner 🤷 Seems low-risk. | 07:01:18 |
| Vladimír Čunát | This morning the runner is logging lots of
sqlx::query: slow statement: execution time exceeded alert threshold
| 07:10:20 |
