| 25 May 2026 |
 hexa (signing key rotation when) | Nothing in particular stands out. Very mundane changes IMO. | 17:26:53 |
 Jeremy Fleischman (jfly) | Could use eyes from someone more familiar with NixOS boot up on this issue: https://github.com/NixOS/infra/issues/1047 | 18:42:01 |
| hexa (signing key rotation when) | beginning with ofborg infra | 19:35:42 |
| hexa (signing key rotation when) | done | 19:59:59 |
| hexa (signing key rotation when) | continuing with non-critical-infra | 20:00:04 |
| hexa (signing key rotation when) | and the builders | 20:08:43 |
| hexa (signing key rotation when) | mimas, titan, pluto | 20:32:29 |
| hexa (signing key rotation when) | I expect hydra to be back in 11m | 20:34:51 |
| hexa (signing key rotation when) | looks like we're early | 20:41:46 |
| hexa (signing key rotation when) | and the macs | 20:49:13 |
| 26 May 2026 |
|  @aangularframework:matrix.org left the room. | 04:12:00 |
|  ·☽•Nameless☆•777 · ± changed their profile picture. | 05:41:22 |
|  phanirithvij changed their display name from loudgolem to phanirithvij. | 11:40:38 |
 vcunat | Uh, sleepy-brown now failed builds due to /nix/var/nix/builds on tmpfs exhausthing the RAM. It seems to happen relatively often and causing real issues, e.g. a couple days ago:
https://github.com/NixOS/nixpkgs/issues/523780 | 19:10:30 |
| vcunat | Well, hopeful-rivest this time, sleepy-brown last time, if I look right. | 19:12:18 |
| vcunat | I'm inclined to at least simply dropping that tmpfs - and thus using the SSD for build temporaries. But if you'd like to have something more elaborate, I don't care, as long as it's reasonably reliable. | 19:16:10 |
| hexa (signing key rotation when) | hopeful-rivest has half the amount of ram than the other builders | 19:49:02 |
| hexa (signing key rotation when) | we should have replaced the box way back when | 19:49:19 |
| 27 May 2026 |
| vcunat | It looks like the next couple of days of our Linux capacity will be spent almost all on NixOS tests (for three different branches). | 08:54:03 |
 Arian | I wish we had better DSL for tests. A lot of them could benefit from some parallelism | 09:11:27 |
| Arian | Very often can cut the runtime in half by just spawning all VMs at the start | 09:11:45 |
| vcunat | Running itself is not a bottle-neck here. | 09:42:33 |
| vcunat | It's getting all dependencies ready and to the builder. | 09:43:02 |
| vcunat | Especially ingestion of all those .drv steps, apparently. | 09:43:36 |
| vcunat | (our *-linux runners get almost idle for... it will be many hours today) | 09:45:14 |
 Alyssa Ross | I concur that we should stop running non-blocker tests | 11:30:57 |
 Winter | https://hydra.nixos.org/build/330335876 why did this create a secret fourth cached/propagated attempt two minutes after a retry i kicked off? | 20:55:34 |
| Winter | * https://hydra.nixos.org/build/330335876 why did this create a secret fourth cached/propagated attempt two minutes after a retry i kicked off (3)? | 20:56:18 |
| Winter | though this one has a bunch of n/as... i feel like yesterday it was doing similar but without n/as 🫠 | 20:57:21 |
|  Carsten Kragelund Jørgensen joined the room. | 21:02:27 |
