| 3 Apr 2025 |
 Mic92 | Haven't had the time yet to git bisected it | 05:21:50 |
 hexa | the constituent globbing is what was test failing in nixos/infra as well | 05:21:50 |
| hexa | Redacted or Malformed Event | 05:21:52 |
| hexa | * arasdfasd' | 05:22:03 |
| Mic92 | I will get to it but in the meantime this branch could pin nix-eval-jobs | 05:23:45 |
| Mic92 | hexa (signing key rotation when): is this still an issue affecting NixOS infra? https://github.com/NixOS/nix/issues/9961#issuecomment-2424164165 | 07:59:49 |
 Vladimír Čunát | AFAIK it's just try and see. | 08:03:23 |
| Vladimír Čunát | Some version combinations suffer from it, some do not. | 08:03:47 |
| Mic92 | Ok. I don't think we have different versions at the moment. I will de-priotize this than from an infra issue (John was working on some VM tests to improve compatibility between different versions in general) and put more the focus on hydra again. | 08:38:18 |
| Mic92 | * Ok. I don't think we have different versions at the moment. I will de-priotize this than from an infra issue (John was working on some VM tests to improve compatibility between different versions in general) and I put more the focus on hydra again. | 08:38:56 |
| Mic92 | Infra meeting is now vcunat hexa (signing key rotation when) Jonas Chevalier Arian | 16:03:26 |
| hexa | won't be able to make it | 16:06:40 |
| hexa | upgrading hetzner macs to 15.4 | 23:55:32 |
| 4 Apr 2025 |
 emily | 15.4 has some weird issues btw | 00:13:35 |
| emily | https://matrix.to/#/!lheuhImcToQZYTQTuI:nixos.org/$BBcedAqqTJbkE1Nhx8vkEMEu1ZuqwKvG4zePhc83fN0?via=nixos.org&via=matrix.org&via=nixos.dev | 00:13:40 |
| hexa | exciting | 00:14:15 |
| hexa | lun: looks like you didn't set the composable kernel packages to big-parallel after all? https://hydra.nixos.org/build/294295249 | 16:51:07 |
| 6 Apr 2025 |
 Lun | It's split into parts only some of which are marked big-parallel
https://github.com/NixOS/nixpkgs/blob/5c0edef29f64ca168f3d6ba0e944c2ae57f693bd/pkgs/development/rocm-modules/6/composable_kernel/default.nix#L8 | 04:57:16 |
| Lun | https://hydra.nixos.org/build/294295249/nixlog/26 _mha was on a 24 core machine so I think that part worked? Do more of the parts need set to big-parallel? | 04:59:26 |
| Vladimír Čunát | The machines have more cores, but you get -j based on the big-parallel flag. | 06:18:46 |
| Vladimír Čunát | Without the flag you only get 2 typically (on current hydra.nixos.org). | 06:19:11 |
| Vladimír Čunát | You can see on that linked page (another tab): https://hydra.nixos.org/build/294295249#tabs-buildsteps
that some of the builds took many hours. | 06:20:55 |
| Vladimír Čunát | * You can see on that linked page (another tab): https://hydra.nixos.org/build/294295249#tabs-buildsteps
that some of the builds took several hours, so those are probably candidate. | 06:21:21 |
| Vladimír Čunát | (I assume they do parallelize well.) | 06:21:33 |
|  sinan changed their profile picture. | 11:32:59 |
| 7 Apr 2025 |
| hexa | rebooting hydra for updates | 13:39:46 |
 John Ericson | hexa (signing key rotation when): once https://github.com/NixOS/hydra/pull/1456 passes, I think we're ready to do another update | 18:54:43 |
| John Ericson | CC Mic92 | 18:54:52 |
| John Ericson | that has all of the extra hydra.nixos.org work merged to master, and Nix 2.28 | 18:55:11 |
| hexa | John Ericson: cool, sounds like we can drop the nej pin then and migrate to master | 20:17:27 |
