| 17 Jan 2026 |
 vcunat | But that's not so bad. One test takes much more time than one normal build. | 19:00:43 |
| vcunat | (by measuring how long it takes Hydra to clear this many-thousand bunch) | 19:01:26 |
 emily | right, I just mean that we shouldn't be seeing all the tests get rebuilt more than every week unless people aren't directing stuff to staging-nixos | 21:44:08 |
 hexa (signing key rotation when) | rebuilds-all-tests targeted against master should get a comment warning similar to backports w/o cherry-picks imo | 21:46:30 |
| emily | yeah, mass rebuilds targeted at master too | 21:54:44 |
| emily | just in those cases committers generally catch it due to experience, but with staging-nixos maybe not yet | 21:55:05 |
| vcunat | 🤔 maybe it would help. We do seem to get the test-rebuild labels:
https://github.com/NixOS/nixpkgs/issues?q=state%3Amerged%20label%3A10.rebuild-nixos-tests%20sort%3Aupdated-desc%20base%3Amaster | 21:57:46 |
| 18 Jan 2026 |
 mdaniels5757 | https://github.com/NixOS/nixpkgs/pull/481205 | 02:31:25 |
| vcunat | Is nix.ci still in use? | 07:36:38 |
| vcunat | Ah, I guess I should ask at #ofborg:nixos.org instead. | 07:37:13 |
|  ·☽•Nameless☆•777 · ± changed their profile picture. | 14:58:35 |
|  isabel changed their profile picture. | 20:43:54 |
| 19 Jan 2026 |
 n4ch723hr3r (putting stuff in your name is cringe) | Redacted or Malformed Event | 01:48:38 |
| n4ch723hr3r (putting stuff in your name is cringe) | * hows the radicle mirror going along? i found this issue should i ask there? | 01:49:52 |
 Arian | https://github.com/NixOS/nix/issues/15016 | 12:43:18 |
| Arian | lol | 12:45:09 |
 Sandro 🐧 | I merged a new usage of DeterminateSystems/magic-nix-cache like half an hour ago 😂 | 13:01:21 |
 yorik.sar | Would it be possible to retrigger this build? https://hydra.nixos.org/build/319408938 - it failed with segfault on configure phase in Hydra, but went pass that point just fine on my machine. Build didn’t finish for me yet though. | 17:01:50 |
 K900 | Done | 17:08:11 |
| 20 Jan 2026 |
| yorik.sar | @K900 thanks! Unfortunately, it still fails. Will look if there’s an issue about that already. Also ran into this issue with Hydra itself: https://github.com/NixOS/hydra/issues/1562 | 14:17:21 |
| vcunat | IIRC there are "imperfections" about logs on Hydra when a build gets retried. | 15:01:05 |
| vcunat | The worst part I guess is that binary cache indexes logs just by derivation, e.g.
https://cache.nixos.org/log/gf2lm8i875fqa55j3gix4rpxm5y8hp3j-itstool-2.0.7.drv
so it's not designed to hold multiple attempts. | 15:02:29 |
| yorik.sar | I don’t quite understand where Hydra gets the old log then. | 15:58:33 |
| yorik.sar | I also wonder how much space these logs take up in the cache bucket. | 15:59:30 |
| vcunat | Logs compress well. IIRC this was investigated when analyzing space usage (and considering garbage collection), and logs weren't taking a significant fraction of our S3 space. | 17:14:07 |
| vcunat | The hydra server also stores logs somewhere, like in local /nix/store or in the postgres? So you might be getting those instead from S3. | 17:14:42 |
| vcunat | * The hydra server also stores logs somewhere, like in local /nix/ or in the postgres? So you might be getting those instead from S3. | 17:14:55 |
| hexa (signing key rotation when) | `/nix/var/log | 17:15:18 |
| hexa (signing key rotation when) | * /nix/var/log | 17:15:21 |
| vcunat | But that dir suffers from the same design problem as the URL, I think. | 17:16:05 |
