| 18 Jun 2026 |
 Mic92 | thx | 08:03:10 |
| Mic92 | https://github.com/NixOS/hydra/issues/1805#issuecomment-4739539396 This might explain a few of the aborts | 08:10:05 |
|  vmfunc joined the room. | 08:17:29 |
 Sergei Zimmerman (xokdvium) | https://matrix.to/#/!CTCrFzsBPYmDLmrja4%3A0upti.me/%24olty6gvo6UpfhC5zyAr0LPIlO3U6CffpDZGhIAS89Ho
Nixpkgs isn’t faring too well too | 08:18:56 |
| vmfunc | hiya, ci looks broken repo-wide right now on nixpkgs. the ci/pinned.json bootstrap can't realise its tooling, hash mismatch on nghttp3-1.15.0 from cache.nixos.org (specified 157s6q…, got 07svi8…), no substituter, which takes out lint / parse, lint / nixpkgs-vet and eval / x86_64-darwin. it's been hitting PRs at random depending on how warm the runner's cache is. seeing it on #532911, #532912, and my #532910. known? pretty sure this is linked to cache | 08:18:59 |
| vmfunc changed their profile picture. | 08:19:39 |
| Sergei Zimmerman (xokdvium) | In reply to @vmfunc:mozilla.org
hiya, ci looks broken repo-wide right now on nixpkgs. the ci/pinned.json bootstrap can't realise its tooling, hash mismatch on nghttp3-1.15.0 from cache.nixos.org (specified 157s6q…, got 07svi8…), no substituter, which takes out lint / parse, lint / nixpkgs-vet and eval / x86_64-darwin. it's been hitting PRs at random depending on how warm the runner's cache is. seeing it on #532911, #532912, and my #532910. known? pretty sure this is linked to cache https://github.com/NixOS/hydra/issues/1804 | 08:19:39 |
| vmfunc | i see | 08:20:27 |
| Sergei Zimmerman (xokdvium) | Whats really scary is that the nixpkgs didn’t get bumped in CI? | 08:23:12 |
| Sergei Zimmerman (xokdvium) | Last bump was https://github.com/NixOS/nixpkgs/pull/530903? | 08:24:17 |
 K900 | I wonder if it's uploading everything it builds | 08:25:01 |
| K900 | Not just everything it doesn't have | 08:25:05 |
| Sergei Zimmerman (xokdvium) | Did existing NARs also get sprayed with garbage? | 08:25:07 |
| K900 | And then something something race something | 08:25:11 |
| Sergei Zimmerman (xokdvium) | Lemme check the last-modified on the narinfo | 08:25:17 |
| vmfunc changed their profile picture. | 08:25:19 |
| vmfunc | investigating rn, drafting a comment give me a moment | 08:26:34 |
 Ryan Burns | the texliveFull failure that raised that issue is a package that hasn't changed since 2026-06-10, so I'd say yes - perhaps a new invalid narinfo was pushed for it | 08:27:28 |
| vmfunc | cache seems repaired now | 08:28:03 |
| Sergei Zimmerman (xokdvium) | On the /nix/store/i8ykl6jr2x9x3bqv2m8l6gn6i2246lyr-nghttp3-1.15.0 side something fishy is definitely going on:
curl https://cache.nixos.org/nar/157s6qjhidk6asyg2qg6ga6s5a78hwg2al9dyaw8sxx8759q2qx6.nar.zst -v |& grep "last-modified:"
last-modified: Thu, 18 Jun 2026 06:34:05 GMT
curl https://cache.nixos.org/i8ykl6jr2x9x3bqv2m8l6gn6i2246lyr.narinfo -v |& grep "last-modified:"
last-modified: Wed, 17 Jun 2026 06:18:34 GMT
| 08:31:44 |
|  Marie joined the room. | 08:37:05 |
 Vladimír Čunát | All these .narinfo do have Compression: zstd | 08:37:46 |
| Vladimír Čunát | so that means that the .narinfo were also uploaded very recently, right? | 08:38:01 |
| vmfunc | my edge serves the nar.zst as last-modified wed 17 jun 06:18, sergei's serves thu 18 jun 06:34 | 08:38:03 |
| K900 | So we're definitely double uploading then | 08:38:19 |
| Sergei Zimmerman (xokdvium) | Yeah, but that can't be right | 08:38:27 |
| Vladimír Čunát | The FOD from texlive was old, surely. | 08:38:40 |
| vmfunc | so it's not local cache warmth like i first guessed, it's edge inconsistency | 08:38:46 |
| Mic92 | If you have theories, please put them in the issue. | 08:38:49 |
| Mic92 | I will not be able to read everything in this chat | 08:39:00 |
