| 24 Jun 2026 |
 Mic92 | builds are stopped when queue runner restarts | 07:33:28 |
 Vladimír Čunát | OK, sounds like it should be rare. | 07:33:53 |
| Vladimír Čunát | At least after we get that dedup which we don't have yet, if I understand you correctly? | 07:34:10 |
 K900 | https://github.com/NixOS/hydra/pull/1811 | 07:39:17 |
| K900 | OK this should do it | 07:39:20 |
| Mic92 | looks good | 07:43:29 |
| K900 | 
Download Tried it on a real log from browser devtools | 07:45:28 |
| K900 | It definitely renders things | 07:45:35 |
| K900 | (also what the fuck is a groonga) | 07:46:10 |
 Sergei Zimmerman (xokdvium) | In reply to @vcunat:matrix.org
🤔 but this deduplication won't persist if we restart the queue-runner and the leftover build result might persist on the builder? Though that seems hard to avoid. It will. Retires on different builder are also a concern. Frankenbuilds are a thing definitely | 07:46:13 |
 Arian | In reply to @vcunat:matrix.org
I still wouldn't be too happy if a builder built stuff against a local version of a dependency instead of waiting for the "official one" which ends up in S3. Yeh these Franken builds are annoying | 07:59:27 |
 dramforever | looks like the new queue runner is supposed to be uploading build logs, so that's odd | 08:03:31 |
| Mic92 | dramforever: check for content-encoding. Earlier version changed didn't included that in the upload | 08:24:25 |
| Vladimír Čunát | GC race probably:
https://hydra.nixos.org/build/332701150#tabs-buildsteps | 10:05:56 |
| Vladimír Čunát | (step 1 + 3) | 10:06:07 |
| dramforever | ... i'm just getting a 404? i don't know what content-encoding i should be checking | 10:14:22 |
| dramforever | or should i just check a newer build | 10:14:49 |
| Mic92 | vcunat: can you check if gc was running at the time? | 10:35:18 |
| Vladimír Čunát | Hydra says the job failed at 9:10:53 UTC. The GC on that builder finished 09:07:23 UTC. | 10:39:07 |
| Vladimír Čunát | So it doesn't look like a GC problem then? | 10:39:47 |
| Vladimír Čunát | * Hydra says the job failed at 9:10:53 UTC. The GC on that builder finished 09:07:23 UTC. (and next one didn't start until 10) | 10:40:02 |
| Vladimír Čunát | * Hydra says the job failed at 9:10:53 UTC. The GC on that builder finished 09:07:23 UTC. (and next one didn't start until 10:00) | 10:40:14 |
| Vladimír Čunát | Ah now. | 10:40:48 |
| Vladimír Čunát | * Ah no. | 10:40:52 |
| Vladimír Čunát | It was queued at that time. I don't know failure time. | 10:41:03 |
| Vladimír Čunát | * Hydra says the job failed at 9:10:53 UTC. The GC on that builder finished 09:07:23 UTC. (and next one didn't start until 10:00) | 10:41:16 |
| Vladimír Čunát | * So it doesn't look like a GC problem then? | 10:41:27 |
| Mic92 | than I need to check the db again. | 10:42:30 |
| Mic92 | it's there | 10:42:32 |
| Vladimír Čunát | Failed 10:02:28 on the builder. | 10:44:37 |
