| 24 Jun 2026 |
 Vladimír Čunát | I suppose we can reapply to master to keep it in -small. | 13:13:04 |
| Vladimír Čunát | * I suppose we can reapply also to master to keep it in -small. | 13:13:12 |
 K900 | I don't think staging-nixos merges to -next? | 13:13:17 |
| K900 | Automatically at least | 13:13:25 |
| Vladimír Čunát | I do that manually when I remember, and I did now. | 13:13:37 |
| Vladimír Čunát | * I do that manually when I remember, and I did when doing these revers. | 13:13:50 |
| Vladimír Čunát | * I do that manually when I remember, and I did when doing these reverts. | 13:13:52 |
| K900 | OK | 13:13:51 |
| Vladimír Čunát | Uh, elated-minsky builder system load is very unstable | 13:21:50 |
| Vladimír Čunát | It goes up to 1.5k | 13:22:15 |
| Vladimír Čunát | (with cpusome ~90%) | 13:23:12 |
| Vladimír Čunát | * (with cpusome ~90%, 96 cores) | 13:23:38 |
 hexa | but minimal system load, so I think we can leave it | 13:34:52 |
| hexa | Redacted or Malformed Event | 13:35:40 |
| hexa | Redacted or Malformed Event | 13:35:44 |
| hexa | libgomp is the primariy thing causing load | 13:36:41 |
| hexa | so probably something not respecting -j2 | 13:36:46 |
| Vladimír Čunát | What do you mean by "minimal system load"? | 13:38:15 |
| hexa | not much time in kernel space | 13:38:58 |
| hexa | so the scheduling cost is not that great | 13:39:45 |
| Vladimír Čunát | When you run 10x+ more threads than you can schedule, you'll be prone to issues, as tests will run 10x+ longer on average, etc. | 13:40:21 |
| hexa | right, but in this case I think we need to find the build going haywire | 13:41:12 |
| Vladimír Čunát | I see several different builds using many cores. | 13:45:37 |
| Vladimír Čunát | All python packages. | 13:46:58 |
| Vladimír Čunát | pytest I suspect | 13:47:58 |
| hexa | scikit-bio is my guess | 13:48:16 |
| Vladimír Čunát | I mean, all the culprits seem to be running pytest tests right now. | 13:48:37 |
| Vladimír Čunát | So perhaps it's what pytest does in usual setups? | 13:49:07 |
| hexa | yeah, but some are doing openmp | 13:49:11 |
| hexa | pytest-xdist in nixpkgs respects NIX_BUILD_CORES | 13:49:27 |
