| 12 Feb 2025 |
 atemu12 | You could try limiting the parallelism | 07:13:04 |
| atemu12 | It's probably hard-coded | 07:13:18 |
| atemu12 | You could also try using the github mirror | 07:13:38 |
 waltmck | it looks like the number of jobs defaults to the number of logical cores from reading the code | 07:13:57 |
| atemu12 | I've never done it myself but it ought to be possible using ROBOTNIX_MIRRORS | 07:14:16 |
| waltmck | I capped the number of jobs to 4 and it appears to be chugging along, I'll see if it works. Might be smart to adjust the code so it works on CPUs with many cores | 07:18:34 |
| waltmck | * I capped the number of jobs to 4 and it appears to be chugging along, I'll see if it works. Might be smart to adjust the default so it works on CPUs with many cores | 07:18:50 |
| atemu12 | Sounds like a good idea | 07:18:56 |
| waltmck | it's line 251 of mk_repo_file.py, just change default=4 or some other small number | 07:19:22 |
| waltmck | I can file a PR later for it if that's easier | 07:19:31 |
| atemu12 | You probably don't gain anything after a few cores because you become network-bound | 07:19:46 |
| atemu12 | Just PR it | 07:20:00 |
| waltmck | huh it's still throttling with 4 jobs, just takes longer. I'll try 2. | 07:33:33 |
