| 28 Feb 2024 |
 hexa | 3 evaluators per host, 15 in parallel | 14:24:49 |
 Alyssa Ross | My latest OfBorg LLVM build for a staging PR on aarch46-darwin timed out with Ninja showing 2533/4188 | 14:25:05 |
| Alyssa Ross | but there's also the test suite after that, which takes a while | 14:25:11 |
| hexa | so 3h? | 14:25:17 |
| Alyssa Ross | so we might need to go to 2.5 or 3 hours | 14:25:17 |
| hexa | 
Download image.png | 14:25:49 |
| hexa | green is the queue for aarch64-darwin | 14:25:54 |
| hexa | * green is the queue for aarch64-darwin for the last 7 days | 14:26:02 |
 Lily Foster | yeah maybe we could just try it? it's definitely wasting time on builds it's not gonna finish. it's hard to tell without data if it finishing the builds at the cost of 3h would actually help or hurt though | 14:26:17 |
| hexa | please do | 14:26:31 |
| Lily Foster | otherwise disabling aarch64-darwin builds on staging PRs can be the backup | 14:26:49 |
| Lily Foster | In reply to @hexa:lossy.network
please do @[cole-h] | 14:26:53 |
| hexa | but we also need another machine to work through that queue in a reasonable time | 14:26:56 |
 adamcstephens | honestly i don't see how we can make things much worse | 14:26:58 |
| hexa | even turning it off wouldn't make it worse 🤷 | 14:27:22 |
| adamcstephens | it would be interesting to see the impact of moving just one of the x86_64-darwin machines to aarch64-darwin | 14:27:36 |
| hexa | I think they are sponsored machines by macstadium | 14:27:55 |
| adamcstephens | it feels wrong to have more capacity for a platform that's dying than the one that is growing | 14:27:59 |
| hexa | and domen is the contact they are used to | 14:28:24 |
| Lily Foster | i mean hydra changed all their x86_64-darwin builders to aarch64-darwin with rosetta, right? | 14:28:28 |
| Alyssa Ross | How much would another machine even cost? | 14:28:32 |
| Alyssa Ross | As long as x86_64-darwin is still tested, people are still going to wait for it. | 14:28:46 |
| Alyssa Ross | So I'm not sure how much moving around what the same number of machines are used for is going to help | 14:29:01 |
| hexa | hetzner stopped offinger m1 | 14:29:08 |
| Alyssa Ross | Unless we actually drop x86_64-darwin, or at least establish a norm that it's fine to completely ignore it. | 14:29:15 |
| adamcstephens | yeah, i'm not suggesting we drop it, but even getting the time for the two to align would be better. | 14:29:57 |
| hexa | hydras mac builders are hosted at hetzner | 14:30:11 |
| hexa | we apparently pay 52EUR/mo for them | 14:30:21 |
| hexa | so not too expensive | 14:30:24 |
| Alyssa Ross | BTW, should I give up waiting for OfBorg to pick up this job scheduled 18 days ago? https://github.com/NixOS/nixpkgs/pull/287727/checks?check_run_id=21435531956 | 14:30:25 |
