| 12 Sep 2021 |
 Sandro | you take the queued numbers from https://hydra.nixos.org/queue-summary ? or is there something fancier? | 19:01:52 |
 Vladimír Čunát | Including last-minute stdenv rebuilds is problematic. Especially the early bootstrapping needs lots of serial time, too. | 19:02:33 |
| Vladimír Čunát | https://hydra.nixos.org/queue-runner-status is another nice URL (and there's no link or button to it AFAIK) | 19:05:07 |
| Vladimír Čunát | Not "fancier" though :-) | 19:05:31 |
| Sandro | hmmm | 19:06:24 |
| Sandro | so you think we can merge it later today or tomorrow? | 19:06:44 |
| Vladimír Čunát | I'm not sure. There are down-sides, e.g. x86_64-darwin in OfBorg will become unusable for some time. | 19:09:15 |
 hexa | 
Download image.png | 19:11:15 |
| hexa | In reply to @vcunat:matrix.org
https://hydra.nixos.org/queue-runner-status is another nice URL (and there's no link or button to it AFAIK) ^ | 19:11:18 |
| Sandro | In reply to @lukegb:zxcvbnm.ninja
https://p.lukegb.com/raw/ViolentlySteadyHound.png heh based on this darwin will take a really long time | 19:11:43 |
| Vladimír Čunát | In reply to @hexa:lossy.network
sent an image. Where is that? | 19:12:14 |
| hexa | frontpage | 19:12:25 |
| Vladimír Čunát | That graph showed early time when stdenv rebuild was starting. But nevertheless, full rebuild certainly takes days on x86_64-darwin, even if there's no other work. | 19:23:41 |
| Vladimír Čunát | * That graph showed early time when stdenv rebuild was starting. But nevertheless, full rebuild certainly takes days on x86_64-darwin, even if there was no other work. | 19:23:48 |
| Sandro | linux does not take days, right? | 19:49:00 |
| Vladimír Čunát | I'd think it would be two days at least. | 19:49:36 |
| Vladimír Čunát | (if there was no other work) | 19:49:48 |
| Sandro | thats pretty long... | 19:50:03 |
| Vladimír Čunát | It's lots of work. It would probably be a bit better if more servers were paid for, though I can't be sure if some (other) bottleneck could appear anyway. | 20:08:48 |
 lukegb (he/him) | In reply to @vcunat:matrix.org
It's lots of work. It would probably be a bit better if more servers were paid for, though I can't be sure if some (other) bottleneck could appear anyway. It's definitely not just that, though. The aarch64-darwin staging-next queue hasn't reduced in size ~for the last day or so, and the only other builds in it are nixpkgs-unstable-aarch64-darwin, which shouldn't be stealing all the cycles | 21:59:50 |
| lukegb (he/him) | I'd expect building the compiler to take a long time, but even for x86_64-darwin it doesn't take more than ~a few hours for it to advance through e.g. one step of building LLVM/Clang/whatever | 22:00:20 |
| 13 Sep 2021 |
| lukegb (he/him) | update: looks like it's making progress now. I'm still curious how it ended up stalled for so long though; the queue-runner-status said all the slots had reservations in them, but the build status page didn't show anything running, hrm | 01:53:44 |
| Vladimír Čunát | I don't know. The jobset only has 50% more shares than staging-next. I don't know details of this accounting, but it sounds likely that staging-next has depleted lots of them recently. | 06:43:06 |
| Vladimír Čunát | * I don't know. The jobset is configured with 50% more shares than staging-next. I don't know details of this accounting, but it sounds likely that staging-next has depleted lots of them recently. | 06:43:31 |
| Vladimír Čunát | Roughly one third of *-darwin is finished on staging-next. | 07:15:17 |
| Vladimír Čunát | Though perhaps that was helped by several thousands of new quick failures :-/ | 10:16:57 |
| lukegb (he/him) | Yeah :/ | 10:33:14 |
| lukegb (he/him) | 
Download Screenshot_20210913-113558.png | 10:36:28 |
| hexa | jonringer: can you merge master into staging-next, some azure-cli conflicts | 13:31:13 |
| hexa | * jonringer: can you merge master into staging-next? some azure-cli conflicts | 13:31:19 |
