| 26 Dec 2024 |
|  phaer changed their display name from phaer to phaer (8650 at 38c3). | 17:41:31 |
|  @lorenzleutgeb:matrix.org changed their display name from Lorenz Leutgeb to Lorenz Leutgeb (📞6343). | 19:33:26 |
| @lorenzleutgeb:matrix.org changed their display name from Lorenz Leutgeb (📞6343) to Lorenz Leutgeb 📞6343. | 19:36:40 |
| 27 Dec 2024 |
|  raitobezarius changed their display name from raitobezarius to raitobezarius (DECT: 3538 / EPVPN 2681). | 07:32:51 |
 K900 | I've merged the gitlab aarch64 test skip | 09:21:34 |
| K900 | Largely because I don't think we'll see any movement on it during the holiday season | 09:21:47 |
| K900 | Starting a new eval | 09:22:13 |
|  tilpner changed their display name from tilpner to tilpner (38c3 3209). | 09:41:20 |
|  garbas left the room. | 10:21:07 |
|  kdn changed their display name from nazarewk to kdn. | 12:38:25 |
| K900 | We have a successful tested job on nixos-unstable | 15:18:36 |
 hexa | finally | 15:26:31 |
| 28 Dec 2024 |
 Gaétan Lepage | The ARM builder has OOM again :/ | 11:30:40 |
| Gaétan Lepage | Sorry | 11:30:48 |
| Gaétan Lepage | * The ARM builder has OOMed again :/ | 11:30:57 |
| hexa | looks to be responsive right now | 11:56:25 |
| Gaétan Lepage | Yes, it's back | 11:57:41 |
| Gaétan Lepage | thanks | 11:58:12 |
|  Bruno Rodrigues changed their profile picture. | 13:02:25 |
 emily | Gaétan Lepage: might be worth limiting -j or --cores when running those large Python builds? it often makes the builders with fewer cores difficult to use too | 14:39:12 |
| Gaétan Lepage | Definitely, it is just not often predictable when such problems will occur.
Certain packages are so RAM hungry at testing that building 2 of them concurrently can fill the ARM box's RAM.
Yet, it also happen to have 64 concurrent builds using less than 50GB... | 15:24:33 |
| 29 Dec 2024 |
|  anund_ joined the room. | 04:41:32 |
|  sinan changed their profile picture. | 07:27:55 |
| Gaétan Lepage | The ARM builders' root partition is 95% full:
[glepage@aarch64:~]$ df -h /
Filesystem Size Used Avail Use% Mounted on
/dev/nvme0n1p2 1.8T 1.7T 93G 95% /
| 08:13:23 |
 Vladimír Čunát | I don't know... simply running nix-store --gc now, but I suspect you would also have permissions for that. | 08:32:51 |
| Gaétan Lepage | In reply to @vcunat:matrix.org
I don't know... simply running nix-store --gc now, but I suspect you would also have permissions for that. Good point, I was not expecting to be able to do so. | 08:33:32 |
|  elikoga changed their display name from elikoga (@38c3 📞488{0,1}) to elikoga (@38c3 📞488{0,1,9}). | 11:03:15 |
 linj | Just noticed this PR https://github.com/NixOS/nixpkgs/pull/354586. So zstd cache is on the plan now? | 15:32:19 |
| Vladimír Čunát | I'm not sure. Hydra doesn't seem to be bottle-necked on compression of outputs anymore. So the tradeoff that remains is CPU vs. network on clients + S3 size, and that one isn't very convincing to me so far (IIRC). | 16:01:41 |
| Vladimír Čunát | * I'm not sure. Hydra doesn't seem to be bottle-necked on compression of outputs anymore, thanks to HW upgrade. So the tradeoff that remains is CPU vs. network on clients + S3 size, and that one isn't very convincing to me so far (IIRC). | 16:02:03 |
