| 15 Jul 2024 |
 zowoq | Buildbot is working again. | 10:37:09 |
|  @dminca:matrix.org changed their display name from dminca to nixpkgs. | 17:28:56 |
| @dminca:matrix.org changed their display name from nixpkgs to dminca. | 17:42:32 |
|  @drupol:matrix.org left the room. | 18:10:12 |
|  ChaosAttractor joined the room. | 18:59:37 |
| 16 Jul 2024 |
|  tomasharkema joined the room. | 01:49:09 |
| zowoq | We've replaced the two darwin CI builders (both m1, 16gb ram, 256gb ssd) with a single machine (m2, 24gb ram, 1tb ssd). | 09:50:13 |
 Gaétan Lepage | Cool ! You also have x86 machine(s) as well, right ? | 11:26:26 |
 Traxys | Is the migration finished? I think there is a worker (worker-021?) that does not have the nix daemon active | 21:38:03 |
| zowoq | I've restarted it but evaluating nixvim seems to causing it to crash. I'll do some testing later to see if it is specific to nixvim or not. | 23:47:33 |
| zowoq | * I've restarted it but evaluating nixvim seems to be causing it to crash. I'll do some testing later to see if it is specific to nixvim or not. | 23:47:58 |
 Soliprem | In reply to @zowoq:matrix.org
I've restarted it but evaluating nixvim seems to be causing it to crash. I'll do some testing later to see if it is specific to nixvim or not. if you're not committed to nixvim, nvf is pretty good | 23:48:10 |
| Soliprem | and hasn't had evaluation errors for me | 23:48:17 |
| Soliprem | or at least in the meantime it can work as a substitute, maybe using the maximal config | 23:48:41 |
| Soliprem | until nixvim is fixed | 23:48:52 |
| zowoq | nixvim isn't broken, I was replying to the previous comment about the broken worker. | 23:50:56 |
| Soliprem | In reply to @zowoq:matrix.org
nixvim isn't broken, I was replying to the previous comment about the broken worker. oh sry | 23:51:07 |
| zowoq | In reply to @glepage:matrix.org
Cool ! You also have x86 machine(s) as well, right ? We don't have native x86 darwin but the aarch64 machine can emulate x86. | 23:54:09 |
| 17 Jul 2024 |
| Gaétan Lepage | Makes sense. Good to know, thanks ! | 06:07:19 |
| Traxys | I tried to improve the situation for nixvim in 1882, empirically it seems to help as locally the build only seems to use 2-3 GiB of RAM, but I saw some error: writing to file: Broken Pipe in buildbot, I'm not sure what caused that :/ | 20:20:54 |
 magic_rb | In reply to @traxys:familleboyer.net
I tried to improve the situation for nixvim in 1882, empirically it seems to help as locally the build only seems to use 2-3 GiB of RAM, but I saw some error: writing to file: Broken Pipe in buildbot, I'm not sure what caused that :/ Im somewhat responsible for the recent buildbot work, ill try to identify where that came from, but my guess is that there is an empty attrset somewhere which buildbot did not agree with. No sure tho, ill take a definitive look tomorrow | 20:28:52 |
| Traxys | Yeah the strange thing is that is does not seem to be deterministic, as when re-triggering the build the same derivation passed | 20:30:55 |
| magic_rb | Yeah this is recent, ill need @zowoq to pull the logs from the journal to see if anything pops up there. This is definitely a bug that needs fixing, if you could open a issue on the buildbot-nix repo id appreciate it | 20:32:38 |
| Traxys | I'll do that! | 20:33:12 |
| magic_rb | Thanks, if you run into anything which looks like a implementation bug ping me | 20:33:41 |
| magic_rb | For the future | 20:33:53 |
 Matt Sturgeon | Anecdotally, I've seen broken pipe errors on nixvim's buildbot in the past. However I think I've only seen them when some other test has failed, leading to the broken pipe.
I may have also seen them when buildbot was having issues due to high load, but I can't remember definitively.
Either way, this is clearly something different, thanks for your interest! | 20:34:59 |
| Traxys | Opened the issue, added some context if it helps in any way (I think the context is unrelated, but in case it's not :eh:) | 20:41:07 |
| Traxys | (I looked at the repo and saw the PR to combine build reports, it might help us I think we had around ~300 checks in nixvim, we used a link farm to make it more manageable on github) | 20:44:06 |
| Traxys | * (I looked at the repo and saw the PR to combine build reports, it might help us I think we had around ~300 checks in nixvim, we used a link farm to make it more manageable on github, thanks :D) | 20:44:14 |
