| 21 Jul 2024 |
 @aloisw:kde.org | In reply to @magic_rb:matrix.redalder.org
How does nixpkgs contribute to the problem? Nixpkgs derivations remember too much of their inputs. | 08:29:44 |
 Traxys | zowoq we merged the PR that reduced the memory usage. We'd need to ask all open PRs to rebase on main, and we should be good to re-enable buildbot | 13:22:20 |
 Matt Sturgeon | Was our stable branch affected by the issue?
Do we need to disable buildbot on that branch until the change is backported, or since it is a low traffic branch is it enough that we simply avoid triggering any CI builds on stable for now? | 13:39:16 |
| Traxys | I don't really know :/ | 14:00:57 |
| Matt Sturgeon | I'm out and about so can't check right now, but we can try running the eval command on our stable branch to see?
nix run nixpkgs#nix-eval-jobs -- --workers 1 --check-cache-status --force-recurse --flake 'github:nix-community/nixvim/nixos-24.05#checks'
Or just enable buildbot and test it on the branch when someone's around to restart it if it crashes? | 14:44:58 |
 zowoq | I think you'll need https://github.com/nix-community/buildbot-nix/pull/217 as well to avoid the rate limit problems https://github.com/nix-community/buildbot-nix/issues/178? | 22:24:25 |
| 22 Jul 2024 |
| Traxys | When I launched a job I did not encounter that, but that may be because it was a single job, and if we switched on buildbot we would have issues due with many PRs | 08:11:48 |
| Traxys | (and I think our stable branch is affected too, so we need to backport some of this) | 08:12:07 |
 Clumsily6239 | Can I login into a user without triggering their ~/.config/autostart and systemd user services? | 12:38:36 |
 K900 | Depends on your definition of "login" | 12:38:58 |
| Clumsily6239 | the standard login from tty? I thought that was the only "login". | 12:40:09 |
| K900 | Log in as root, sudo -u <your user> | 12:40:39 |
| K900 | e.g. | 12:40:43 |
| Clumsily6239 | Awesome, thank you. | 12:41:35 |
| 23 Jul 2024 |
| zowoq | https://docs.github.com/en/apps/oauth-apps/building-oauth-apps/differences-between-github-apps-and-oauth-apps
The rate limit for GitHub Apps using an installation access token scales with the number of repositories and number of organization users. Conversely, OAuth apps have lower rate limits and do not scale.
Maybe the size of nix-community means the rate limits aren't a problem.
| 01:05:52 |
| zowoq | I checked stable and it seems to use a lot less memory than it was on main so I've enabled buildbot again. | 01:06:07 |
 Gaétan Lepage | Thanks a lot zowoq ! | 06:05:01 |
|  Ezzobir Bezziou joined the room. | 08:23:46 |
 emily | does the community builder push things to a binary cache? | 08:33:24 |
| zowoq | No, only the CI builds are pushed to the cache. | 09:08:25 |
| emily | gotcha | 09:33:31 |
| emily | I guess there's nothing stopping you using Cachix yourself on the builder to push to your own cache | 09:33:54 |
 Mic92 | @emilazy:matrix.org: with so many trusted users the binary cache would become a bit dodgy. | 11:19:08 |
| emily | indeed | 11:19:30 |
| emily | my use-case is transporting this digiKam build into a throwaway VM to test it :) | 11:19:41 |
| emily | but I'll just nix-copy-closure | 11:19:49 |
| Mic92 | It's cooler if we don't need to trust everyone as much and instead provide access more freely | 11:20:02 |
| emily | (I don't even set the community builder as a remote builder, I just SSH in to build stuff) | 11:20:30 |
| emily | thanks again for all the work running these builders btw, and sorry for building multiple WebKits lately :) | 11:21:18 |
| Mic92 | nix-fast-build can automate the copy also. But unsure if it fits your workflow | 11:21:26 |
