| 23 Jun 2021 |
 hexa | Redacted or Malformed Event | 14:11:44 |
| hexa | wouldn't expect it unless someone takes a special interest, the maintainers certainly won't | 14:12:28 |
 Linux Hackerman | I saw that, should I ask him about it? :> | 14:12:31 |
| hexa | lol :) | 14:12:36 |
| Linux Hackerman | he's just around the corner | 14:12:40 |
| hexa | ask him about his general maintainer status in nixpkgs | 14:12:46 |
 pennae | haven't trusted networkd too much since it kept segfaulting on our router config D: otherwise we'd probably use that too just for convenience | 14:15:54 |
| hexa | pennae: looks like it needs to have a user configured at compile time, which would mean some tie-in with a module creating that user | 14:16:20 |
| hexa | networkd runs pretty well on multiple machines here, even on one carrying a bgp fulltable | 14:18:29 |
| pennae | never quite found out why it segfaulted here either, but it seemed to dislike vlan ifs in bridges for some reason | 14:19:44 |
| pennae | but only sometimes | 14:20:04 |
| hexa | that is a pretty common setup, best take it to #networking:nixos.org | 14:22:17 |
 kraftnix | I also have seg faults with networkd when building up network configurations sometimes as well. The deepest I got on the debugging was that for some reason when I had a bad networkd config (some bad nix code), /etc/ld-nix.so.preload was missing when running the nix build, which causes a seg fault. The only work around I had was rolling back changes until I found the bad piece of code. I am still not unsure whether it's directly related to networkd or not, but it does happen when I am writing networkd configurations | 14:23:12 |
| hexa | rough. can we move it to #networking:nixos.org still? | 14:24:00 |
| hexa | * rough. can we move it to #networking:nixos.org still? if both of you have issues like this it should probably be investigated further. | 14:24:21 |
| hexa | pennae: on the topic of privsep, dhcpcd should get a privsep user when we can ensure it is used through the module, which in turn ensures the user exists | 14:26:14 |
| hexa | at one point there as a dhcpcd user allocated, see e9cd877921cc4164aa0ba57ae1fd7526ec6440c3 | 14:28:37 |
| hexa | * at one point there even was a dhcpcd user allocated, see e9cd877921cc4164aa0ba57ae1fd7526ec6440c3 | 14:28:40 |
| pennae | that's been a while | 14:29:06 |
| hexa | would you want to look into that further? | 14:29:12 |
| pennae | not sure we're qualified | 14:29:49 |
| hexa | can only encourage you to try 😀 | 14:34:18 |
| pennae | might try at some point. the last module we tried to modify kind of fizzled in review 😶 | 14:35:38 |
| hexa | there are even nixos tests that you can use to verify you didn't break anything :) | 14:35:42 |
| pennae | grubbing around in system daemons is a bit scary tbh
at least mosquitto isn't that important (and also had tests (that were half broken)) | 14:39:27 |
| hexa | mosquitto was also haphazardly bumped a major version number just before the release | 14:51:25 |
| hexa | ignoring the resulitng breakage, not nice. | 14:51:43 |
| pennae | and our PR to fix the module being extremely limited has seen basically no review since ... when was it, beginning of may? | 14:53:53 |
| pennae | at leas the bump didn't break anything here :/ | 14:54:31 |
| pennae | oh hey, you're on that one as a reviewer hexa 😛 | 15:26:13 |
