| 19 Feb 2023 |
 @elvishjerricco:matrix.org | and we implement it in terms of the x-systemd.requires-mounts-for= FS option | 16:20:05 |
 @lily:lily.flowers | That actually sounds like a pretty good idea | 16:23:31 |
| @lily:lily.flowers | I like that | 16:23:42 |
| @elvishjerricco:matrix.org | issue about the depends thing: https://github.com/NixOS/nixpkgs/issues/217179 | 16:25:42 |
| @elvishjerricco:matrix.org | I think I found a way to do the flushBeforeStage2 thing for the networkd PR, but networkctl down is neither failing nor taking interfaces down... am confused | 22:29:30 |
| @elvishjerricco:matrix.org | it requires bringing dbus into initrd because that's how networkctl interacts with networkd | 22:31:12 |
| @elvishjerricco:matrix.org | and that was annoying in its own right | 22:31:17 |
| @elvishjerricco:matrix.org | and then it just... almost worked | 22:31:27 |
| @lily:lily.flowers | In reply to @elvishjerricco:matrix.org
and then it just... almost worked Anything interesting in logs? | 22:34:43 |
| @elvishjerricco:matrix.org | not that I can see | 22:34:50 |
| @elvishjerricco:matrix.org | though I haven't tried with systemd debug log level yet | 22:35:01 |
| @lily:lily.flowers | In reply to @elvishjerricco:matrix.org
it requires bringing dbus into initrd because that's how networkctl interacts with networkd (also does this work with services.dbus.implementation = "broker"?) | 22:35:03 |
| @elvishjerricco:matrix.org | oh my initrd implementation is far too dumb to care for that option lol | 22:35:21 |
| @elvishjerricco:matrix.org | * oh my initrd dbus implementation is far too dumb to care for that option lol | 22:35:27 |
| @lily:lily.flowers | In reply to @elvishjerricco:matrix.org
though I haven't tried with systemd debug log level yet Yeah prob try that. Looking at networkctl code I don't see anything obvious unless networkd is just ignoring dbus commands totally or something | 22:35:31 |
| @lily:lily.flowers | In reply to @elvishjerricco:matrix.org
oh my initrd dbus implementation is far too dumb to care for that option lol I figured so, just wanted to make sure it wouldn't blow up with it :) | 22:35:41 |
| @elvishjerricco:matrix.org | yea next thing is too look at the logs | 22:36:08 |
| @lily:lily.flowers | If you want to push to a public branch, I might have time to tinker with it too. I'm assuming you're doing this in a nixosTest | 22:36:18 |
| @elvishjerricco:matrix.org | kind of | 22:36:39 |
| @elvishjerricco:matrix.org | I just added a sleep 5 && networkctl status command to the end of a nixos test and have been observing that output :p | 22:37:22 |
| @elvishjerricco:matrix.org | er, to the end of the service I made to bring down interfaces | 22:37:49 |
| @elvishjerricco:matrix.org | so it definitely happens in initrd after the networkctl down commands are run | 22:38:40 |
| @elvishjerricco:matrix.org | and they still show as configured.. | 22:38:50 |
| @elvishjerricco:matrix.org | * and they still show as configured... | 22:38:51 |
| @elvishjerricco:matrix.org | Hm but this would be a good test to have. Due to... particulars, I would have to make an ExecStopPost line that would trigger emergency should it fail | 22:41:00 |
| @lily:lily.flowers | In reply to @elvishjerricco:matrix.org
and they still show as configured... I mean they'll still show as configured after going down, right? Just the operational state would change I think | 22:41:57 |
| @elvishjerricco:matrix.org | oh, that's news to me | 22:42:20 |
| @elvishjerricco:matrix.org | I thought down would bring them to an unconfigured state, releasing their DHCP leases and whatnot | 22:42:37 |
| @elvishjerricco:matrix.org | I have a system to test that on... one sec | 22:42:45 |
| @lily:lily.flowers | Ah, confusingly configured is also an operational state | 22:42:53 |
