| 7 Mar 2024 |
edef | like, what i want from the incentive alignment is avoidance of "no, we won't pay you to spend time hacking on making this system more stable, but we will pay you to respond to the emergencies this creates" | 00:06:25 |
adamcstephens 🐝 | makes sense to me | 00:07:20 |
Wanja Hentze | In reply to @edef1c:matrix.org part of the thing here is, you can prevent a lot of emergencies by doing consistent work to make them unlikely and handleable yes. I'm currently on-call. The last time I was paged was months ago, and it was a false positive. the time before that was years ago. | 00:07:30 |
adamcstephens 🐝 | that's the dream rotation | 00:07:45 |
raitobezarius | In reply to @whentze:matrix.org yes. I'm currently on-call. The last time I was paged was months ago, and it was a false positive. the time before that was years ago. do we need to make things spicier in nixos | 00:07:51 |
raitobezarius | sorry | 00:07:53 |
edef | In reply to @whentze:matrix.org yes. I'm currently on-call. The last time I was paged was months ago, and it was a false positive. the time before that was years ago. and thus it would be a reasonable bet for me to take that you work in an environment where those incentives are well-aligned | 00:08:01 |
adamcstephens 🐝 | i've definitely had on-call rotations that were much more painful than a couple times a year :) | 00:08:09 |
Wanja Hentze | if you do a good job on the infra and have reasonable requirements (this one is important), on-call can be relatively okay | 00:08:19 |
Wanja Hentze | that being said I would currently not join a 2nd rotation because I simply need time off for real once in a while | 00:11:49 |
Wanja Hentze | I can help with infra work though if you want me to beat a NixOS over the head or replace a hard drive or whatever | 00:13:21 |
@linus:schreibt.jetzt | Yeah I'm not sure "on-call" makes that much sense here anyway? I'd consider it to be a best-effort thing, and I expect that if stuff catches fire people will scramble to fix it, but nobody needs to be paged out of whatever they're doing for it | 00:15:25 |
Wanja Hentze | idk, crates.io has a rotation and I consider that less essential | 00:16:07 |
@linus:schreibt.jetzt | we also have the advantage of the cache not being super critical for nix/nixos to be usable | 00:16:16 |
@linus:schreibt.jetzt | like if it goes down for a day, that would be much less of a problem than the ubuntu package servers being down for a day would be for ubuntu | 00:17:10 |
@linus:schreibt.jetzt | ergh, I should sleep, not think about this, for now | 00:19:46 |
@linus:schreibt.jetzt | good night :D | 00:19:48 |
Wanja Hentze | eh, for a loooot of NixOS users, the cache being down means nixos-rebuild just becomes a desk warming app | 00:19:53 |
edef | yeah | 00:20:03 |
edef | from the cache logs, i know we have a lot of users in like, low-GDP countries you've never heard of before | 00:20:30 |
Wanja Hentze | we can actually build our closures on our CI infra but I'm sure many can't | 00:20:36 |
edef | i don't think they would appreciate the desk warming since it is coming out of their power bill | 00:20:46 |
Wanja Hentze | and in my hobby usage I definitely can't | 00:20:55 |
edef | nor do they likely have the RAM to actually build most things | 00:21:11 |
raitobezarius | In reply to @edef1c:matrix.org from the cache logs, i know we have a lot of users in like, low-GDP countries you've never heard of before i wonder if low GDP countries can download fast enough some of our stuff though | 00:21:31 |
raitobezarius | and if they don't end up passing --no-substituters or something | 00:21:42 |
edef | i'm working from cache logs, so they do seem to hit it | 00:22:01 |
edef | or they wouldn't show on my radar at all | 00:22:05 |
raitobezarius | ah right | 00:22:05 |
edef | bandwidth will hopefully improve in the long run if we can get deduping to work well and carry all the way through to the frontend protocol | 00:22:55 |