ChanServ changed the topic of #nixus to: Nixus is an experimental deployment tool for NixOS systems - https://github.com/Infinisil/nixus - https://logs.nix.samueldr.com/nixus/
drakonis has quit [Quit: ZNC 1.8.1 - https://znc.in]
drakonis has joined #nixus
drakonis has quit [Quit: ZNC 1.8.1 - https://znc.in]
drakonis has joined #nixus
bkv has joined #nixus
bqv has quit [*.net *.split]
bkv is now known as bqv
drakonis has quit [Quit: ZNC 1.8.1 - https://znc.in]
drakonis has joined #nixus
<bqv> was hoping updating nix might fix it
<bqv> apparently still broken
<bqv> maybe i should make a bug... i'm just barely even sure what's going wrong
<infinisil> bqv: Solve bugs, don't make more!
<bqv> i wish i could, but this seems opaque as hell to me
<bqv> i get that something's going wrong with the protocol/transport
<bqv> that's it
<infinisil> bqv: Maybe you could try to figure out which commit introduced it
<infinisil> Because I'd assume this wasn't a problem at some point
<bqv> would be way easier if i didn't have to update both machine's nix's and nix-daemons
<bqv> actually... nix doesn't use the daemon if you're root, right?
<bqv> infinisil: oh huh
<bqv> you know what makes it work
<bqv> sshing as root
<bqv> amazing
<infinisil> Interesting
<bqv> yeah, works fine now
<bqv> guess i'll leave it at that
drakonis has quit [Quit: ZNC 1.8.1 - https://znc.in]
<bqv> i'd like a success action, for each host
<bqv> my previous deploy mechanism allowed me to git tag the system
<bqv> it'd be nice not to have to parse the output for that sort of thing
<infinisil> That might be a good idea, I'll think about it a bit
<infinisil> bqv: Probably something like https://github.com/NixOS/nixops/pull/1245
<bqv> hmm yeah, though perhaps that's a bit heavy handed
<infinisil> Ideally I'd rather have a good solution over a patchy one
<bqv> since at the moment you just have a script, it might make sense to just have json output of the full final state, so it's nice and pipelineable
<bqv> that way it remains nice and hackable and you don't gain a craptonne of systemd cruft
<infinisil> I mean ideally all machines are rolled back when something goes wrong
<infinisil> Such that exit code 0 means everything good, everything else means an error
<bqv> well, that works too, but that's not the case right now
<infinisil> Oh and actually, you can add deployment actions yourself
<bqv> via the deployScripts?
<infinisil> nodes.foo.deployScriptPhases = lib.dag.entryAfter [ "switch" ] '' <script> ''
<bqv> yeah, i thought of that, but it felt very wrong adding git commands that depend on PWD in there
<bqv> i suppose i could make it check the PWD first, too
<bqv> maybe i'll do that for now...
<infinisil> Or `git -C ${toString ../.} ..`
<infinisil> (or toString wherever your root is)
<bqv> yeah not sure i can accurately get that path from a flake
<bqv> but i can definitely check for it
<bqv> there's an elegance to nixus, i love it
<infinisil> Hehe
{`-`} has joined #nixus