03:15
cole-h has quit [Ping timeout: 264 seconds]
03:17
cole-h has joined #nixops
07:10
kaliumxyz has quit [Remote host closed the connection]
07:12
cole-h has quit [Quit: Goodbye]
07:20
andi- has quit [Ping timeout: 260 seconds]
07:27
kaliumxyz has joined #nixops
08:05
teto has joined #nixops
08:10
tewfik has joined #nixops
08:11
<
tewfik >
DigitalKiwi: sure thing ;)
08:11
tewfik has quit [Client Quit]
09:13
<
adisbladis >
Anyone up for a review call today? :)
09:21
<
{^_^} >
nixops#1331 (by dhess, 6 weeks ago, open): eval-machine-info.nix: add nixpkgsPath argument.
09:59
meh` has joined #nixops
10:31
andi- has joined #nixops
10:37
<
manveru >
adisbladis: is there anything controversial about it? i'm all for this change :D
10:38
<
manveru >
just wish it'd have happened years ago
10:40
<
manveru >
maybe i'd even display a warning when it's not passed: "Your deployment depends on <nixpkgs>, which is known to cause different behaviour depending on your channel configuration" or something
10:44
<
adisbladis >
manveru: I'm not sure the design is right
10:44
<
adisbladis >
And I'm not clear on how it relates to flakes
10:45
<
adisbladis >
Mostly I just want to involve more people in decision making & reviewing =)
10:46
<
manveru >
yeah, this is pretty much a conflict with the flakes PR
10:47
<
manveru >
but mostly because your PR still relies on <nixpkgs> :P
10:47
<
manveru >
which is evil
10:49
<
adisbladis >
Yeah, it's a bit of a bootstrapping issue going on here
10:54
<
adisbladis >
I think there are som quick wins to at least reduce the dependency on <nixpkgs>
11:27
<
{^_^} >
nixops#1368 (by adisbladis, 7 minutes ago, open): Minimise the dependency on <nixpkgs>
11:27
<
adisbladis >
It was pretty easy to get the dependency down to only `zipAttrs`
11:28
<
manveru >
zipAttrs isn't exactly a big function :)
11:30
<
adisbladis >
Nah, I just want to avoid copy-pasting the entire chain of functions leading up to it
11:33
<
adisbladis >
It's only 8 lines actually
11:33
<
adisbladis >
Once you get rid of the old nix compat bits
11:33
<
adisbladis >
Alright, pushed :)
11:34
<
manveru >
there we go :)
11:34
<
manveru >
could be even smaller, but whatever
11:35
<
adisbladis >
It's literally copy-paste from nixpkgs
11:35
<
adisbladis >
If you want to improve something suggest a change ;)
11:42
<
manveru >
`zipAttrs = set: __listToAttrs (map (name: { inherit name; value = __catAttrs name set; }) (__concatMap __attrNames set))`
11:42
<
manveru >
well, change that __ to builtins if you like
11:44
<
adisbladis >
manveru: Nice, thanks
11:45
<
manveru >
nixops is really becoming a bit handsome now :)
11:45
<
manveru >
i can't believe how much work you guys put into it
11:48
<
adisbladis >
Happy to hear that
11:49
<
manveru >
btw, is there any news on the deploy hooks front?
11:49
<
adisbladis >
You mean nixops#1245 ?
11:50
<
manveru >
hm, that's on the server side
11:50
<
manveru >
i mean, the target server
11:50
<
adisbladis >
Please elaborate :)
11:51
<
manveru >
i need hooks on the nixops side to run stuff before deploy based on the network config :)
11:51
<
adisbladis >
nixops#1318 ?
11:51
<
adisbladis >
I'm planning on getting that one merged today
11:52
<
manveru >
ah, so that would be via plugins
11:52
<
manveru >
do i really have to write python again? :P
11:52
<
adisbladis >
Explain your use case a bit more please :)
11:53
<
manveru >
well, we modified our nixops to have a `preDeploy` attribute for each machine, which specifies some script that does whatever...
11:54
<
adisbladis >
Ah, we could make a plugin for that functionality :)
11:54
<
manveru >
then we collect them using `nixops eval` and run them before deploy
11:54
<
manveru >
i see my eval pr wasn't merged either... and has conflicts now :|
11:54
<
manveru >
it had a bug anyway
11:57
<
manveru >
but yeah, it might be more suited to a nixops plugin indeed, so we don't have to eval the config twice
11:58
<
manveru >
though that's less of a performance issue with flakes :)
11:59
<
adisbladis >
manveru: If you want to we could have a chat about this. I'd love to understand your use case better.
11:59
<
adisbladis >
I really want plugin hooks to be able to provide what you need.
12:01
<
manveru >
so atm we have two major uses of the hooks
12:02
<
manveru >
one is to generate wireguard keys for the whole network, the other to fetch secrets from vault to put into the nixops secrets...
12:04
<
adisbladis >
The first one sounds
_very_ similar to the old encrypted links
12:05
<
manveru >
just that we need wireguard so people can connect from their home machines :)
12:05
<
manveru >
yep, that'd be how it's done
12:05
<
adisbladis >
Yeah, WG is obviously better, I'm just observing the very similar needs
12:06
<
adisbladis >
Or maybe you'd want a per-machine hook ?
12:07
<
adisbladis >
You can still iterate over machines from the Deployment object
12:07
<
adisbladis >
And then you'd only need to fetch once for your entire deployment and stick those attributes on each machine
12:07
<
adisbladis >
So maybe what you have there is good enough?
12:08
<
adisbladis >
Actually I just remembered nixops#1280
12:09
<
manveru >
that is also damn nice
12:10
<
adisbladis >
So for secrets I think keyCommand is the much better option
12:10
<
gchristensen >
nixops probably shouldn't go out of its way to support generating keys in a central place
12:10
<
manveru >
yep, definitely beats writing a plugin :D
12:11
<
manveru >
thanks a lot
12:11
<
adisbladis >
gchristensen: Nah, I think the post_wait hook introduced in nixops#1318 is better
13:32
<
adisbladis >
Is anyone up for a call? It feels wrong to merge my own PRs without anyone else reviewing them =)
14:09
<
manveru >
adisbladis: well, i can join you for a bit
14:09
<
manveru >
if it's not longer than half an hour or so :)
14:55
meh` has quit [Ping timeout: 240 seconds]
14:55
meh` has joined #nixops
15:08
virus_dave has joined #nixops
15:08
virus_dave has quit [Client Quit]
15:08
virus_dave has joined #nixops
15:10
virus_dave has joined #nixops
15:10
virus_dave has quit [Client Quit]
15:10
virus_dave has joined #nixops
15:10
virus_dave has quit [Client Quit]
16:09
meh` has quit [Ping timeout: 256 seconds]
17:10
teto has quit [Quit: WeeChat 2.8]
17:47
cole-h has joined #nixops
17:59
meh` has joined #nixops
18:29
teto has joined #nixops
18:47
meh` has quit [Ping timeout: 256 seconds]
19:25
meh` has joined #nixops
21:29
tokudan has quit [Remote host closed the connection]
21:32
tokudan has joined #nixops
22:00
globin has joined #nixops
22:11
<
cole-h >
adisbladis++
22:11
<
{^_^} >
adisbladis's karma got increased to 100
22:54
meh` has quit [Ping timeout: 256 seconds]
23:17
virus_dave has joined #nixops
23:18
virus_dave has joined #nixops
23:18
virus_dave has quit [Client Quit]
23:19
virus_dave has joined #nixops
23:19
virus_dave has quit [Client Quit]
23:19
virus_dave has joined #nixops
23:20
virus_dave has joined #nixops
23:20
virus_dave has quit [Client Quit]
23:21
virus_dave has joined #nixops
23:21
virus_dave has quit [Client Quit]
23:21
virus_dave has joined #nixops
23:21
virus_dave has quit [Client Quit]
23:22
virus_dave has joined #nixops
23:23
virus_dave has joined #nixops
23:23
virus_dave has quit [Client Quit]
23:23
virus_dave has joined #nixops
23:24
virus_dave has quit [Client Quit]
23:24
virus_dave has joined #nixops
23:24
virus_dave has quit [Client Quit]
23:25
virus_dave has joined #nixops
23:25
virus_dave has quit [Client Quit]
23:26
virus_dave has joined #nixops
23:26
virus_dave has quit [Client Quit]
23:26
virus_dave has joined #nixops
23:26
virus_dave has quit [Client Quit]
23:27
virus_dave has joined #nixops
23:27
virus_dave has quit [Client Quit]
23:28
virus_dave has joined #nixops
23:28
virus_dave has joined #nixops
23:29
virus_dave has quit [Client Quit]
23:29
virus_dave has joined #nixops
23:29
virus_dave has quit [Client Quit]
23:30
virus_dave has joined #nixops
23:30
virus_dave has quit [Client Quit]
23:31
virus_dave has joined #nixops