<Lisanna>
hey, would anyone have any ideas about why my hydra channel's binary cache isn't working? It only works if I explicitly add it using --option binary-caches
<clever>
Lisanna: did you add it to nix.conf?
<Lisanna>
and the binary-cache-url file is accessible
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/f4yyh
<clever>
Lisanna: cache.nixos.org is a default hard-coded into nix itself
<Lisanna>
ugh
<clever>
for when nix.conf is missing
<Lisanna>
wonderful
<clever>
you will have to add the binary cache to nix.conf
<Lisanna>
when I had a channel and binary cache setup manually in the past it would work
<Lisanna>
but since moving to hydra I haven't been able to get it to go
<Lisanna>
yeah I know I can add it to nix.conf, but isn't it supposed to be automatic?
<clever>
i thought it would be too, but the code says the last piece is missing
<Lisanna>
otherwise what's the point of the binary-cache-url file
<clever>
Lisanna: unfinished plans?
<Lisanna>
"A file containing the URL to a binary cache (such as https://cache.nixos.org). Nix will automatically check this cache for pre-built binaries, if the user has sufficient rights to add binary caches."
<Lisanna>
that's in the nix-channel manpage
deepfire` has quit [Ping timeout: 264 seconds]
<clever>
are you a running these commands as root?
<clever>
Lisanna: there WAS a binary-caches-files option, that was a list of files containing binary caches
<Lisanna>
ok, and it defaulted to whatever root had installed
<clever>
`git log --patch` claims that option never actually worked
<clever>
and it was already commented out in the docs
<clever>
no code can be found that ever made this work
<clever>
time for you to file a bug!
<Lisanna>
lol
<Lisanna>
but this is insane
Guest7961 has quit [Remote host closed the connection]
<Lisanna>
no way something like this would get overlooked
<Lisanna>
.......right???
<Lisanna>
:/
<Lisanna>
*greps through the open issues*
<elvishjerricco>
Why would adding a `plugin-files` line to `nix.extraOptions` cause `illegal configuration line 'plugin-files ...'`?
<tester123>
Haskell stack problem: 1. Running `stack --version` yields "Version 1.6.3 x86_64". 2. But I need Version >=1.7. 3. The current version on nixpkgs search is "1.6.5". 4. I think normally on non-NixOS you could just run "stack upgrade" in your home folder, but that doesn't seem to work. What do?
<clever>
elvishjerricco: what version of nix are you running?
<Lisanna>
elvishjerricco forget the equals sign?
<elvishjerricco>
clever: I've tried it with `nix` and `nixUnstable` from both `channel:nixos-18.03` and `channel:nixpkgs-unstable`
<Lisanna>
(I do that all the time)
<elvishjerricco>
...
<elvishjerricco>
i forgot the equals sign
<Lisanna>
ha
justbeingglad has joined #nixos
<elvishjerricco>
Lisanna: Thanks :P
JasonGrossman has quit [Remote host closed the connection]
testuser has quit [Ping timeout: 256 seconds]
ryantrinkle has joined #nixos
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
cdepillabout has joined #nixos
rschm has quit []
<cdepillabout>
I have a C library I'm trying to write a nix expresssion for. It is using cmake to build and it is linking to the protobuf library.
<cdepillabout>
It seems to be building fine. It correctly produces a lib*.so file. Running nix-build finishes without error.
<cdepillabout>
However, when I run `ldd` on the resulting lib*.so file, it shows that the libprotobuf.so.14 lib can't be found: "libprotobuf.so.14 => not found"
<cdepillabout>
Is this expected behavior? Is something strange going on?
TacNuke has quit [Read error: Connection reset by peer]
andymandias has quit [Ping timeout: 268 seconds]
aarvar has joined #nixos
roconnor has quit [Ping timeout: 240 seconds]
roconnor has joined #nixos
andymandias has joined #nixos
<cdepillabout>
Hmm, after playing around with this a little, it looks like one of my dependencies (from nixpkgs) is using a different version of protobuf than the one I am specifying.
<cdepillabout>
I guess I need to rebuild all of my dependencies with the older version of protobuf I want to use, but I'm not sure how to write that.
fragamus has joined #nixos
realrokka has quit [Read error: Connection reset by peer]
realrokka has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos
<cdepillabout>
I'm importing <nixpkgs> in my default.nix file (well, actually taking it as an argument, and if it is not passed in, importing it), and I would like to override the protobuf version in nixpkgs.
asuryawanshi has quit [Ping timeout: 240 seconds]
phreedom has quit [Remote host closed the connection]
phreedom has joined #nixos
jperras has quit [Quit: WeeChat 2.1]
Jason_Grossman has joined #nixos
Jason_Gr` has joined #nixos
justbeingglad has joined #nixos
justbeingglad has left #nixos [#nixos]
<vaibhavsagar>
can you use an overlay to override protobuf in nixpkgs?
<vaibhavsagar>
or maybe just the protobuf version of the dependency that is using a different one
<cdepillabout>
@vaibhavsagar, I'll have to look into using overlays. I've never used them yet.
<cdepillabout>
Part of my problem is that I'm not sure which dependency is pulling in a different version of protobuf. So I'm not sure which to change.
<cdepillabout>
Oh great, thanks! I'll try that.
<vaibhavsagar>
if you grep for "protobuf =" in the nixpkgs source there are a couple of places where the default protobuf is overridden
<vaibhavsagar>
so you could use the same overlay method to switch out the protobuf version that is passed to whichever dependency you want to change
Jason_Grossman has quit [Read error: Connection reset by peer]
roconnor has quit [Remote host closed the connection]
lord| has quit [Quit: WeeChat 2.1]
roconnor has joined #nixos
lassulus_ has joined #nixos
sbdchd has quit [Remote host closed the connection]
iyzsong has joined #nixos
<bluej774>
When building nix, do you have to do something special to build it with statically?
lassulus has quit [Ping timeout: 268 seconds]
lassulus_ is now known as lassulus
<bluej774>
*statically linked?
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
worldofpeace has quit [Quit: Konversation terminated!]
worldofpeace has joined #nixos
rprije has joined #nixos
Jason_Gr` is now known as Jason_Grossman`
<tim`>
Hi, I'm trying to deploy to a machine using `nixops deploy -dname`, and the process exits with the error pasted here: https://lpaste.net/6314792580248764416 . I'm not sure how to begin debugging this -- any tips?
fragamus has joined #nixos
justanotheruser has joined #nixos
Jason_Grossman` has quit [Remote host closed the connection]
spear2 has joined #nixos
ericsagn1 has quit [Ping timeout: 256 seconds]
<clever>
tim`: looks like a systemd issue on the remote end, try rebooting that machine
Jason_Grossman has joined #nixos
lord| has joined #nixos
iyzsong has quit [Read error: Connection reset by peer]
mizu_no_oto has joined #nixos
<tim`>
clever: that fixed it, thank you!
worldofpeace has quit [Remote host closed the connection]
mizu_no_oto has quit [Quit: Computer has gone to sleep.]
ndowens04 has joined #nixos
tzemanovic has joined #nixos
<ndowens04>
this is weird, I ssh into this machine and it will give errors : http://dpaste.com/1M710P5
tzemanovic has quit [Ping timeout: 264 seconds]
<ndowens04>
nvm seems to be just shell type
bluej774 has quit [Quit: Leaving]
roconnor has quit [Remote host closed the connection]
roconnor has joined #nixos
roconnor has quit [Remote host closed the connection]
roconnor has joined #nixos
Jason_Grossman has quit [Ping timeout: 264 seconds]
ndowens04 has quit [Quit: WeeChat 2.1]
nvh^ has quit [Ping timeout: 256 seconds]
silver_hook has quit [Ping timeout: 256 seconds]
coffeecupp__ has joined #nixos
blankhart has quit [Remote host closed the connection]
<mjrosenb>
if I have a foo.nix in a project that I'm working on, and it depends on a library (snap), which depends on another library (heist), which is kind of broken; it requires an older version of a library than gets provided by default. is there a way I can request it to be built against an older library just for this program?
tertle||eltret has quit [Quit: Connection closed for inactivity]
Xiro` has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
tzemanovic has joined #nixos
<Lisanna>
so, in a nix machines file, the maximum number of jobs is local to each submitter, right? so, if alice and bob both have example.com with 4 as the maximum number of builds, then that means that each one can submit 4 for a total of 8? what happens if the example.com machine has nix.maxJobs set?
<Lisanna>
trying to figure out to create the ideal remote building setup that is shared by several users, so far have not found any clear documentation on this
tzemanovic has quit [Ping timeout: 248 seconds]
<clever>
Lisanna: i believe the remote nix-daemon will block until a slot is free
<Lisanna>
so, nix.maxJobs is the total number of jobs that the local nix-daemon can be running, regardless of feature types, and the max jobs number in machine files are local self-imposed limits?
<Lisanna>
(and therefore less important?)
<clever>
Lisanna: ive also noticed in older versions (not sure if its changed), but a remote "slave" will sometimes forward things to its own slaves, so you could use a single slave as a hub to load-balance between more
<clever>
yeah
<Lisanna>
clever I would love it if that was the behavior, I could setup a forwarding machine with more fine-grained control over limits, although the features would have to be preserved
<Lisanna>
and localhost building is bugged ):
<clever>
Lisanna: i think it would preserve features
<clever>
but that forwarding might have been a nix1 "bug"
<Lisanna>
bleh
<clever>
you will have to test nix2 and see what it does
<clever>
cannot build on 'ssh://clever@192.168.2.167': cannot connect to 'clever@192.168.2.167': ssh: connect to host 192.168.2.167 port 22: No route to host
<clever>
error: a 'x86_64-darwin' is required to build '/nix/store/rwm1hd7lxa1kmlgbgqbdvb3by9rd4xr9-name.txt.drv', but I am a 'x86_64-linux'
<clever>
but my mac is off, so the build fails
<Lisanna>
I dunno about writing a multiplexer, don't have the resources to do that sort of thing, just trying to figure out the most ideal setup for shared build resources
<Lisanna>
I'm assuming that I can choose each client's build machines file values
<Lisanna>
eh, actually, those would really actually only be helpful with the nix1 "bug" I guess
<clever>
ok, i have configured my laptop to think the nixos desktop is a mac slave
<clever>
then i configured the desktop to use an actual mac slave
<Lisanna>
interesting
<Lisanna>
so, laptop forwards darwin builds to the nixos desktop, which then forwards to a real darwin machine?
<clever>
in theory, yes
tzemanovic has joined #nixos
winem_ has joined #nixos
<clever>
Lisanna: the laptop is now copying inputs to the desktop...
<clever>
error: build of '/nix/store/11vdgj2s54nzryzj3snl3v1fyb85v0p5-name.txt.drv' on 'ssh://root@192.168.2.15' failed: a 'x86_64-darwin' is required to build '/nix/store/11vdgj2s54nzryzj3snl3v1fyb85v0p5-name.txt.drv', but I am a 'x86_64-linux'
<clever>
that is a very nix2 style error
<clever>
it may require a nix1 daemon on the central hub
pie_ has quit [Read error: Connection reset by peer]
<Lisanna>
so, I guess a follow-on question I'd have is: if I set 5000 as the max jobs in a local machine file, but set nix.maxJobs on the target machine to 10, and launch 5000 jobs, will any of those 500 of those get queued on that one machine, refusing to be offloaded to others even if no slots are available locally on that machine?
<Lisanna>
guess it's time for some testing
<Lisanna>
5000*
<clever>
it will probably run out of build users and hard-fail
<clever>
i think it needs to allocate a nixbld1 user before it can block on maxJobs
<clever>
and the number of build users is dynamicaly based on the bigger value, between 32, and maxJobs
<clever>
Lisanna: cant seem to reproduce that proxy like "feature"
<Lisanna>
darn
<clever>
1.10 doesnt like the new nix schema, and 1.11+ wont do it
<clever>
you would need a 17.09 VM to test older versions
<Lisanna>
wish remote building just worked like a normal job scheduler
<clever>
you may have better luck just using hydra
<clever>
push all code to github, let hydra build it all and manage the queue
<Lisanna>
my use case is multiple users needing to run builds using special features / system resources
<Lisanna>
interactively
<Lisanna>
like, nix-build -A myThing
<clever>
just set some limits and tell users to retry when things fail?
<Lisanna>
gonna experiment to see how the remote building handles some odd limits
* clever
heads off to bed, 4am!
winem_ has quit [Ping timeout: 264 seconds]
joelpet has quit [Ping timeout: 240 seconds]
<Lisanna>
goodnight ^^
<Lisanna>
wtd
<Lisanna>
wtf*
<Lisanna>
nix.maxJobs does nothing
<Lisanna>
I set it to 2, but I'm able to locally launch 4 builds in parallel
<bluej774>
How do I make a nix release? I've tried `make dist` but that complains about missing target perl/configure.
inquisitiv3 has joined #nixos
<tilpner>
bluej774 - Look at release.nix
testuser has joined #nixos
<tilpner>
E.g. nix-build release.nix -A tarball
<bluej774>
Ah, okay, so I need to do it from inside an existing nix system.
<bluej774>
How would I cross-compile it?
<bluej774>
Like, let's say I'm on a x86_64 system, but I want to make a aarch64 tarball.
rprije has quit [Ping timeout: 260 seconds]
<tilpner>
nix-build release.nix -A build.aarch64-linux
<tilpner>
Maybe that'll work with qemu-user setup properly
hotfuzz has quit [Ping timeout: 245 seconds]
<bluej774>
Hmm... I'll give it a show.
<bluej774>
*shot
pemeunier has quit [Ping timeout: 255 seconds]
davidak has quit [Ping timeout: 240 seconds]
hotfuzz has joined #nixos
pemeunier has joined #nixos
davidak has joined #nixos
<MichaelRaskin>
Lisanna: I think maxJobs is about what Nix client will ask as a part of a single invokation, not the maximum number of parallel builds
<Lisanna>
ah
<Lisanna>
the nix manual sort of just mentions that system load is taken into account in a very hand-wavy way and doesn't give any details. can I trust it to load-balance stuff in a sane way?
<Lisanna>
there doesn't seem to be any way to globally control this
andymandias has quit [Ping timeout: 260 seconds]
<MichaelRaskin>
Oh the global load
<MichaelRaskin>
It is more about build-cores
<MichaelRaskin>
Which passes both -j and -l
<MichaelRaskin>
Sometimes
<MichaelRaskin>
Package buildsystem dependent
pemeunier has quit [Ping timeout: 240 seconds]
<Lisanna>
There's "cores" which is how many concurrent tasks per build, then there's "max-jobs" which is how many jobs can be launched at a time for a single build
<MichaelRaskin>
And cores might affect how different builds are dividing the global core pool
<Lisanna>
a resource manager is the right tool for managing this kind of stuff
<Lisanna>
like slurm
<MichaelRaskin>
Sometimes people talk of removing drv files as a notion; I am slighlty afraid of that, because it would take away our last chance for a tunable build multiplexer
illabout has joined #nixos
palo has joined #nixos
palo has quit [Changing host]
palo has joined #nixos
ericsagnes has joined #nixos
cdepillabout has quit [Ping timeout: 245 seconds]
deepfire` has joined #nixos
knupfer has joined #nixos
<Lisanna>
OK, so nix.nrBuildUsers looks like nix.nrBuildUsers is the only way to set a global limit on how many builds can be executing at once
<Lisanna>
since each build needs its own build her
<Lisanna>
user
<Lisanna>
wish I could assign specific nix build users to system features
joelpet has joined #nixos
joelpet has quit [Changing host]
joelpet has joined #nixos
blankhart has quit [Ping timeout: 264 seconds]
Jason_Grossman has joined #nixos
sir_guy_carleton has quit [Quit: WeeChat 2.0]
spear2 has quit [Remote host closed the connection]
bluej774 has quit [Quit: Leaving]
et4te has joined #nixos
fader has quit [Ping timeout: 256 seconds]
<LnL>
MichaelRaskin: only the physical files AFAIK
et4te has quit [Ping timeout: 248 seconds]
<MichaelRaskin>
LnL: The problem is that without physical files we won't have any clear way to refer it in an cross-user way
knupfer has quit [Ping timeout: 245 seconds]
<LnL>
I would assume will still be a way to reference derivations
justbeingglad has joined #nixos
justbeingglad has left #nixos [#nixos]
FRidh has quit [Read error: Connection reset by peer]
<inquisitiv3>
I'm planning to install NixOS on my laptop as my daily driver. Do I need to install the drivers for Intel's wifi chip manually or are they automatically included in the hardware configuration?
<tilpner>
inquisitiv3 - I think nixos-generate-config gets it right
<illabout>
I'm trying to work on converting a somewhat-complicated multi-project program at work to build with Nix. A few of the sub-projects (sub-packages?) are Haskell projects. Up until now, we've been using stack to build the Haskell projects. They are all being built with the same stackage lts (lts-10.1). Is there a way to easily use the same Haskell dependency package versions from an lts release when building with Nix?
<illabout>
For instance, let's say that nixpkgs is using version 4.16 of lens, but lts-10 is using version 4.15 or lens. I'd like to use lens-4.15 when building my Haskell packages instead of 4.16 as defined by the nixpkgs I'm using.
klntsky has quit [Remote host closed the connection]
klntsky has joined #nixos
<inquisitiv3>
tilpner: Nice! One less thing to think about :)
<Jason_Grossman>
inquisitiv3: You MIGHT have to set allowUnfree = true;
<inquisitiv3>
Jason_Grossman: Okey, I'll keep that in mind. Thanks for the heads up!
<inquisitiv3>
Is there any way to check that?
<inquisitiv3>
Jason_Grossman: ^
stepcut has quit [Remote host closed the connection]
<Jason_Grossman>
inquisitiv3: I don't know a way. Safest to add the setting, then try removing it (after you've got a working system to roll back to).
stepcut has joined #nixos
<Jason_Grossman>
inquisitiv3: The great think about NixOS is that it's easy to do that.
yellowbrick has joined #nixos
<Jason_Grossman>
* thing
<yellowbrick>
hi all. My epson perfection v500 photo is listed in the epkowa backend, but even adding hardware.sane.extraBackends = [ pkgs.epkowa ]; it still does not work. Anything come to mind? :)
<inquisitiv3>
Jason_Grossman: Okey, I'll have to do that. Didn't find any answer after a quick Google search.
<benny>
bendlas: cask is still broken, you're executing cask-package-toolset which is not the upstream cask script
stepcut has quit [Ping timeout: 255 seconds]
rprije has quit [Ping timeout: 240 seconds]
<Jason_Grossman>
inquisitiv3: Have fun!
<benny>
I've removed the reference to a package in all-packages.nix but I'm still able to install it with nix-env -I $NIXPKGS -i <package>, how can that be?
orivej has quit [Ping timeout: 260 seconds]
<benny>
maybe -I just adds to the path, is there a way to force it to only look in my modified $NIXPKGS dir?
andreabedini has joined #nixos
<benny>
solution: NIX_PATH=nixpkgs=$NIXPKGS
Mitaka89 has joined #nixos
Mitaka89 has quit [Max SendQ exceeded]
syd has joined #nixos
<syd>
Hi! Is this the right place to ask beginner Nix (the package manager) and nixpkgs questions?
init_6 has joined #nixos
<benny>
bendlas: I also don't understand the purpose of the top-level cask definition. it's already available and includes the script
<benny>
syd: sure, just be ready to wait a bit as it's not always busy
<syd>
I'm trying to build a haskell package (of my own making)
<syd>
and the haskell package uses TH to grab binary files from ../assets
<syd>
which is fine when I build with stack
<syd>
but obviously nix argues that those don't exist when it is building my package
<syd>
Is there any way to tell nix that the "../assets" dir is also part of the source?
<syd>
Is there any way to tell nix that the "../assets" dir is also part of the 'sources'?
tmaekawa has joined #nixos
<benny>
I'm new myself but I think if you want to go above the package's level you need to make ../assets its own package and use that package as a buildInput
<syd>
yeah I'm okay with that, let's see if I could do that
<syd>
wait, wouldn't that package just be somewhere in the store then? instead of at "../"
<benny>
syd: yeah, but then you would reference it in buildInput with its unique name and sha and it would be present in your build env
<benny>
that's the way to do pure packaging as I understand it
<benny>
but maybe this is not optimal and just having it live in the package directory instead of above it is the way to go, you could try looking through nixpkgs source for people that packaged haskell stuff to get an answer for what is the right way to do this
<syd>
Yes, I may just have to move the assets
<syd>
which is too bad, but oh well
inquisitiv3 has quit [Ping timeout: 256 seconds]
<benny>
I'm having my own problems trying to get something that works in a side effects heavy package manager and converting it into pure nix due to the complexity involved
<syd>
well, let me know if I can help
<benny>
I've abandoned that to just use docker, it was taking up too much for too little benefit :P
<Lisanna>
how do I enable secret-key-files in hydra?
<Lisanna>
ah, binary_cache_secret_key_file
<benny>
how do I get stuff included in 18.03? it's already in master. do I contact whoever accepted the PR?
knupfer has joined #nixos
<Lisanna>
wait, I set that and reloaded hydra, but stuff in its binary cache is still not signed. Do I need to purge everything and rebuild it?
<{^_^}>
[nixpkgs] @LnL7 opened pull request #42415 → curl: disable default CA bundle → https://git.io/f4yjJ
<MichaelRaskin>
I think nix-store can sign paths in store…
<MichaelRaskin>
So even if you do purge/rebuild there shouldn't be a need to actually do the builds
<MichaelRaskin>
(But I don't know details about Hydra)
<Lisanna>
MichaelRaskin well, it uses this method for specifying the store dir, although I don't know if the signing applies to things going in or things coming out: "store-uri = file:///nix/store?secret-key=/run/keys/hydra.example.com-1"
<Lisanna>
must be on things going in, since just setting that isn't enough for the client machine with require-sigs to accept anything from it
justanotheruser has quit [Ping timeout: 264 seconds]
<MichaelRaskin>
You could try to ask nix-store to sign everything already there with this key
<Lisanna>
MichaelRaskin okay, not seeing the flag for that though
<Lisanna>
nix-store can generate the keypair, but I don't see anything about retroactively signing anything
<MichaelRaskin>
nix sign-paths
<Lisanna>
oh cool!
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Lisanna>
okay, the operation in theory completed, no sign from the client that anything has changed, still complaining about lacking a valid signature.
<MichaelRaskin>
You specified the key in the sign-paths command, right?
<Lisanna>
yeah
<Lisanna>
hmm, interesting, nix copy seems to be happy with the state of affairs
<Lisanna>
I can copy things directly over ssh without having to specify --no-check-sigs
<Lisanna>
so wtf is hydra doing
knupfer has quit [Ping timeout: 265 seconds]
rardiol1 has quit [Ping timeout: 256 seconds]
<MichaelRaskin>
Maybe you should clear the cache itself now and let it re-push?
<MichaelRaskin>
Sounds like you have unsigned narinfos there that it doesn't recreate
<MichaelRaskin>
Maybe clearing out only narinfos would be enough…
<Lisanna>
ah, that's a good idea
<Lisanna>
not sure how to do that though
phdoerfler has quit [Quit: Leaving.]
<MichaelRaskin>
Where your cache lives?
<Lisanna>
also, wait, I thought narinfo files were only present in binary caches? I specified the hydra store as /nix/store (following an example)
<Lisanna>
so, shouldn't it just be using the live /nix/store on the machine?
<MichaelRaskin>
I mean, Hydra web server process should be under a single UID
<MichaelRaskin>
I would really hope it is not a build user that is used to access the key
philippD has joined #nixos
<{^_^}>
[nixpkgs] @clefru closed pull request #42362 → Force positive stackprotector detection on x64 for Linux >=4.16 → https://git.io/f4Dnh
<Lisanna>
these are the hydra users that I see from inspecting its nixos module: "hydra" "hydra-queue-runner" "hydra-www"
<Lisanna>
and they all have access to the key
<MichaelRaskin>
Hmm
<Lisanna>
Caught exception in Hydra::View::NARInfo->process "cannot open file '/run/keys/hydra.example.com-1': Permission denied at /nix/store/scmd0s9zjn59nldmch7pl477absmg46n-hydra-perl-deps/lib/perl5/site_perl/5.24.3/x86_64-linux-thread-multi/Nix/Utils.pm line 37."
<Lisanna>
That's the message being sent with the 500
<MichaelRaskin>
I would just put inotify on the file and see who wants it
<{^_^}>
[nixpkgs] @bandresen opened pull request #42419 → Emacs cask version update and wrapper → https://git.io/f4SeR
<flyx>
I just saw that nixos-rebuild installed gtk3 along with x11 on my server. I do not want x11 on my server. how can I find out why gtk3 was installed?
b has joined #nixos
<Lisanna>
inotifywait is showing me that it's accessed but not by whom :/
<{^_^}>
[nixpkgs] @Ma27 opened pull request #42420 → nixos/npm: add `npm` to the system closure by default → https://git.io/f4Sea
<Lisanna>
flyx nix why-depends
<MichaelRaskin>
process id maybe?
<MichaelRaskin>
Wait
<MichaelRaskin>
You said you gave access
<MichaelRaskin>
How? Adding to a group?
<MichaelRaskin>
Did you restart Hydra to make sure new groups are applied to the process?
<Lisanna>
MichaelRaskin made a dummy file
<Lisanna>
and changed the location
<Lisanna>
"/tmp/testkey OPEN" is the only message I get
<Lisanna>
lol
inquisitiv3 has joined #nixos
<MichaelRaskin>
Oh well
<MichaelRaskin>
So, what about the groups?
<{^_^}>
[nixpkgs] @bandresen opened pull request #42421 → pythonPackages.python-periphery: init at 1.1.1 → https://git.io/f4SeX
<MichaelRaskin>
Or just attach strace to Hydra already
<{^_^}>
[nixpkgs] @basvandijk opened pull request #42424 → pgmanage: remove deprecated postage alias → https://git.io/f4Svt
<Myrl-saki>
Sweet jesus, all Inux distros either look the same, or look crappy.
<Myrl-saki>
Linux*
<Myrl-saki>
Err, not distro. Destop Environment.
<init_6>
so make it asome
<Myrl-saki>
I think XFCE's the only one that deviates from the norm which I actually like.
<Myrl-saki>
init_6: Did you mean to say "make some"
<init_6>
i mean that i say
<Lisanna>
flyx curious, why do you care so much if x11 is on your server or not? trying to save on disk space?
<flyx>
Lisanna: I want to figure out whether I did something wrong in my configuration
<Myrl-saki>
flyx: Can you link us your configuration?
<Myrl-saki>
I can nix-uild it.
<Myrl-saki>
nix-build*
<flyx>
Lisanna: nixos-rebuild builds /nix/store/0j2jqd2mdyczx6vgvaqb64zzwmv8bp0g-nixos-system-nixos-18.03.132750.91b286c8935.drv, but I do not know how to transform that to a package name usable with nix why-depends
<{^_^}>
[nixpkgs] @rycee pushed commit from @pasqui23 to master « home-manager:2017-12-7 -> 2018-06-14 »: https://git.io/f4Sfe
iyzsong has quit [Ping timeout: 264 seconds]
<MichaelRaskin>
You are in a position of a software engineer who does admin stuff because spending effort on Hydra is still less effort than fighting the alternative options of CI?
iyzsong-x has joined #nixos
<MichaelRaskin>
(otherwise how the hell would you manage to avoid knowing all the group gotchas, and hearing about ACLs)
<Lisanna>
yeah, sysadmin is not my profession
<MichaelRaskin>
Not even automating QA (which might be distinct on sysadmin)
<MichaelRaskin>
It's more that you obviously grabbed quite a few quite powerful servers, which means you probably do use the company servers.
tzemanovic has quit [Ping timeout: 256 seconds]
<MichaelRaskin>
I imagine how annoyed you should have been when requesting the servers to redo CI from scratch, but correctly
<Lisanna>
yeah, I'm not happy with the current state of affairs
<Lisanna>
but it's also not my primary job, it's something I work on in the background
tzemanovic has joined #nixos
<MichaelRaskin>
That I understand
<MichaelRaskin>
Well, at my former job I participated in a micro-private-cloud setup for an adjacent project (I had dedicated servers for mine, and I was nomically software developer) just because it was painful to watch the alternative options happen
<Lisanna>
haha
<Lisanna>
yeah, it's painful watching my coworkers when they work on the "official" build/test system
<MichaelRaskin>
Ended up with Ganeti, because a proper OpenStack deployment would ideally use as many coordination servers as we wanted to have _total_ servers
<lejonet>
My brain is not cooperating with me today, how do I use a .nix file as the basis for a nix-shell? (I'm trying to create a FHS shell with buildFHSUserEnv, but running nix-shell fhs-env.nix gives me "cannot convert a function to JSON")
<Lisanna>
lejonet does fhs-env.nix return the raw result from buildFHSUserEnv?
<lejonet>
it kindof feels like the buildFHSUserEnv somehow has changed in what attributes it wants or something
<lejonet>
I guess it would also work to somehow edit the rpaths on the binary I'm trying to run, is there any cli tool to help with that outside of creating a full package and all?
<MichaelRaskin>
patchelf
sorixelle has quit [Ping timeout: 265 seconds]
<MichaelRaskin>
Although creating a package to run patchelf might be easier than manually using it more than once…
sorixelle has joined #nixos
<lejonet>
I guess I'll have to find all the paths in the nix-store myself to use patchelf tho? :P
logzet has joined #nixos
<MichaelRaskin>
Or ask Nix to do that for you while expanding variable substitutions in runCommand!
<lejonet>
And how would I do that dark magic? :P
Huddo121 has quit [Remote host closed the connection]
<gchristensen>
MichaelRaskin: 'ey, look at my PM :P
grw has joined #nixos
<Lisanna>
hrm, even with builders-use-substitutes and the binary caching working it's still insisting on downloading quite a bit of stuff to my local machine before it starts attempting to build remotely.
<Lisanna>
also autoPatchelfHook is your friend
<lejonet>
MichaelRaskin: I'm guessing runCommand isn't a cli tool, but something to run inside a nix-shell expression or so?
<MichaelRaskin>
runCommand is a function available at nixpkgs top-level pkgs set
Anthony_Bourdain has joined #nixos
tex has joined #nixos
Anthony_Bourdain has quit [Remote host closed the connection]
<lejonet>
meh, ofc libGl isn't an attribute of nixpkgs (trying runCommand in a nix repl '<nixpkgs>') but nix search libGl gives that nixpkgs.libGl is supposed to exist :(
<gchristensen>
libGL not libGl
<tex>
Hi! How to write to file in my default.nix used to setup C++ dev enviroment? I need to get paths to some buildInputs... ``` with import <nixpkgs> {}; { cppEnv = stdenv.mkDerivation { buildInputs = [ stdenv gcc6 ... }; } ```
<lejonet>
gchristensen: that is what I guessed, but it says libGl in the nix search output :(
<tex>
The file needs to be in the same folder as the default.nix...
<lejonet>
gchristensen: might be my font and coloring that makes it odd for me
<{^_^}>
[nixpkgs] @clefru opened pull request #42430 → Disable hardening the stackprotector for spl/zfs. → https://git.io/f4SJc
<gchristensen>
ah
agander has joined #nixos
<lejonet>
When running runCommand from a nix repl, where does $out gets created? $(pwd)/result?
<__monty__>
clever: Does toxvpn require root?
<MichaelRaskin>
lejonet: nix-repl evaluates, but doesn't build
Anthony_Bourdain has joined #nixos
<MichaelRaskin>
lejonet: replace stdenv with stdenv.all
<MichaelRaskin>
Then everything will work
<lejonet>
MichaelRaskin: yeah, I figured that when it outputs the drv path, but was wondering how that did the evaluation of relative paths and so, I'm guessing it'll take whatever pwd that you're in if you do a nix-build on that drv?
<MichaelRaskin>
Yeah, or nix-store -r and it will just output $out without symlinks or GC roots
<MichaelRaskin>
lejonet: your fhs.nix is fine, if you put stdenv.all instead of stdenv
<gchristensen>
$out is always in the nix store, ./result is a result symlink to the nix store
<lejonet>
MichaelRaskin: yeah, just tried that, its now claiming I don't have unprivileged_userns_clone = 1 even tho I do :P
<lejonet>
MichaelRaskin: thx for the help :)
<MichaelRaskin>
Are you inside a container?
<lejonet>
gchristensen: ah true
<lejonet>
MichaelRaskin: nope, shouldn't be
<MichaelRaskin>
Can you use unshare?
<MichaelRaskin>
(also, if you are inside unshare/nsjail, exit it)
brandon_ has quit [Remote host closed the connection]
<lejonet>
O.o it would seem like kernel.unprivileged_userns_clone is 0, even tho I have boot.kernel.sysctl."unprivileged_userns_clone" = 1; in my config
<MichaelRaskin>
Well, something else can set sysctl…
<MichaelRaskin>
Also, is it Nixpkgs kernel?
<lejonet>
set it myself so I can use this lol :P
<lejonet>
I'm running nixos yes, using copperhead_stable
Ariakenom has quit [Remote host closed the connection]
<MichaelRaskin>
ij: This job is not a member of the latest evaluation of its jobset. This means it was removed or had an evaluation error.
<lejonet>
Because the sandbox works atleast, after setting max_user_namespaces to a high enough number :P
Ariakenom has joined #nixos
<MichaelRaskin>
Hydra says this if you click on a job
<tex>
Will set enviroment variable instead of file...
<ij>
gotcha
<gonz_>
Is there any way to list which keymaps are available for `i18n.consoleKeyMap` and `services.xserver.layout` respectively?
<gonz_>
I seem to remember Svorak A5 being available in one of those
<gonz_>
That would be something like sv-dvorak-a5
<lejonet>
MichaelRaskin: sure, I understand if they want to disable it as a default, but shouldn't my own settings overrule that, because well, I'm the user and my settings should be honoured firstly? :P
<MichaelRaskin>
Maybe your settings are just in the place where they don't get applied at all
<lejonet>
MichaelRaskin: which would be mean, as they are in my "top-level" configuration.nix, i.e. the nix file that imports all the other files
tex has quit [Ping timeout: 260 seconds]
<MichaelRaskin>
Are the settings in question in /etc/sysctl.d/nixos.conf ?
<{^_^}>
[nix] @dtzWill opened pull request #2250 → repl: use `nix build` for building instead of `nix-store -r` → https://git.io/f4SJX
<MichaelRaskin>
What is the status of systemd-sysctl
<lejonet>
MichaelRaskin: yep, unprivileged_userns_clone=1 is in that file
<lejonet>
There is the error: "Jun 09 02:03:17 achilles systemd-sysctl[12231]: Couldn't write '1' to 'unprivileged_userns_clone', ignoring: No such file or directory" meh :P
<lejonet>
I guess I need to prefix it with kernel. then
<lejonet>
In my naivety, I thought boot.kernel.sysctl would do that :)
<MichaelRaskin>
Well, not all sysctl settings are kernel.* I think
<lejonet>
quite a few aren't, dunno why I thought that honestly
<lejonet>
was quite a while since I conf'd this desktop, and haven't really used nix-shell or anything needing unpriv userns clone in quite a while, hence why I didn't catch it until now
yellowbrick has quit [Quit: Page closed]
<lejonet>
and now to nuke saleae for crappy applications "I cannot find the platform module xcb! The only available platform plugins are: xcb" -.-
<rardiol1>
Someone here runs btrfs on /? I tried updating and now the system won't boot, errors during stage1 mounting / saying something like "failed to open_ctree", "failed to read ctree -5" "failed to read system array". I can use the btrfs fs fine from another linux or grub, so the fs itself is fine
Jason_Grossman has quit [Remote host closed the connection]
Jason_Grossman has joined #nixos
Jason_Gr` has joined #nixos
Jason_Gr` has quit [Remote host closed the connection]
Jason_Grossman has quit [Remote host closed the connection]
<rardiol1>
viric: hi. the irc logs say you had a similar problem. Do you remember how you fixed it by any luck?
agander_ has quit [Read error: Connection reset by peer]
<MichaelRaskin>
rardiol1: can you mount this exact partition from a LiveUSB?
michaelpj_ has joined #nixos
<rardiol1>
MichaelRaskin: I have it mounted right now from mint, you mean specifically a nixos liveusb?
<rardiol1>
I can try, just need to download and reboot, you think nixos is unable to mount the fs for some reason?
<MichaelRaskin>
Hm. Specifically a NixOS liveusb might show if there is a kernel version difference, but if Mint mounts it, then it's probably not corrupted
<MichaelRaskin>
I guess checking dmesg on Mint to see that mounting went without compliants might be a good idea
<lejonet>
rardiol1: I have a similar error on one of my servers, does it say "Invalid argument" in the console when it errors out?
<lejonet>
I haven't been able to fix it, but setting boot.shell_on_fail and running btrfs device scan and then mounting it just as the initscript tries to, works
<lejonet>
sometimes you have to do the btrfs device scan twice, so it seems like it might be related to some type of initial probing of the layout that goes awry
<rardiol1>
MichaelRaskin: I moderately suspect a kernel version problem. Sort of long history: I tried updating the nixos partiion a few months ago, but I had a similar problem, that was generation 58 on the boot menu. For some reason I stopped using nix for a few months, but I could still boot by using the previous generation (57) on the boot menu. A couple days ago I boot on gen 57, try running "sudo nixos-switch --upgrade" again,
<MichaelRaskin>
You don't use any subvolumes, by the way?
<MichaelRaskin>
For kernel version problem checking a NixOS liveusb might work as diagnostic…
<rardiol1>
at some point, the screen goes black, I do a REISUB, and for some reason generation 57 disappeared ( I Suspect I ran a delete I shouldn't have, but I thought it should still be there), and the gen 59 with a more recent kernel fails very similarly to 58, just a different message, I think one says Failed to read system array and the other failed to read ctree
<rardiol1>
MichaelRaskin: I do actually
init_6 has quit [Ping timeout: 264 seconds]
<rardiol1>
lejonet: Yes, that's what it says after the error
<MichaelRaskin>
Wonder if there is a change in subvolume syntax for mounts…
<MichaelRaskin>
(I never used subvolumes with BtrFS, though)
<michaelpj_>
some packages provide systemd user units in `$out/lib/systemd`. However, these end up in `/run/current-system/sw/lib/systemd`, which doesn't seem to get picked up anywhere. I think these *should* get picked up - for example evince has a daemon which is started by such a unit, but it can't start it because the unit file isn't anywhere systemd can find
<michaelpj_>
anyone know if this is supposed to work?
<MichaelRaskin>
I think not
<lejonet>
MichaelRaskin: possibly, because when I run the command manually, I let it use the default subvolume (which is the rootfs subvolume) instead of doing subvol= or subvolid= as an option
<hyper_ch>
hmmm, I tried to setup openvpn's sftp chroot but it's not working... where can I find the sshd_config ?
<lejonet>
maybe I should try to specify subvol= and see if that errors out
<MichaelRaskin>
And does your configuration specify subvol?
<lejonet>
MichaelRaskin: iirc it does
<hyper_ch>
in /run/current-system/etc/ssh there is no sshd_config
<infinisil>
hyper_ch: /etc/ssh/sshd_config
<rardiol1>
here's the options line from hardware-configuration.nix: options = [ "defaults" "autodefrag" "compress=lzo" "subvol=/nixos1/" ];
<hyper_ch>
infinisil: does not exist here
<lejonet>
Yep, it does, let me try that a quicky, maybe "just" removing that would make it not error out
<infinisil>
hyper_ch: Check the command line of the sshd process, it might be there then
<infinisil>
hyper_ch: Not sure what this has to do with sftp though, I just know my sshd_config is in /etc/ssh
<rardiol1>
lejonet: thanks for the hint, I vaguely remember this being related to some kernel parameters like "dobtrfs rootfstype=btrfs rootfsparams". But how do I set boot.sheel_on_error when I don't have a running system? Run from usb and do a chroot?
<lejonet>
MichaelRaskin: :/ specifying subvol=rootfs (as is in the conf) in the shell_on_fail shell works like a charm :/
<MichaelRaskin>
Hmmm
<lejonet>
rardiol1: just appending boot.shell_on_fail to the grub entry works (press e when it presents the grub menu)
Aleksejs is now known as rvsiers
<rardiol1>
lejonet: as a kernel parameter? just "boot.shell_on_fail" or "boot.shell_on_fail=true" or something?
<lejonet>
rardiol1: as a kernel parameter, boot.shell_on_fail is what you need
<lejonet>
=true will make it ignore it :( (I tried that the first times and wondered why I wasn't getting the shell option)
<lejonet>
MichaelRaskin: interesting to note for my case is that the btrfs errors from dmesg is from the other disk in the raid1, not the one it tries to mount, so maybe there is some type of discrepancy between the 2 disks in the raid1 or such
<rardiol1>
ok, thanks, will try that , bye
jensens has quit [Ping timeout: 260 seconds]
agander_ has joined #nixos
rvsiers is now known as Aleksiers
agander__ has joined #nixos
rardiol1 has quit [Ping timeout: 260 seconds]
worldofpeace has joined #nixos
agander__ has quit [Client Quit]
agander__ has joined #nixos
agander__ is now known as agander
jD91mZM2 has joined #nixos
agander_ has quit [Ping timeout: 264 seconds]
<worldofpeace>
Anyone have a way to use plasma5 with a wayland session?
TonyTheLion has quit [Remote host closed the connection]
fader has joined #nixos
MercurialAlchemi has quit [Ping timeout: 260 seconds]
andymandias has quit [Ping timeout: 256 seconds]
tzemanovic has quit [Remote host closed the connection]
andymandias has joined #nixos
tester123 has quit [Ping timeout: 260 seconds]
lanelet has joined #nixos
michaelpj_ has quit [Remote host closed the connection]
rardiol1 has joined #nixos
<rardiol1>
lejonet: thanks, it worked on the third try for some reason. Is there some trick on what to do after mounting? just ctrl-d? And I'm supposed to pick the normal interactive shell or PID 1?
agander has quit [Ping timeout: 265 seconds]
<lejonet>
rardiol1: I pick the normal one and after I've mounted, just ctrl-d and it'll continue where it stopped
<rardiol1>
weird, I thought I did that once and it just kernel panic. Maybe the fact I had scroll lock on the second time to try to see what went wrong changed something?
<lejonet>
no idea
<rardiol1>
and in any case, I bet our problem is that stage 1 is not doing "btrfs device scan", which I remember is fixed by using "dobtrfs" as a kernel parameter on gentoo, maybe nixos hasn't done this fix on its stage 1?
<lejonet>
Well for me it says "Scanning for Btrfs devices" just before mounting, so it seems to atleast attempt to do it
fendor_ has joined #nixos
<lejonet>
I'm wondering what is the actual difference between the env you get in the interactive shell and the env that the initscript has
igo95862 has joined #nixos
<lejonet>
Because somewhere there the answer probably lies
tzemanovic has joined #nixos
fendor has quit [Ping timeout: 264 seconds]
<rardiol1>
nixos/modules/tasks/filesystems/btrfs.nix does seems to exec "btrfs device scan"
tzemanovic has quit [Ping timeout: 256 seconds]
<lejonet>
Maybe its some type of race condition? That the initscript tries to mount the btrfs fs too fast?
steell_ has joined #nixos
tycho01 has joined #nixos
<rardiol1>
hmm, maybe, I remember someone mentioning this among a multitude of archlinux posts about similar problems
blankhart has joined #nixos
<lejonet>
Are you using simple btrfs or any of the raid modes too? Because it might be something in the raid modes that does it otherwise (I'm running RAID1)
<rardiol1>
simple btrfs, just with multiple devices
<lejonet>
hmm, so it could still be some type of sync between the devices, and the btrfs device scan returns too fast or something like that (i'd not be surprised of btrfs device scan is done async)
tzemanovic has joined #nixos
<{^_^}>
[nixpkgs] @mredaelli opened pull request #42436 → epkowa: added Perfection V500 support → https://git.io/f4SLd
<rardiol1>
Adding a "sleep 5" just after the scan would fix it then?
<lejonet>
in theory, yeah, if it is some type of race condition, and the only difference between the initscript mount and the interactive shell mount is the timing after btrfs device scan
<Myrl-saki>
yyadavalli: Nix packages (often) don't have Systemd units, as what samueldr has said, but you can use a NixOS configuration, compile it, then check the file tree.
<yyadavalli>
samueldr: I have one in ~/.nix-profile/lib/systemd/user for syncthing
<samueldr>
though (as I can see in syncthing) the upstream-packaged services look like they are installed
<samueldr>
you probably can do something like systemctl --user link $HOME/.nix-profile/lib/systemd/user/[service_name]
<WinchellsM>
2. But I need Version >=1.7.
<WinchellsM>
3. The current version on nixpkgs search is "1.6.5".
<WinchellsM>
4. I think normally on non-NixOS you could just run "stack upgrade" in your home folder, but that doesn't seem to work.
<WinchellsM>
What do?
<nh2[m]>
what's the simplest way to nixify a python script? I have one that depends on the `boto` python packae, and I want to run it from a cronjob in nixos. How should I wrap it so that its dependencies are nicely declared?
<Myrl-saki>
WinchellsM: Do you want to update Stack the nix way?
<yyadavalli>
samueldr: Looks like it is setup to work correctly, ExecStart is pointed correctly
<yyadavalli>
let me try linking it
<WinchellsM>
Myrl-saki: How would I do that?
<nh2[m]>
WinchellsM: stack 1.7 builds reasonably well in current nixpkgs, perhaps with some patches. See https://github.com/nh2/static-haskell-nix/tree/master/static-stack for how I made stack 1.7 work with some recent version of nixpkgs even though master had 1.6.5
<Myrl-saki>
WinchellsM: I'll go check the NixOS manual.
<samueldr>
yyadavalli: within nixos (may not be an issue for nix on other distros) sometimes execstart isn't enough
<samueldr>
e.g. missing entries in PATH
<samueldr>
as systemd (and it's a good thing) clears the environment
<yyadavalli>
ohk
<Myrl-saki>
WinchellsM: Seems like you do it shi way.
<Myrl-saki>
WinchellsM: this*
<nh2[m]>
Myrl-saki: so far I'm not using any function yet; happy to get recommendations. Looking at buildPythonPackage, it's unclear to me how I should tell it that the code to execute is my single python file
<nh2[m]>
or how it would it figure out which the main file is (as most usages seemt to take a directory as `src`)
<yyadavalli>
samueldr: systemctl failed to link; Too many levels of symbolic links; I am assuming its a bad idea to point to the store directly
<samueldr>
ooh, I didn't know it wouldn't like using symbolic links :/
<samueldr>
the issue with the store path, yyadavalli, would be that you'd miss on updates
<samueldr>
(I only learned recently about `systemctl link` and haven't yet learned about its limitations)
cfricke has quit [Quit: WeeChat 2.1]
<WinchellsM>
Myrl-saki: I can paste that into my configuration.nix?
<Myrl-saki>
nh2[m]: `buildPythonPackage` is used for pip packages. I think. My work consists of Nixing Python programs, but I'm not familiar with Python packaging.
<yyadavalli>
Can we configure where systemd looks for services when using `systemctl enable`?
<Myrl-saki>
WinchellsM: Yeah. How much do you understand of what I'm doing?
tycho01 has quit [Quit: Page closed]
<Myrl-saki>
WinchellsM: I haven't tried that out yet, give me 5 or so mins.
<WinchellsM>
Ok
<WinchellsM>
Thanks
warbo has joined #nixos
<WinchellsM>
I understand it 60%
<Myrl-saki>
WinchellsM: Do you know how to use nix-build?
<WinchellsM>
I prefer to use my configuration.nix as much as possible since I have duplicate installs; keeping things in my configuration.nix allows me to just pass that file back and forth among machines to keep everything consistent.
<FRidh>
nh2[m]: if it is just a single file, but with Python dependencies, then I recommend using `runCommand`. Add `python.withPackages(ps: with ps; [ boto ])` to the `buildInputs` and `runCommand` will patch the shebang of your script to point to this env, assuming your script has a shebang.
<warbo>
hi I'm trying to go from nixos/nixpkgs 17.09 to 18.03, but I seem to be hitting a problem with my config.packageOverrides
<Myrl-saki>
WinchellsM: I think you're mixing `nix-build` and `nix-env` with my statement. :P `nix-build` will just add to /nix/store, and not affect global state. It's very nice for prototyping Nix expressions.
<warbo>
the error I'm getting is error: attempt to call something which is not a function but a set, at /nix/store/22h8f98hya17zh12prqnxkjlcf8bx0xs-nixos-18.03.132748.68e02f8ff21/nixos/pkgs/top-level/stage.nix:106:5
<Myrl-saki>
WinchellsM: If you're feeling a little brave, then you can run the global garbage collector. Otherwise, you can use nix-store --delete, I think.
<FRidh>
nh2[m]: runCommand is simpler, yet powerful enough for what you need. In my opinion it makes the code more readable.
<warbo>
thing is, this problem only occurs for a project which uses a pinned version of nixpkgs (16.03)
<Myrl-saki>
nh2[m]: Let me check my Github for a second. :P
<jD91mZM2>
WinchellsM: That won't install anything, just build it and make a `results` directory in your current folder
<WinchellsM>
As in run `sudo nix-store --delete` without any other arguments?
<WinchellsM>
Ok
<warbo>
if I tell that project to use 18.03 then it doesn't have that problem (but hits a bunch of others since it's not designed to run with 18.03)
<Myrl-saki>
WinchellsM: I'm not so sure right now.
<Myrl-saki>
WinchellsM: I never felt the need to delete packages.
<warbo>
Myrl-saki: whenever I ask something in IRC, but find the solution elsewhere, I always come back and mention the answer
<warbo>
Myrl-saki: since I'll inevitably end up Googling the same thing later, and find my own question in the results :P
<jD91mZM2>
WinchellsM: Yeah you won't need to run nix-store if you really wanna delete the uninstalled package. Otherwise the next GC will take care of it anyway.
<Myrl-saki>
warbo: We need more people like you. :P
rardiol1 has joined #nixos
* warbo
notes that nix-store --delete can take a LONG time if it's not been done for a while/ever
<Myrl-saki>
Uh, crap.
<Myrl-saki>
How do you unset a value? The `stack` in hackage-packages sets a `revision`.
<Myrl-saki>
So it's not a simple overrideCabal.
<Myrl-saki>
Tried `revision = null;` and `revision = "";` with neither working.
<Myrl-saki>
WinchellsM: This might be a bit more complicated. :P
<Myrl-saki>
warbo: Yeah, but it's already been passed to mkDerivation.
roconnor has joined #nixos
<warbo>
ah, maybe overrideAttrs?
<rardiol1>
lejonet: tried adding a sleep 10 after the scan, no luck :( Weird thing is that I tried running with boot.trace, and by the order of the things on dmesg, it looks like the boot system actually ran the HD initialization, with the messages about "SCSI disk attached, "sdb3 detected", "write protection is off" and so on and on AFTER the scan. so maybe the problem is that the scan is running BEFORE disk init somehow?
kp__ has joined #nixos
<Myrl-saki>
warbo: That's what I'm doing ish. I'm using overrideCabal, but there doesn't seem to be a way to unset it.
<rardiol11>
lejonet: fixed it, the sleep needs to be BEFORE the scan, so it has time to init the HD I suppose
<ndowens04>
hey Myrl-saki
<Myrl-saki>
ndowens04: Is it about not being able to mosh to a NixOS system?
rardiol1 has quit [Ping timeout: 260 seconds]
<lejonet>
rardiol11: interesting
<Myrl-saki>
ndowens04: Something related to waiting for port 60000 or the like?
<lejonet>
rardiol11: you running it on mechanical disks or SSDs? (I have 2 SSDs)
<warbo>
hmm, looks like nixpkgs 16.09 fails but nixpkgs 17.03 works
<rardiol11>
lejonet: mechanical
<lejonet>
rardiol11: even more interesting then that it tries to poke em "too fast"
<warbo>
did something substantial change in the way overrides are done between those 2 versions?
<ndowens04>
Myrl-saki: it doesnt give an error on mosh-server side, seems to be running, just I can't connect to it on my phone. FUnny thing was before I had to reinstall to adjust partition size I had it working but also could be becase I upgraded to unstable
<warbo>
is that when "overlays" were added? (still haven't looked into what those are :P )
<Myrl-saki>
WinchellsM: Welps, stack fails to build for me.
<WinchellsM>
Myrl-saki: I'm here
<Myrl-saki>
ndowens04: Specifically just from your phone?
<WinchellsM>
Myrl-saki: Sorry about that.
<Myrl-saki>
src/Stack/Prelude.hs:15:5: error:
<rardiol11>
lejonet: dmesg has the sleep start at "1.439305", and the last message relating to HW init at "2.927167", so 10 seconds is probably more than needed
<Myrl-saki>
Ambiguous occurrence ‘stripCR’
<Myrl-saki>
It could refer to either ‘X.stripCR’,
* warbo
sunk a LOT of time into trying to make stack work, on a few separate occasions, before giving up entirely
<Myrl-saki>
WinchellsM:
<ndowens04>
yup it says no response from mosh server but mosh server looks to be running according to systemctl and looking in ps -A
<Myrl-saki>
ndowens04: If it's no response from Mosh server, then have you checked your firewall settings?
<lejonet>
rardiol11: so btrfs device scan is most likely run async then, which is why when it returns, it actually isn't finished yet in actuality
* warbo
subsequently gets quite annoyed to see stack referred to as "just works", "less complicated than cabal", etc. to newbies :P
<{^_^}>
[nixpkgs] @FRidh merged pull request #42410 → nixos/screen: add `pkgs.screen` to the system closure → https://git.io/f4yH0
<Myrl-saki>
warbo: TBF, it just worked for me, just that Nix works even better for me. :P
<ndowens04>
Myrl-saki: i have firewall disabled completely on nixos, and i know it isn't the router as I have port forwarding and it was working perfectly on Arch
<rardiol11>
lejonet: I think not, because if that was the case, the sleep AFTER the scan would fix it. The actual problem is that the scan is running before the HD is initalized, so the scan starts, sees no HDs to scan and says "ok, nothing to scan, scan done!"
<Myrl-saki>
ndowens04: Hmm
<Myrl-saki>
ndowens04: Sorry, can't help you. That's the only problem I encountered with mosh.
<lejonet>
rardiol11: hmm, yeah, that seems more likely
<warbo>
aha, according to the release notes nixpkgs 17.03 (a) introduced "overlays" and (b) rewrote overridePackages to be based on overlays
<ndowens04>
yea it'sbeing weird to me, i have unfreeAllow enabled in configuration.nix and i tseems to ignore that as well
<manveru>
doesn't look like mosh actually has a service
<Myrl-saki>
It bootstraps itself using SSH(port 22), but it runs mosh-server on port 60000..., (user changeable) IIRC.
<Myrl-saki>
manveru: Yeah, but neither should it be. It should be an application.
<ndowens04>
i created a systemd service file to make it start on auto which i did before reinstall and worked fine
<Myrl-saki>
manveru: `mosh-server` is 1:1 with mosh-client.
<Myrl-saki>
I..think?
<Myrl-saki>
At least that's how I understand it.
<manveru>
so mosh connects over ssh and then builds its own connection from that?
<ndowens04>
hmm port 60000 is stealthed so that's closed
<WinchellsM>
Myrl-saki: I got an empty message from you above?
<Myrl-saki>
From what I understand, yeah. `mosh-client` SSHs to the server, starts a mosh-server, along with a unique token/key, and communicate with that.
Drakonis has joined #nixos
<manveru>
looks like it
<Myrl-saki>
WinchellsM: Yeah. Backlog a bit and you'll see the error I encountered. It's non-trivial to fix.
<Myrl-saki>
WinchellsM: By that, I mean it requires cooperation from Stack devs.
<Myrl-saki>
WinchellsM: Or a Nix user who's patient enough.
<Myrl-saki>
WinchellsM: The fix is basically to make it either a qualified import or to specifically set `Prelude.` or `X.`. I think.
<rschm>
it is Compiling Network.Wai.Application.Static ( Network/Wai/Application/Static.hs, .stack-work/dist/x86_64-linux/Cabal-2.0.1.0/build/Network/Wai/Application/Static.o
<Myrl-saki>
rschm: Very first question is "how to reproduce"
illabout has quit [Ping timeout: 260 seconds]
<Myrl-saki>
What command did you run?
<clever>
__monty__: yeah, toxvpn needs root to create the tun interface
<Myrl-saki>
rschm: And what package failed to build, according to Nix?
<clever>
__monty__: but you can do (i think) -u foo, to make it drop root after doing so
<rschm>
it is stack build --nix --file-watch
<Myrl-saki>
rschm: You should see something like
<Myrl-saki>
builder for '/nix/store/rw1z3qiphrhgax7cpndhzg1c9yh60agq-stack-1.7.1.drv' failed with exit code 1
<Myrl-saki>
Oh.
<Myrl-saki>
I'm not familiar with stack + nix.
<rschm>
in the link it is suggested that zlib is added to the dependencies.
<Myrl-saki>
Yeah.
<rschm>
but that didn't resolve the issue on nixos.
<Myrl-saki>
Wait, I think I actually do have that.
<{^_^}>
[nixpkgs] @dezgeg pushed commit from @Pneumaticat to master « mpage: fix urls from mesa → glu transition »: https://git.io/f4SOD
tzemanovic has joined #nixos
andymandias has quit [Ping timeout: 264 seconds]
<__monty__>
clever: Installing the tuntap cask with brew got me further. But getting terminated by a different signal now: fish: 'sudo toxvpn -u toonn' terminated by signal SIGSEGV (Address boundary error)
<clever>
__monty__: try running it under gdb to get a backtrace
<rotaerk>
hmm question for those familiar with both nix and cabal: if I need to `cabal install` some things for my emacs plugins to work, should I just do that in the general context, or within a nix-shell, or should I configure it in my shell.nix somehow?
<clever>
rotaerk: you should build it with nix-build i think, look into cabal2nix
<__monty__>
clever: During startup program terminated with signal SIGTRAP, Trace/breakpoint trap.
<__monty__>
And then "No stack."
<clever>
__monty__: dang
oida has joined #nixos
<rotaerk>
cabal2nix gives the derivation corresponding to a cabal file, which can in turn be run with mkDerivation inside of haskellPackages, but that's not what I'm trying to do here
Acou_Bass has quit [Read error: Connection reset by peer]
<clever>
rotaerk: if you need to execute the code inside it, then its best to build it with cabal2nix
Acou_Bass has joined #nixos
iyzsong-x has quit [Ping timeout: 265 seconds]
rardiol1 has quit [Ping timeout: 268 seconds]
<__monty__>
clever: Hmm, seems to have solved itself, no longer getting killed. I'm gonna assume it's some mac os guardian feature. Still getting "operation not permitted setting ip" and "error setting dest ip: operation not permitted" though.
<clever>
__monty__: try without the -u flag
Acou_Bass has quit [Read error: Connection reset by peer]
<__monty__>
clever: Difference between add and whitelist?
<clever>
__monty__: whitelist lets a person connect, but does not inform them about wanting to connect over the onion network, so they will never know to accept
<clever>
__monty__: add will whitelist them, and send a friend request over the onion routing, so they get a log msg saying to whitelist you back
<{^_^}>
[nixpkgs] @matthewbauer pushed to master « boost: use default toolset option »: https://git.io/f4S34
Acou_Bass has joined #nixos
<clever>
but if your going to ssh into both ends and add eachother, it doesnt matter much which you use
myshoe has quit [Quit: Lost terminal]
myshoe has joined #nixos
<__monty__>
clever: How do I test it's working? Ping doesn't but maybe it shouldn't.
<clever>
__monty__: first check `list` to see if it says its online
joepie91_to_go has joined #nixos
<clever>
ping will still depend on the normal firewall rules, and nixos blocks it by default
Xiro` has quit [Ping timeout: 265 seconds]
<joepie91_to_go>
so since upgrading to 18.03 on my thinkpad x270, I've noticed that everything seems to have generally become more sluggish
<joepie91_to_go>
I'm not seeing this on my desktop
<joepie91_to_go>
has something significant changed in terms of power management or scheduling in 18.03?
<__monty__>
clever: There's a lastonline value. No on/offline status though.
Acou_Bass has quit [Client Quit]
<clever>
__monty__: then its not connected yet, do both ends show the other in list?
<clever>
__monty__: try stopping and restarting both ends and then checking list again
<__monty__>
clever: Getting "writing to routing socket: file exists" "add net 10.0.0.1: gateway ...: file exists" on one end, on the other list shows an empty ownip.
<rotaerk>
hmm I think what I need is a way to build several haskell packages *and* include the bin folders in their store locations into the PATH
<clever>
__monty__: you may need to remove the routes with `ip route` or whatever the darwin tool is
<clever>
the darwin side hasnt had as much testing
<warbo>
rotaerk: I'd say it depends how you're emacs is set up
rardiol1 has joined #nixos
<warbo>
if it's an unmodified emacs package, you could stick them in your user config or something
<warbo>
if the emacs package already has a bunch of modifications you could put them in there
<clever>
rotaerk: if you need to build and get it in PATH, then cabal2nix is part of the answer, you then need to either: add that to systemPackages, nix-env -i it, or wrap emacs to add those to PATH for emacs
<warbo>
(what clever said)
<rotaerk>
hmm k
<__monty__>
clever: Hmm, route delete says it's not in the table, but toxvpn still says file exists.
<warbo>
I define a package called 'all' in my ~/.nixpkgs/config.nix, which includes everything I want in my user profile
<clever>
__monty__: what about route print?
<warbo>
that way I can e.g. update everything at once by installing the latest `all` package
<rotaerk>
interesting
<__monty__>
clever: Ah, had the wrong /24 suffix.
<{^_^}>
[nixpkgs] @dtzWill opened pull request #42454 → llvm-{4,5,6}: unconditionally set triples → https://git.io/f4S39
<__monty__>
Hmm, ping still won't work.
<__monty__>
Get network is unreachable for ssh too.
<manveru>
damn sshd... keeps segfaulting when i try to fix the path
<clever>
__monty__: maybe test it on 2 linux machines first?
<__monty__>
clever: I guess : /, will have to wait till monday then.
<fresheyeball>
File "/nix/store/zcib77h4v8ljvg0x5a39yfj822zs66wq-nixops-1.6/lib/python2.7/site-packages/nixops/deployment.py", line 731, in worker
<fresheyeball>
raise Exception("unable to activate new configuration")
<fresheyeball>
Exception: unable to activate new configuration
<fresheyeball>
and my vm is borked
<clever>
fresheyeball: there is another error above that
<fresheyeball>
Timeout, server 178.128.189.79 not responding.
<fresheyeball>
that is the only other one I see
<clever>
then nixops cant connect to the vm, so it has no way to deploy
<fresheyeball>
but the closures copied successfully
<fresheyeball>
doesn't that mean it was connected?
<clever>
did it give any other errors?
<fresheyeball>
nope
<clever>
can you reboot the vm or confirm its IP?
<fresheyeball>
Well after I try a deploy the vm is borked
<fresheyeball>
and I cant ssh into it anymore
<fresheyeball>
but before I deploy I can ssh into it just fine
<fresheyeball>
at that IP
<fresheyeball>
clever: can I increase the timout? It might just be that
<fresheyeball>
?
<clever>
probably wont help, which vm provider is it?
<fresheyeball>
clever: digital ocean
<fresheyeball>
I converted the droplets to nixos via nix-infect
<clever>
can you see any errors in the console?
<fresheyeball>
I dont see any errors in the digital ocean web console
<clever>
are you able to login from that console?
thc202 has quit [Ping timeout: 240 seconds]
WinchellsM has quit [Quit: Lost terminal]
<fresheyeball>
curently? or after nixops deply?
<clever>
after th deploy
<fresheyeball>
currently* deploy*
<fresheyeball>
I will do a deploy and see
<fresheyeball>
sec
fendor_ has quit [Remote host closed the connection]
<fresheyeball>
copying closure...
<rotaerk>
hmm my global cabal version is G, and if I run nix-shell, it's version S. when I run nix-shell, and then within it I run emacs, emacs sees version G of cabal
<rotaerk>
why?
<manveru>
clever: i'm not sure how this gdb magic works... can you maybe help me? :)
<manveru>
clever: i added `"--with-default-path=/nix/var/nix/profiles/default/bin:/run/current-system/sw/bin:/run/wrappers/bin"` to the openssh configureFlags, and that makes it segfault when i connect to the server
igo95862 has quit [Quit: igo95862]
<warbo>
OK, I've not found the solution to my packageOverrides problem, but I've found a workaround
<clever>
manveru: id start by turning on coredumpctl in the nixos options, then running crashing it, and checking coredumpctl for a coredump
<warbo>
I'm now importing nixpkgs 18.03, with all of my overrides, and using that to fetch the source for the desired nixpkgs version (16.03 in the case of this project), then importing that with all of my overrides
<rotaerk>
I wonder if I need to setup a special override of emacs within my nix-shell
<rotaerk>
for the purpose of working on that one project
<fresheyeball>
rotaerk: you can checkout nix-buffer
<warbo>
rather than trying to use a "bare" <nixpkgs> to apply all of my overrides to a selection of nixpkgs versions at once
<warbo>
rotaerk: have you tried nix-sandbox for emacs?
<rotaerk>
no
<rotaerk>
never heard of nix-sandbox
<manveru>
rotaerk: i'm using nix-sandbox
<manveru>
still had to patch a bunch of functions
<manveru>
but at least it's possible
<warbo>
it traverses a directory tree looking for shell.nix or default.nix and if found, uses that to provide executables
<rotaerk>
I see
<warbo>
I use it with flycheck so that the right ghc, etc. is used
<warbo>
it can be slow on the first invocation, but is cached for subsequent uses
<manveru>
clever: what do i do with that?
<warbo>
although I still found it to be excruciatingly slow, so I made some custom wrapper functions that start the nix-shell build asynchronously and fail immediately
<clever>
manveru: then run bt in there
<manveru>
did you see my paste?
<clever>
manveru: ah, checking
<warbo>
so that all invocations, including the first, are instant; but they all fail until the background builder has finished
<clever>
manveru: ah, you have to start sshd manually after doing `ulimit -c unlimited`
<clever>
manveru: stop the systemd service, then read its .service file and run it manually
<warbo>
for flycheck that's fine, since it means my buffer never freezes, and I start getting hints once the shell/sandbox is built:)
* warbo
should really start pushing his mountain of Nix-related patches upstream
<joepie91_to_go>
contains malware, maintainer is refusing to take it seriously and either doesn't know what he's doing, or is actively complicit
<infinisil>
MichaelRaskin: The bot won't link to #<number>'s by himself
<infinisil>
If you mean that
<joepie91_to_go>
cc gchristensen
<MichaelRaskin>
Which might be a pity in context…
<infinisil>
Yeah
<infinisil>
I should implement that
tzemanovic has joined #nixos
slyfox has quit [Quit: \o/]
steell_ has joined #nixos
<manveru>
joepie91_to_go: is that also on linux?
seku has joined #nixos
<manveru>
joepie91_to_go: seems like an issue with the installer they're using on windows
<MichaelRaskin>
,pr = 0. Sorry. 1. If you complain about PR being in limbo, always provide a number («I have an unmerged PR (#23924)…»), bot will print PR title and it might bring you an impulse-merge. 2. Please look up who commits similar work and ping these people (or request-review), keeping up with all the PRs is not really possible.
<{^_^}>
0. Sorry. 1. If you complain about PR being in limbo, always provide a number («I have an unmerged PR (#23924)…»), bot will print PR title and it might bring you an impulse-merge. 2. Please look up who commits similar work and ping these people (or request-review), keeping up with all the PRs is not really possible.
<infinisil>
Thanks \o/
<joepie91_to_go>
manveru: *this particular issue* only affects the Windows installer; however, the maintainer's response makes all alarm bells go off and I absolutely don't feel comfortable stating that the underlying issue is Windows-release-only :)
kazoo_keygen has joined #nixos
tzemanovic has quit [Ping timeout: 256 seconds]
<manveru>
joepie91_to_go: well, fine with me, i'm oldschool and use ftp directly :P
<joepie91_to_go>
the maintainer is either incredibly incompetent or actively malicious; in both cases you probably don't want to trust their Linux release either
kazoo_keygen has quit [Client Quit]
<seku>
heh, seems i joined too late for an interesting discussion
<infinisil>
joepie91_to_go: Maybe worth mentioning in #nixos-security
<rotaerk>
#38316 is the one I submitted a while ago
<clever>
fresheyeball: and if you use the DO console to login as root?
<fresheyeball>
yes
<clever>
fresheyeball: and if you check ifconfig, does it have an ip?
<fresheyeball>
Iyes
<fresheyeball>
yes it appears to have a public IP
<clever>
fresheyeball: is it the right one?
<fresheyeball>
but actually I am wrong, the console is up, but I can't login
<fresheyeball>
the IP looks right
<fresheyeball>
I am wondering now if this completely overwrote the configuration.nix, and now my ssh key is not in there anymore
<clever>
fresheyeball: start over, set a password with passwd and confirm you can login as root from the console first, then deploy again
<clever>
fresheyeball: nixops just ignores the configuration.nix entirely
<clever>
you have to put your keys into the nixops deployment files
<mupf>
https://github.com/NixOS/nixpkgs/issues/42216 no one even reacted yet. If anyone has an idea how to fix this, I would really grateful. Because I could really use that package. :D
<fresheyeball>
clever: can you clue me into your thought process here?
<clever>
and ssh keys wont help over the DO console
<clever>
fresheyeball: i suspect you didnt configure the network properly, so when nixops deploys, it just turns off the network card
npmccallum has quit [Ping timeout: 276 seconds]
<fresheyeball>
clever: what do you mean by configure the network?
<infinisil>
rotaerk: The error you get there should be pretty self explanatory, I suggest to just not build the derivation that causes that error (playonlinux here). Probably you can just remove playonlinux from the command line the error displays and run it again
<clever>
fresheyeball: does DO require static IP setup or does it provide dhcp?
<rotaerk>
oh the one that prevented me from testing with nox-review?
<infinisil>
Yeah
<fresheyeball>
clever: static IP is what I am using
<fresheyeball>
clever: ok, I have a fresh ubuntu vm
<fresheyeball>
now I am running nix-infect
<rotaerk>
infinisil, how do I "just not build the derivation that causes the error"?
<clever>
fresheyeball: you need to include that static ip config in the nixops files, or it will just shut off the network
<rotaerk>
with that nox-review command
akapav has joined #nixos
<infinisil>
rotaerk: See the last line of the error
<fresheyeball>
clever: you mean other than deployment.targetHost ?
<clever>
fresheyeball: targetHost tells nixops what to ssh into, but does not configure a static ip
<rotaerk>
infinisil, right, but I'm not calling nix-build
<clever>
fresheyeball: so nixos will try to run dhcp, and DO wont answer, so it just never comes back online
<rotaerk>
not sure how I would change that command that's failing
<infinisil>
rotaerk: Yeah, but nox is, and whether you call it or nox calls it shouldn't make a difference
<clever>
fresheyeball: you must copy the configuration.nix into the nixops files
<fresheyeball>
clever: interesting, how do I tell nixops to configure a static ip?
<clever>
fresheyeball: the same way you configure every other nixos thing in nixops
<fresheyeball>
clever: I only see staticIP options in the manual for GCE
<clever>
fresheyeball: read the existing configuration.nix file, that already has the config you need
<rotaerk>
infinisil, still not clear on what you're suggesting... are you suggesting that I modify the source for nox-review, rebuild it, and then run it again?
<infinisil>
No..
<rotaerk>
(modify it such that I remove playonlinux from its nix-build call)
<fresheyeball>
clever: ok so I can see that after infect finishes
<clever>
fresheyeball: yeah
<infinisil>
rotaerk: nix-build -A winePackages.base -A libretro.stella -A .... /home/rotaerk/.nox/nixpkgs
<rotaerk>
what about it?
<infinisil>
To call that!
<fresheyeball>
clever: ok, I have another infected machine I can use as an example
<rotaerk>
okay, so you want me to manually call the build ... but will that help nox-review work? wouldn't it try to do the full build with playonlinux when it's run?
<fresheyeball>
I need to know what options I need to copy over to my nixops stuff
<infinisil>
rotaerk: It might take too long though, wine is pretty big. You should probably change the base branch to staging because of that
<fresheyeball>
and which options I should not copy over
<clever>
fresheyeball: everything under networking.interfaces i believe
<rotaerk>
I don't actually know what nox-review even does, I'm just running it because the github says to
<infinisil>
rotaerk: Nox review doesn't do anything other than building the things
<rotaerk>
so it just figures out WHAT to build, and then runs the build ... so me taking the nix-build command at the end of it, removing the erroneous package, and running it, has the same effect
<infinisil>
rotaerk: Eh but it doesn't matter much if you do so as MichaelRaskin suggested in his comment
Xiro` has quit [Ping timeout: 260 seconds]
<infinisil>
rotaerk: Most likely (I don't know nox internals)
<fresheyeball>
I ra:n nixops ssh -d web-infect web
<fresheyeball>
I ran: nixops ssh -d web-infect web
<fresheyeball>
it just hangs
<clever>
fresheyeball: maybe?, try to do a root login in DO and then a nixos-rebuild rollback
<infinisil>
Check logs in the web console
<fresheyeball>
clever: I never set a root password
<fresheyeball>
I am rebuilding
robstr has quit [Quit: WeeChat 1.9.1]
<fresheyeball>
infecting...
<fresheyeball>
ok so it infected
<fresheyeball>
now question is
<fresheyeball>
is there additional stuff I should try copying over?
nvh^ has joined #nixos
<fresheyeball>
or should I just set the root password and see some logs when it fails?
<infinisil>
Did you not check the logs?
<fresheyeball>
I deleted that vm
<infinisil>
Oh right you couldn't even access them
<fresheyeball>
now I have a fresh infection
roconnor has quit [Remote host closed the connection]
patrl has joined #nixos
roconnor has joined #nixos
<ndowens04>
still havent' figured out why i can't connect to mosh from my iphone yet, even set the connect option to mosh --server=/run/current-system/sw/bin/mosh-server
<ndowens04>
if I ssh into my machine and run that command it works
<{^_^}>
[nixpkgs] @matthewbauer pushed to master « picolisp: broken on darwin »: https://git.io/f4SW2
<Lisanna>
can hydra be configured to do things on build success? like tag builds?
<Lisanna>
(in the scm repos)
<ndowens04>
I upgraded system to nixos-unstable and i am getting arning: name collision in input Nix expressions, skipping '/home/ndowens/.nix-defexpr/channels/nixos' even though there is only one channel listing nixos https://nixos.org/channels/nixos-unstable
<Lisanna>
ndowens04 the "nixos" channel is just a different release branch of nixpkgs
<Lisanna>
oh, sorry, misread your thing I think
<{^_^}>
[nixpkgs] @matthewbauer pushed to master « gravit: only build on linux »: https://git.io/f4SWw
<ndowens04>
i deleted that dir to start over and did a channel force update to regen it and well getting that err
<{^_^}>
[nixpkgs] @matthewbauer pushed to master « chocolate-doom: only build on linux »: https://git.io/f4SWP
<ndowens04>
so use to "conventional" linux distros haha
<ndowens04>
not sure how to get rid of that error and wondering if it is gonna cause issues
<samueldr>
ndowens04: both your (root's) and your own channels are listing the same name (nixos) I think
<samueldr>
your system (root's) and your own channels **
<samueldr>
if you run nix-info, I believe it would confirm it
<ndowens04>
yup
<ndowens04>
so remove channel from user ?
<samueldr>
the biggest issue is that since it's skipping your own channels, a `[non-sudo] nix-channel --update` will seemingly do nothing
<samueldr>
ndowens04: yes, use the system's (root's) in that case
<ndowens04>
if i sudo nix-channel --update i get Nix search path entry '$HOME/.nix-defexpr/channels' does not exist, ignoring
<samueldr>
and with `sudo -i nix-channel --update`?
<samueldr>
I'm unsure if something changed with regards to how `sudo` operates on unstable, as I don't recall similar issues earlier this year
<ndowens04>
nope just as root
<ndowens04>
nvm its workingnow
<clever>
ndowens04: also, after removing a channel from a user, you still have to --update as that user to apply the removal
<samueldr>
clever: oh, good to know
warbo has left #nixos ["ERC (IRC client for Emacs 25.3.1)"]
<ndowens04>
wth nix-env -u
<ndowens04>
warning: name collision in input Nix expressions, skipping '/root/.nix-defexpr/channels/nixos'
<ndowens04>
y
<clever>
ndowens04: did you run nix-channel --update as the user you removed nixos from?
<ndowens04>
but nothing yup
<elvishjerricco>
Sweet. Using `nix-plugins` and `deployment.keys` from NixOps, I now have secure, declarative key management. Keys are extracted at eval-time from my `.password-store`, which prompts for a password to decrypt.
<clever>
elvishjerricco: nice
rardiol1 has quit [Read error: Connection reset by peer]
<elvishjerricco>
clever: Good tip on the `exec` thing. `nix-plugins` makes that really nice
<samueldr>
I'm don't know if relying on `<nixpkgs/...>` is the best idea, since I'm not using `<nixpkgs>` to build everything, `pkgs` comes from elsewhere and has overlays added
<elvishjerricco>
samueldr: It's possible for an `imports` to change `pkgs.path`, so trying to reference `pkgs.path` in `imports` is a paradox
freeman42x[nix] has quit [Ping timeout: 256 seconds]
<samueldr>
clever: my modules-foo may not be strong enough, looks like this is an internal API, that pkgs_path would be the one at the current point? The initial one given?
<clever>
samueldr: everything you put into _module.args, is passed to every module
<clever>
that is where pkgs itself comes from
<clever>
so you can just add your own custom things, and then use { pkgs_path, pkgs, lib, config, ... }:
<samueldr>
so it would be whatever I pass to eval-config I gather
tzemanovic has quit [Ping timeout: 248 seconds]
<clever>
yeah, thats one place you can insert it
<elvishjerricco>
Warning, I've had lots of problems with recursion using _modules.args
jtojnar has quit [Remote host closed the connection]
<clever>
it cant depend on the args its creating
philippD has quit [Client Quit]
philippD has joined #nixos
jtojnar has joined #nixos
<samueldr>
obviously, this is also an XY problem, where I'm asking this question, but another solution where I can add imports relying on `pkgs`'s path would be fine too :)
<{^_^}>
[nixpkgs] @matthewbauer pushed 3 commits to master: https://git.io/f4SlO
<{^_^}>
[nixpkgs] @dtzWill opened pull request #42463 → llvm_{38,39}: unconditionally set triples, not just w/musl → https://git.io/f4Slc
rardiol1 has quit [Remote host closed the connection]
<samueldr>
looks like that, at least for `imports`, this still causes an infinite recursion
<rardiol1>
I suppose it's supposed to say something else? I think there has been a relevant update in the last couple years?
<clever>
strange, what nix command gave that error?
<clever>
warning*
<elvishjerricco>
My bluetooth headset tries to auto-connect to whatever it was last connected to when I turn it on. For some reason this doesn't work quite right on NixOS. My machine connects to it (preventing other machines from connecting to it), but it doesn't send any audio and the headset never makes the beep that it uses to indicate something connected to it.
<elvishjerricco>
Anybody have any experience with this sort of problem on NixOS?
<elvishjerricco>
Works fine on macOS and iOS.
<worldofpeace>
Yep, are you using a DE?
<elvishjerricco>
Yea, xfce
<worldofpeace>
I think this is a problem with it not setting the audio profile for that device to A2DP. It's even worse in gnome.
<samueldr>
elvishjerricco: using pulseaudio? does it switch into HSP or Off in pavucontrol?
<elvishjerricco>
I am using pulseaudio. Let me check
<samueldr>
this week's release of pulseaudio changes default to A2DP worldofpeace
<samueldr>
so, hopefully soon nixos-unstable, then nixos-stable
<elvishjerricco>
Oh and btw, if I tell my machine to disconnect from the headset and then manually reconnect, it works fine
mounty has joined #nixos
<samueldr>
it's a bummer my headset seems to have an issue and reconnects to nothing :/
<worldofpeace>
yay! On nixos-unstable as long as the tests work :)
<samueldr>
so I can't even help
<worldofpeace>
There's some pacmd hack for it but I can't remember
<samueldr>
yay, got to stage-2 with my custom init
<rardiol1>
clever: nixos-rebuild. actually , I think it makes sense, I tried deleting something in the hope that it would just reset to default, but apparently not. What is supposedd to be in .nix-defexpr and NIX_PATH?
<elvishjerricco>
Well now I'm having a different bluetooth problem. `Connection Failed: DBusFailedError: Protocol not available...` :P
<elvishjerricco>
Not sure what's going wrong there
<clever>
rardiol1: with or without sudo?
roconnor has quit [Remote host closed the connection]
<rardiol1>
clever: what with or without sudo? the deleting? NIX_PATH?
<worldofpeace>
elvishjerricco: What's weird is that It works great in kde, aweful everywhere else
<clever>
rardiol1: nixos-rebuild
roconnor has joined #nixos
<elvishjerricco>
worldofpeace: Is there a workaround?
<rardiol1>
clever: with
<worldofpeace>
elvishjerricco: is it blueman?
<elvishjerricco>
worldofpeace: yea
<clever>
rardiol1: run `sudo -i` then check to see if nixos-rebuild warns again
<worldofpeace>
Yeah I'll try to find the workaround but It seems blueman has some problems of it's own too
<rardiol1>
clever: apparently not. what is nix-defexpr for anyway?
<clever>
rardiol1: thats where nix-env and nix-channel store channels
<worldofpeace>
elvishjerricco: first try `sudo rfkill unblock bluetooth' then 'sudo systemctl restart bluetooth'
<rardiol1>
clever: but, the user is not supposed to have channels, right? so I can delete it?
<elvishjerricco>
worldofpeace: Nope, `Protocol not available`, still :/
tzemanovic has joined #nixos
<clever>
rardiol1: ~/.nix-defexpr/channels should be a symlink
<elvishjerricco>
This was working earlier today. Ugh
<clever>
rardiol1: and the warning is more likely saying that somebody deleted it, when it should have existed
<samueldr>
elvishjerricco: I, very not often, have to delete and re-pair, not sure if it's that error message
<samueldr>
since the issue isn't closed, don't think it was in the last pulseaudio update :/
<elvishjerricco>
samueldr: Yep, been using your thing for a while now :)
<samueldr>
I knew I shared it with one of you two :)
<rardiol1>
clever: but NIX_PATH doesn't have a $HOME in it apparently
<elvishjerricco>
Oh I was wrong. It pairs, but doesn't connect
<elvishjerricco>
Still getting the same error
<elvishjerricco>
Might try rebooting...
tzemanovic has quit [Ping timeout: 256 seconds]
<clever>
rardiol1: sudo messes with env vars
<samueldr>
bluetooth on linux still feels so clunky :/
<worldofpeace>
Do you hear like a buzzing?
<elvishjerricco>
worldofpeace: Nope.
<samueldr>
and since every piece of hardware is its own special snowflake...
<worldofpeace>
It prob not worse battling :) Unless it does it a million more times
<elvishjerricco>
Alright rebooting fixed the connection problem
<Lisanna>
does nix's remote building have the sense to avoid machines that have very little free memory available?
<Lisanna>
"If there are multiple machines of the right type, Nix will prefer the fastest, taking load into account."
<Lisanna>
^ I have no idea what that means
<clever>
Lisanna: it locally tracks how much "load" each machine has, by counting how many jobs it has assigned to it
<clever>
Lisanna: and there is a speed-factor in the /etc/nix/machines file
<elvishjerricco>
samueldr, worldofpeace: Alright so when it does that auto connect thing, it doesn't even show up in pavucontrol
<Lisanna>
clever ok, so each client will just try to keep its builds balanced?
<clever>
Lisanna: yeah
<worldofpeace>
elvishjerricco: Yep I've seen that.
<elvishjerricco>
Ah, so the audio profile is "Off" by default
<Lisanna>
so for the machine overall the load *should* be more or less balanced globally
<elvishjerricco>
But trying to change it gives me `Failed to change profile to a2dp_sink`
<worldofpeace>
Do 'pacmd ls'
<clever>
Lisanna: but if each end-user does 1 job, they will all put it on the 1st file in /etc/nix/machines
<Lisanna>
how does it pick the first machine, if two identical machines are available
<Lisanna>
oh, crap
<worldofpeace>
Find the index of the bluetooth device.
<Lisanna>
that's lame
<clever>
Lisanna: rotate the file, so each end-user has a different idea of "1st" in the file
<elvishjerricco>
worldofpeace: Index?
<Lisanna>
lol
<clever>
Lisanna: thats basically what dns does :P
<worldofpeace>
elvishjerricco: the output will say 'index: somenumber newline name: <bluez_card something something'
<Lisanna>
maybe I could setup some kind of dns proxy system to use as a front for this remote building
<clever>
Lisanna: ssh wont be happy with you, mitm alerts everywhere
<Lisanna>
so it tries to resolve builder.example.com and then the dns will decide which machine to return based on actual system load / availability
<Lisanna>
clever inb4 I just make all the machines have the same host identity :^)
<worldofpeace>
elvishjerricco: pacmd set-card-profile number a2dp_sink
<Lisanna>
or will ssh still bitch about that, since now there'd be duplicate entries
<clever>
Lisanna: probably not
<elvishjerricco>
worldofpeace: The index is in the output of what?
<clever>
it cant tell the difference between duplicate hosts and just 2 IP's for the same host
<elvishjerricco>
Also, if I disconnect and manually connect, it automatically sets the right profile
<worldofpeace>
elvishjerricco: that's because autoconnect it not setting the profile i think
<Lisanna>
ugh, even though I have builders-use-substitutes turned on it's still insisting on downloading everything to the initiating machine before launchin the remote build
<Lisanna>
this is killing my laptop
<clever>
Lisanna: of course, it doesnt know what the remote machine has yet, until it has downloaded everything locally and is ready to start pushing things to the remote box
<Lisanna>
clever so that options just saves outbound traffic from the initiating machine to the target machine, not inbound traffic to the initiating machine
<clever>
Lisanna: yeah
<elvishjerricco>
worldofpeace: So how can i find this index?
<Lisanna>
bleh
<clever>
Lisanna: its for when the slave is on another lan, and cache.nixos.org can out-do your upload pipe
<Lisanna>
yeah, but my dl pipe isn't that great either
<Lisanna>
I wonder if anyone has written anything for complete remote building, i.e., "take this nix code and initiate the build from a remote host"
<elvishjerricco>
worldofpeace: Ah
<Lisanna>
I like keeping all my nix code on my laptop for speed of editing / convenience but having my laptop be the build coordinator here is really bleh