<clever>
thomasd: which derivation is getting a 1.0.3 of julia?
eadwu has joined #nixos
Guanin has quit [Ping timeout: 252 seconds]
<thomasd>
clever: like, which attribute?
<clever>
thomasd: yeah, or its name, from the .drv it started building
<thomasd>
clever: do you mean to say that it _should_ be using julia-1.1.0?
<thomasd>
clever: (at first glance)
<clever>
thomasd: your original msg says 1.0.3 is the worng one, so which derivation is using itt?
<thomasd>
clever: can I check with nix-shell which derivation is being used instead of filtering through a ton of logs from `nix-build -A parallel-julia`
<clever>
thomasd: yeah, you can `nix-shell default.nix -A parallel-julia.env` and then `ghc-pkg list`
ma27 has joined #nixos
<thomasd>
clever: julia is a system dependency
<thomasd>
clever: so it does not show up in `ghc-pkg list`
<clever>
thomasd: ah, then youll want something more like `env | grep --color julia`
o1lo01ol1o has quit [Remote host closed the connection]
<cthachuk>
clever: thanks! that makes good sense and boost is compiling from source now. Apologies for the many questions, but now I'm curious why building boost from source is only using 1 of my 16 cores. Is there a setting in my configuration.nix I need to configure to enable multiple threaded builds by default by gcc?
ayerhart has quit [Read error: Connection reset by peer]
<gchristensen>
buildCores iirc
<clever>
yep, nix.buildCores
<cthachuk>
thanks both; I've learned a lot about nix from this channel today
mkoenig has joined #nixos
ayerhart has joined #nixos
kreisys has joined #nixos
Edes has joined #nixos
wfranzini has quit [Remote host closed the connection]
<gchristensen>
=)
selfsymmetric-mu has quit [Ping timeout: 264 seconds]
fusion809 has quit [Remote host closed the connection]
Rusty1 has joined #nixos
<bsima>
how do i call npm install inside the context of a derivation? i think its failing bc of read-only filesystem
<aleph->
bsima: What's the deriv you're trying? Paste it please.
<yokyolu>
what's the difference between callHackage and hackage2nix?
emily has quit [Remote host closed the connection]
emily has joined #nixos
charukiewicz has joined #nixos
<charukiewicz>
What's the best way to browse nixops options? E.g. I see many deployment.ec2.* options defined in the source code of the NixOS/nixops repo along with descriptions, but what's the best way to browse these other than reading the source code?
vk3wtf has joined #nixos
<clever>
yokyolu: i think hackage2nix turns a hackage snapshot into a haskellPackages, but callHackage lets you just directly load a given version of a given package
<charukiewicz>
simpson: ah yes! thank you. I glossed over that in my first search.
sigmundv__ has quit [Ping timeout: 250 seconds]
ng0 has quit [Ping timeout: 256 seconds]
<infinisil>
Hey, when I change a gid/uid for a user with a specific name, NixOS doesn't actually do the change, but instead warns "not applying GID change of group foo"
<infinisil>
Can I somehow make the change?
tertl3 has joined #nixos
mobile_c has joined #nixos
<tertl3>
not sure why i just installed this OS, but here goes
<clever>
infinisil: thats to prevent the horid mess that happens when the uid changes, and the files arent chown'd over
<infinisil>
clever: But I want to chown the files over (with systemd.tmpfiles or so)
<ajs124>
if you're using this for a "service user", why not use systemd's dynamicUser stuff? or is this for a "real" user?
<clever>
infinisil: i would just do the standard linux solution, manually edit /etc/passwd to sync it over, manually chown the home dir, and then add a nixos twist on the end, fix /var/lib/nixos/uid-map
<infinisil>
ajs124: It's for a service user, but I can't use dynamicUser (because it interacts with other services, need to be able to set group and stuff)
<infinisil>
clever: I guess that works.., very ugly though
reallymemorable has quit [Ping timeout: 246 seconds]
<infinisil>
And I can't really put this in a NixOS configuration
<clever>
infinisil: in the past, nixos would forget the uid when you delete a user from the cfg, and it may wind up with a different uid when you re-enable, causing similar breakage
<clever>
infinisil: the uid-map file sovles that, by remembering the uid even if you delete a user
<clever>
and its only a one-time thing, for when the uid defined in the nixos module is changed
<infinisil>
clever: Why can't it just update {u,g}id-map with the latest id if it's assigned statically?
<infinisil>
Probably it could, just not implemented
phizzz has quit [Quit: leaving]
<clever>
infinisil: the problem, is that you must find every file owned by that user, and fix the uid
<clever>
infinisil: which means recursively searching every mounted FS
<infinisil>
I guess it makes more sense for services (which is what I'm thinking of), where they know exactly where their files are and they can fix them up themselves
<infinisil>
(with e.g. systemd.tmpfiles)
<infinisil>
But really, NixOS should support that
<infinisil>
imo
<clever>
infinisil: update-users-groups.pl would need to be updated, to support forcing the uid change
<infinisil>
If I'm assigning `users.users.foo.uid = 10452`, I'm already looking for manually changing permissions
<infinisil>
Yeah
<clever>
infinisil: if you had assigned that from the first deployment, it would always have that uid, and the problem wouldnt exist
<infinisil>
Yeah
<infinisil>
Although, I guess it doesn't make much sense to want a specific number as a static id
emily has quit [Remote host closed the connection]
<infinisil>
So an alternative would be to just do a read from uid-map and import the static ids you want into your configuration.nix
<infinisil>
Doesn't necessarily work with multiple machines though (sharing the same configuration.nix or parts of it)
<clever>
infinisil: the whole point of the uid-map, is that it keeps them static for you, so you dont have to set things in configuration.nix
<charukiewicz>
Has anyone dealt with using nixops to deploy to EC2 instances and after deployment assigning them to target groups for the purpose of adding them to a load balancer? There's a ton of options for configuring individual instance deployment, but I don't see much in the way of AWS load balancer config.
<infinisil>
clever: Yeah true dat, I'm having the scenario of not restoring /var/lib/nixos on a new install in mind
<infinisil>
(but restoring the services directories
<clever>
DigitalKiwi: another package appears to be doing a --replace, to just delete that dependency
<mek42_laptop>
what happens if there is a power loss during nix-env --upgrade?
<clever>
mek42_laptop: at the top-level, none of the changes will happen, the entire nix-env operation is atomic
<clever>
mek42_laptop: at lower levels, some FS's can corrupt some storepaths, leaving random (recently edited) files truncated, `nix-store --verify --check-contents` will find that
jasongrossman has joined #nixos
jasongro` has joined #nixos
reallymemorable has joined #nixos
jasongrossman has quit [Ping timeout: 257 seconds]
Supersonic has quit [Ping timeout: 258 seconds]
jasongro` has quit [Quit: ERC (IRC client for Emacs 26.1)]
emily has quit [Remote host closed the connection]
o1lo01ol1o has quit [Remote host closed the connection]
Edes has quit [Remote host closed the connection]
rcshm_ has quit [Remote host closed the connection]
rcshm_ has joined #nixos
counting1ort has joined #nixos
<DigitalKiwi>
i got it further
kp__ has quit [Quit: WeeChat 2.3]
<DigitalKiwi>
it needs something else that isn't in nixpkgs but is on pypi or i could build from source maybe... it's by the same author as the one i'm trying
rcshm_ has quit [Read error: Connection reset by peer]
xkapastel has joined #nixos
andreabedini has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
marusich has quit [Remote host closed the connection]
yokyolu has quit [Ping timeout: 250 seconds]
reallymemorable has joined #nixos
ayerhart has quit [Quit: ayerhart]
slack1256 has joined #nixos
ayerhart has joined #nixos
<slack1256>
Does nix-serve allow passing nixpkgs revisions via nix-channel --updates ?
<clever>
slack1256: nix-serve just shares whatever is currently in /nix/store/, and doesnt really care about nixpkgs versions
<ar1a>
whenever i home-manager switch or nixos-rebuild switch, avahi breaks until i reboot. for example `ping: nobelium.local: System error` i can't resolve any .local domains. avahi-daemon.service is running and doesn't report any errors. my avahi config is https://github.com/ar1a/dotfiles/blob/master/nixos/configuration.nix#L90-L95
<slack1256>
clever: Got it, So how do I get a common nixpkgs revision on nix-channel?
<ar1a>
i'm running on nixpkgs-unstable but it's been a problem ever since i installed nixos at the start of the year
<clever>
slack1256: if you nix-channel --update all the machines at the same time, then they will usually match up, you could also use things like nixops to manage things more centrally
<clever>
ar1a: what does `sudo nix-channel --list` report?
<clever>
ar1a: it is not safe to run nixos from the nixpkgs channels, and can potentially brick the machine
<clever>
ar1a: if you want the nixos unstable channel, then you must use nixos-unstable
rcshm__ has quit [Read error: Connection reset by peer]
<ar1a>
clever: ah yeah i don't actually need nixpkgs-unstable anymore, how can i downgrade
<clever>
ar1a: just re-run nix-channel --add, with the name set to nixos
rcshm_ has joined #nixos
<ar1a>
alright, ill downgrade and reboot and see if the issue is still there
rcshm_ has quit [Read error: Connection reset by peer]
<clever>
ar1a: its not really a downgrade, nixpkgs-unstable and nixos-unstable both follow master, and there is only a 2 hour difference between the versions
rcshm_ has joined #nixos
<clever>
ar1a: the major difference, is that nixos-unstable wont update, if things like grub are broken, so it is far less likely to brick the entire machine
rcshm__ has joined #nixos
rcshm_ has quit [Remote host closed the connection]
o1lo01ol1o has quit [Remote host closed the connection]
<{^_^}>
[nixos-weekly] @domenkozar pushed to netlify « WIP »: https://git.io/fhFtb
<nimblepoultry>
hi, could someone clarify how I should be importing and using `nixos/modules/profiles/hardened.nix` and `pkgs/os-specific/linux/kernel/hardened-config.nix` in my configuration.nix ? The Profiles section of the manual seems to only refer to nix-env?
o1lo01ol1o has joined #nixos
<clever>
ping: system76.local: System error
<clever>
ar1a: looks like mine is also broken, done many switch's
jackdk has quit [Ping timeout: 245 seconds]
<ar1a>
clever: glad to know its not something ive done
<clever>
nimblepoultry: <nixpkgs/nixos/modules/profiles/hardened.nix> or (pkgs.path + "/nixos/modules/profiles/hardened.nix")
<clever>
ar1a: given what youve said, i'm not sure any of my machines have working .local, since i rarely reboot and have switched on nearly all of them, let me see if i can fix it locally....
<nimblepoultry>
thanks clever
<clever>
ar1a: `avahi-browse -at` can find system76, so avahi itself is working
<ar1a>
interesting, that command doesn't show up all my .local domains here
<ar1a>
it only shows my laptop and my homelab, not the laptop my gf is using nor my main pc
<ar1a>
all which use .local most of the time
<clever>
+ enp3s0 IPv4 system76 _ssh._tcp local
<clever>
+ enp3s0 IPv4 system76 [a0:af:bd:82:39:0d] _workstation._tcp local
<clever>
it doesnt show any .local's, but it can see the laptop
<{^_^}>
[nixos-weekly] @domenkozar pushed to netlify « WIP »: https://git.io/fhFqT
<clever>
ar1a: hmmm, if i stop nscd, then ping fails much faster...
<clever>
openat(AT_FDCWD, "/run/opengl-driver/lib/libnss_mdns.so.2", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
o1lo01ol1o has quit [Ping timeout: 245 seconds]
<clever>
and ping is trying to load the libnss_mdns library, to add .local support to itself
pie___ has joined #nixos
<{^_^}>
[nixos-weekly] @domenkozar pushed to netlify « WIP »: https://git.io/fhFqk
emily has quit [Remote host closed the connection]
<{^_^}>
[nixos-weekly] @domenkozar pushed 3 commits to master: https://git.io/fhFqY
<clever>
ar1a: looking like it, investigating that end now...
<{^_^}>
[nixos-weekly] @domenkozar pushed 2 commits to production: https://git.io/fhFq3
vmandela has joined #nixos
<{^_^}>
[nixos-weekly] @garbas pushed commit from NixOS Weekly Robot to gh-pages « Preview of '26eeed0b0f8a622a939a9aa852cd9fc94366f778' commit built by Travis-CI \#496337900 »: https://git.io/fhFqs
<ar1a>
clever: just a quick tangent, what's your uptime?
<clever>
ar1a: 31 days on the desktop where i'm debugging
<ar1a>
dayum
<clever>
61 days on the laptop
<clever>
34 days on the nas
<{^_^}>
[nixos-weekly] @garbas pushed commit from NixOS Weekly Robot to gh-pages « Release of (commit '9c39604fcf8892d47d9d088edf57bd98f1a5d689') built by Travis-CI \#496337993 »: https://git.io/fhFqG
<ar1a>
our router has 60 days and our vps has 102
<clever>
34 days on the router
<clever>
i think there was a power blip ~34 days ago, lol
<clever>
ar1a: only thing that comes to mind i the uppercase letters in the hostname
<clever>
and in my case, the testing is happening between 2 nixos machines
<clever>
ar1a: the nas.local was also failing, but when i added publish.address = true; to it, then it started to work
vk3wtf has quit [Ping timeout: 258 seconds]
<ar1a>
ill try changing them to all lower
endformationage has quit [Quit: WeeChat 2.3]
Rusty1 has quit [Quit: Konversation terminated!]
<ar1a>
strange, so now it can ping one box but cant ping another. im trying to ping my gfs computer and shes running `sudo tcpdump -i wlp4s0 -p -n port 5353` and on my machine i run `ping uranus.local` - nothing is showing up on her tcpdump
<ar1a>
from my ip at least
<clever>
ar1a: do you see packets leaving your machien, on 5353?
illegalprime has quit [Ping timeout: 246 seconds]
<ar1a>
same command on mine?
<clever>
yeah
rcshm_ has quit [Read error: Connection reset by peer]
rcshm_ has joined #nixos
<clever>
a common thing when debugging networking isuses, just tcpdump all the machines!
<ar1a>
16:51:42.531289 IP 172.19.128.114.5353 > 224.0.0.251.5353: 0 A (QM)? uranus.local. (30)
<ar1a>
looks like it
<ar1a>
dunno what that 224 ip is though
<clever>
are the packets leaving x? do they pass thru the router? do they reach the target?
<clever>
but under linux, its just a plain old nvme device
<clever>
just `zpool add tank cache /dev/nvme1n0p1` and your done
<ar1a>
you call your hdd array tank?
<ar1a>
nice
<clever>
only in example commands :P
<clever>
i try to match the pool name and hostname up
<ar1a>
i use rpool because the wiki told me to
<ar1a>
🤷
<ar1a>
my gf uses cuffs because on the old install when cryptsetup would ask for a password in the locking screen we edited it so it would say "please unlock the fucking cuffs: "
<ar1a>
lol
<{^_^}>
[nixpkgs] @vcunat pushed 2 commits to release-18.09: https://git.io/fhFmr
<clever>
lol
<clever>
previously, nixos required some fairly ugly hacks (luks key for swap+root on a 3rd partition, that is luks'd with a passphrase)
<clever>
and i didnt like that, so my zfs is on lvm (for swap), and lvm is on luks
slack1256 has quit [Remote host closed the connection]
<clever>
but nixos has recently gained the ability to temporarily save your passphrase to an env var, and use it on all disks
<clever>
so if you use 1 passphrase on both the zfs and swap, it just unlocks both at once
<ar1a>
i saw that option yeah
charukiewicz has quit [Quit: WeeChat 2.3]
<{^_^}>
[nixpkgs] @FRidh pushed 192 commits to staging-next: https://git.io/fhFmD
<{^_^}>
[nixpkgs] @FRidh pushed 739 commits to python-unstable: https://git.io/fhFmp
rcshm_ has quit []
<eyJhb>
And I have NO clue why.. Giving it a couple of reboots fixes it.. But nothing functions pretty much, until I don't have those failed services
<ar1a>
write a cronjob that runs start every 10 minutes?
<eyJhb>
That seems like a very very bad solution
<clever>
eyJhb: does systemd register it as failed in `systemctl status`? you could change the restart conditions to have systemd itself restart on failure
<eyJhb>
clever: it shows as degraded, the output from `systemctl
<eyJhb>
status` can be seen in the gist `systemctl-status` :D
<eyJhb>
But `systemctl
<eyJhb>
failed` does yes, also shown :)
<clever>
eyJhb: then you likely just need to set the right restart conditions on the service
<eyJhb>
But the `user-dir`, is a NixOS service, right?
lawlesseel has quit [Ping timeout: 256 seconds]
<eyJhb>
I haven't set it or activated it at all, so no clue how I would do that
<clever>
eyJhb: i dont see a user-dir service on my box
ij has joined #nixos
<{^_^}>
[nixpkgs] @dywedir opened pull request #56129 → alacritty: add wayland, libxkbcommon to rpath → https://git.io/fhFYL
<eyJhb>
clever: Any idea how I can check what has activated that service?
<clever>
ah, its user-runtime-dir, thats why i didnt see user-dir on my box
<eyJhb>
Might be because of lightdm as a login service? It is listed under "login stuff" in systemd
ij has quit [Ping timeout: 246 seconds]
<clever>
eyJhb: i use slim, and havent had any problems, despite slim being "deprecated"
<eyJhb>
clever: but you still have user-runtime-dir as a service running?
<clever>
user-runtime-dir@1000.service loaded active exited /run/user/1000 mount wrapper
<clever>
eyJhb: yep
<eyJhb>
Can't I just disable it all, or do I need a DisplayManager ?
<eyJhb>
Hmm, but then it would still fail I would guess.... Are you running the stable channel?
<clever>
tmpfs on /run/user/1000 type tmpfs (rw,nosuid,nodev,relatime,size=3284088k,mode=700,uid=1000,gid=100)
<clever>
eyJhb: that unit is responsible for creating this tmpfs, if you login via any method (graphical, or ssh)
<clever>
and if you logout from everything, it will destroy it
<clever>
i'm runnign nixos-unstable
<eyJhb>
But then, it wouldn't be any different using slim or lightdm, correct?
knupfer has joined #nixos
<clever>
eyJhb: yeah, they should perform identially
<clever>
eyJhb: do you have any weird things in your fileSystems. config?
<eyJhb>
But there might be patches in nixos-unstable then...
<eyJhb>
I am seeing more and more, that my "channels" are very very much screwed up..
<clever>
eyJhb: what does `sudo nix-channel --list` report?
xkapastel has joined #nixos
<eyJhb>
clever: just updated it, currently I am just tracking all the unstable in the whole wide world - https://termbin.com/5ole
vk3wtf has joined #nixos
<clever>
eyJhb: yeah, that looks like it should work
<eyJhb>
But there must be a better way to do this.. Because I would like it to track my own repo... But a quest for another time
daniele- has quit [Quit: daniele-]
zupo has joined #nixos
<eyJhb>
clever: The joys of having VirtualBox activated... It likes to rebuild all the time. Loving it
<clever>
eyJhb: i dont have the extension pack enabled, so it mostly comes from the cache
emily has quit [Remote host closed the connection]
Tucky has joined #nixos
<eyJhb>
Yeah, I have it enabled, as it is needed for most folder sync (as far as I remember)... I just rebuild it 2 days ago, becaue I updated my stable channel back then too!...
emily has joined #nixos
alex`` has joined #nixos
<clever>
eyJhb: i have shared folders without it
<eyJhb>
Considering disabling it, but sometimes I use the VNC etc. for testing a project for work, so sometimes I need it besides folder sync
<clever>
and its 4am here, i should get some sleep
<eyJhb>
clever: 09 am here, currently having a lecture on - Engineering Mathematics for electronical engineers
<eyJhb>
Enjoy your sleep and thanks!
infinii has joined #nixos
Havvy has joined #nixos
ircuser5678 has joined #nixos
<ircuser5678>
Hi. I've made an upgrade from 17.09 to 18.09 and while the system works fine I have a small problem with dhcpcd. The message during bootup is "A startjob is running for dhcp client..." and it slows down the boot by about 20 seconds. The thing is, I'm using a static ip address setup on this machine. With 17.09 I didn't had this problem and I made the necessary changes regarding "network." changes from 18.03.
<pbb>
I did set it to postgresql_11. Postgresql 11.1 is not available anymore in the current nixpkgs and I'm blocked from upgrading my system.
<pbb>
Until either my applications or the database fix their shit
<pbb>
cc ocharles
jackdk_ has joined #nixos
<pbb>
or is it maybe necessary to manually reimport the databases for the upgrade? I'm still confused why postgresql_96 has an own package as a minor version, while 11.x shares one package.
<etu>
pbb: fwiw, I know that postgres was more "breaky" in the past even on minor versions, I think they have said that they should stop doing that... But yeah...
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
gagbo has quit [Quit: I'm out !]
<etu>
like 9.5 -> 9.6 required you to manually migrate the database afaik
zupo has joined #nixos
zupo has quit [Client Quit]
hedning has quit [Remote host closed the connection]
Makaveli7 has joined #nixos
gagbo has joined #nixos
hedning has joined #nixos
<pbb>
etu: if 11.1 -> 11.2 requires manual intervention, it should have a seperate package, no?
hedning has quit [Remote host closed the connection]
hedning has joined #nixos
<pbb>
ín any case it should not break when updating channels with my system configuration
<ar1a>
it probably wasn't tested
<pbb>
it was tested, by me :p
<etu>
pbb: Yeah, I'd agree with that
<etu>
But that upgrade-path was probably not tested :/
<pbb>
for me the only viable option is now to fix it to a custom 11.1 package in my system configuration, so that tomorrow's autoupgrade doesn't break everything again
hedning has quit [Remote host closed the connection]
<etu>
pbb: Sometimes I've just done a fetchUrl on an older version of unstable and overwrite that specific package that broken in an overlay :p
<etu>
That way I don't need to compile anything :D
<pbb>
ah, I will not put it in an overlay, but just pass the other package to the module directly
<pbb>
and I want to run 11.1 with current dependencies and stuff, even if I have to compile it
arjen-jonathan has joined #nixos
<ar1a>
are you sure its not something you did?
Glider_IRC__ has joined #nixos
<laas>
would anyone have any clue as to why wireless networking would only work in the live NixOS installation environment and not when it's actually installed on my laptop? I have enabled non-free firmware.
<teto>
laas: could be kernel issue/unloaded module or service. Nothing in journalctl or dmesg ?
<laas>
my internet connection on my phone is spotty atm so sorry if my replies don't reach you
<laas>
my internet connection on my phone is spotty atm so sorry if my replies don't reach you
__Sander__ has joined #nixos
<{^_^}>
[nixpkgs] @jacereda opened pull request #56136 → cereal: run test suite on ghc-8.6 → https://git.io/fhF3q
<{^_^}>
[nixpkgs] @jacereda opened pull request #56137 → cereal: run test suite on ghc-head → https://git.io/fhF3Y
<laas>
to restate my problem: dmesg says nothing and wpa_supplicant gives no errors. I'm using 4.20.10 with unfree drivers too. Running `ip a` shows me that I'm connected but it only shows a link/ether field and an inet6 field
<ircuser5678>
Hi. I've made an upgrade from 17.09 to 18.09 and while the system works fine I have a small problem with dhcpcd. The message during bootup is "A startjob is running for dhcp client..." and it slows down the boot by about 20 seconds. The thing is, I'm using a static ip address setup on this machine. With 17.09 I didn't had this problem and I made the necessary changes regarding "network." changes from 18.03.
gagbo has quit [Ping timeout: 240 seconds]
<laas>
however pinging an IPv6 address also seems to tell me that the Network is unreachable
<ar1a>
you always have an ipv6 address its the link local
<{^_^}>
[nixpkgs] @jacereda opened pull request #56138 → psqueues: run test suite on ghc-8.6 → https://git.io/fhF3s
markus1189 has quit [Ping timeout: 246 seconds]
<ar1a>
you havent connected to the AP
<ar1a>
either use wpa_supplicant or network manager
<{^_^}>
[nixpkgs] @jacereda opened pull request #56139 → psqueues: run test suite on ghc-head → https://git.io/fhF3Z
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/fhFZu
hedning has joined #nixos
reallymemorable has joined #nixos
wfranzini has quit [Ping timeout: 250 seconds]
ninjin has quit [Remote host closed the connection]
ninjin_ has joined #nixos
<nyanloutre[m]>
Hello, I was looking at the KDE update script as I was wondering if anyone tried to make it use the metadata files (which contain file hashes) instead of downloading every archive and compute the hash afterward
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/fhFZr
<nyanloutre[m]>
Is there anything preventing this to work ?
reallymemorable has quit [Ping timeout: 244 seconds]
<simpson>
nyanloutre[m]: If the hope is to reuse other folks' hashes in Nix expressions, then unfortunately those hashes are usually not reusable as-is.
<nyanloutre[m]>
even though it's sha256 ?
<nyanloutre[m]>
because this is something that work with fetchfrompypi for example (I can use the hash found on the pypi website)
<simpson>
Hashes of single files, like tarballs, can be fine; hashes of directories usually aren't.
<nyanloutre[m]>
in the case of kde apps it seems to be hashes of tarballs
emily has quit [Remote host closed the connection]
<eyJhb>
Can nix-collect-garbage show how much space would be freed up?
emily has joined #nixos
<betaboon>
eyJhb: --dry-run
<srhb>
eyJhb: Yes, with --dry-run, but it can't simulate deleting generations (the gcroots)
<eyJhb>
That just shows the profiles deleted, not space :/
<eyJhb>
`nix-collect-garbage
<eyJhb>
dry-run
<eyJhb>
delete-older-than 30d` currently
<srhb>
Huh, I was quite sure that it used to work.. Maybe that's just my imagination.
ottidmes has joined #nixos
<eyJhb>
And now I am at it.. Is `nix-store
<betaboon>
srhb: same here
<eyJhb>
optimise` also a good way to go?
<srhb>
eyJhb: I always use optimize
<eyJhb>
Hmm, it doesn't.. I think it only does when NOT doing a dry-run :/
<eyJhb>
Well, got 30 GB of space back! ;)
<eyJhb>
Now trying
<eyJhb>
optimise.. Wondering how long it will take
<srhb>
eyJhb: You can enable auto optimization to have it happen whenever a store path is installed
jacereda has joined #nixos
arjen-jonathan has joined #nixos
orivej has quit [Ping timeout: 245 seconds]
<jacereda>
What's the process to be added to extra-users for @GrahamcOfBorg? Just a PR to its repo?
<srhb>
jacereda: Basically yes.
<jacereda>
srhb: ok, thanks
init_6 has quit []
<eyJhb>
Would be a nice addition I would guess, but I am just curious how much space will be saved!
<eyJhb>
Any good optimisations for SSDs?
<srhb>
eyJhb: In general, I think the best approach is having a compressed filesystem with auto optimization
<srhb>
eyJhb: I usually have about twice as much real data as I have compressed.
<{^_^}>
[nixpkgs] @callahad opened pull request #56147 → linux: Enable support for Self-Encrypting Drives → https://git.io/fhFn6
rprije has quit [Ping timeout: 264 seconds]
<eyJhb>
srhb: compressed filesystem? zfs? btrfs?
<eyJhb>
Hmm, yeah. Just did a little cleanup generally.. Got from 90% used, to 30% used currently /250 GB ssd), and still compressing and cleaning up
<srhb>
eyJhb: zfs.
<srhb>
eyJhb: Right now my /nix/store has a compressratio of 2.42x :)
ThatDocsLady has joined #nixos
<eyJhb>
srhb: running any encryption on top of that?
<srhb>
eyJhb: Below, actually.
<srhb>
eyJhb: luks -> zfs
ThatDocsLady has quit [Read error: Connection timed out]
<eyJhb>
Considering it, but doesn't that have some complications? Have heard zfs is not happy without ECC memory, and I am running on a x230
ThatDocsLady has joined #nixos
<srhb>
eyJhb: No filesystem is happy without ECC, but I certainly trust ZFS more than basically any other.
hedning has quit [Quit: hedning]
<eyJhb>
Considered btrfs more than zfs. But I like the decuplication of them..
<srhb>
Don't use dedup on zfs.
<eyJhb>
Why not?
<srhb>
You'll run out of memory and then it'll eat your system.
<srhb>
You can google for details :)
<eyJhb>
Is btrfs any better at it?
<srhb>
I wouldn't use btrfs at all.
lassulus has quit [Ping timeout: 246 seconds]
<srhb>
(It ate my system)
<eyJhb>
Hmm, might consider trying out zfs.. But currently ext4 is serving me well!
<etu>
eyJhb: I swapped away from btrfs to ext4 because I had bad luck with btrfs.
<etu>
eyJhb: But to each their own, I also hear success-stories from btrfs from people from time to time
<ottidmes>
eyJhb: I am a long time btrfs user, it works great until it doesn't, which will be when it bricks your file system due to bugs, throwing all the guarantees you thought you would get with it out of the window, so I am going with ZFS with all my new systems
* etu
still have hopes for bcachefs
<eyJhb>
So.. ZFS > More tested ?
<srhb>
etu: Me too
<ottidmes>
my desktop has been running btrfs for 5 years without any issue (I am not using any advanced features though), but on one of my servers, it got bricked, unable to recover, and had to reinstall
<jasongrossman>
I too use a system that can't take ECC. I'm sure ZFS is at least as good as anything else. The recommendation to always use ECC if you use ZFS was never right, unless it was a shorthand for "always use ECC with any file system", which is OK if you can but useless advice if you can't.
<srhb>
jasongrossman: Exactly.
<srhb>
I think it's just that zfs users tend to be more concerned about non-ecc as people, not because of zfs..
<jasongrossman>
srhb: Exactly.
schjetne has quit [Ping timeout: 250 seconds]
<eyJhb>
But.. Don't think my x230 would enjoy ECC, but at least it has 16 GB of ram.. The old gal...
* etu
is very concerned about using ecc wherever possible but have never used zfs... yet....
<jasongrossman>
Another advantage of ZFS is that a lot of the people in this group use it, so you can come here for help.
<jasongrossman>
This should really be in #nixos-chat.
<eyJhb>
Not saying anything, but.. Tried to get help with what the hell g++ package in called in NixOS, still don't know :p But yeah, should be..
<jasongrossman>
Of course a lot of us (although not me) use ext4 too.
<eyJhb>
But I really don't fancy reinstalling my OS:.. :/
<srhb>
eyJhb: qyliss: No, it'll happen on registration instead.
<srhb>
eyJhb: Which is far nicer.
<eyJhb>
registration?
<srhb>
eyJhb: (In other words, you should run it once to catch old paths even after activating it)
<eyJhb>
rebuild too?
<srhb>
eyJhb: When a store path is added, basically.
<etu>
eyJhb: cloud hosted irc clients
vmandela has quit [Quit: Leaving]
<srhb>
eyJhb: I don't understand what you mean. The idea is that whenever a new path is added, it is checked for duplicate files, and those are hard linked together.
<eyJhb>
So when I add new stuff in configuration.nix and rebuild, it will check for that too, as there might be somehting ned added to nix/store, yeah
<eyJhb>
Seems nice!
infinee has joined #nixos
<eyJhb>
"IRC! Internet Relay Channel. It's how hackers talk when then don't want to be overheard" - IRCCloud. Just wanted to share...
<{^_^}>
[nixos-weekly] @domenkozar pushed to simplify-ci « Simplify CI due to netlify »: https://git.io/fhFcK
justanotheruser has quit [Quit: WeeChat 2.2]
srl295 has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixos-weekly] @domenkozar opened pull request #82 → Simplify CI due to netlify → https://git.io/fhFcX
wfranzini has quit [Remote host closed the connection]
Phillemann has quit [Quit: WeeChat 2.3]
<eyJhb>
Does anybody have a nice way, of e.g. specifying that all pdf should open with zathura, all pngs, with x and so on? Can't seem to find any nice elegant solutions
wfranzini has joined #nixos
timor has joined #nixos
<__monty__>
xdg-open?
<ij>
globally, declaratively or just at all?
<__monty__>
Or use ranger's rifle : >
<{^_^}>
[nixos-weekly] @domenkozar merged pull request #82 → Simplify CI due to netlify → https://git.io/fhFcX
<{^_^}>
[nixos-weekly] @domenkozar pushed 2 commits to master: https://git.io/fhFcF
<eyJhb>
xdg-open opens stuff, I want to specify how it will open it :) ij globally would be fine? But just.. Generally..
<__monty__>
eyJhb: That's part of xdg-open. I'm not sure how to set it but I know you can set it.
<Twey>
How should one repair a corrupted db.sqlite?
<Twey>
I was hoping nix-store --repair --verify would do it, but it tries to connect to the daemon, and the daemon doesn't work without a good db.sqlite
dermetfan has quit [Ping timeout: 240 seconds]
silver has joined #nixos
zupo has joined #nixos
ij has quit [Ping timeout: 250 seconds]
<eyJhb>
Twey: nixos-rebuild boot
<eyJhb>
repair ?
<eyJhb>
Sorry.. I cannot do double dash... Need to fix it soon..
<eyJhb>
But nixos-rebuild boot -.-reapir (minux the .)
<srhb>
Twey: Usually dump and re-import it. Making sure to take backups
<srhb>
Twey: There's some sqlite pragma to make it check it
<srhb>
Twey: So basically, whatever tools sqlite recommend.
mmlb has quit [Ping timeout: 255 seconds]
wfranzini has quit [Remote host closed the connection]
adamt has joined #nixos
<WilliamHamilton[>
I'm repeatedly encountering the error `GLib-GIO-ERROR **: 14:14:51.146: No GSettings schemas are installed on the system` when trying to launch a program with steam-run
alex`` has joined #nixos
<{^_^}>
[nixpkgs] @NeQuissimus pushed 0 commits to xsel-08-01-10: https://git.io/fhFC8
<sphalerite>
Does anyone know how to get libvirt USB redirection working on nixos? Relevant config pieces are: `virtualisation.libvirtd.enable = true; users.users.linus.extraGroups = [ "libvirtd" ]; environment.systemPackages = with pkgs; [ spice_gtk virtmanager ];`
<sphalerite>
when I try to redirect a USB device from virt-manager, I get "Error setting USB device node ACL: 'Error setting facl: Operation not permitted' (0)"
hyper_ch2 has joined #nixos
steell has joined #nixos
<sphalerite>
Twey: yeah, you basically can't repair it :(
<{^_^}>
[nixops] @AmineChikhaoui pushed to master « manual: fix output of vbox destroy example »: https://git.io/fhFW7
regulus_ has quit [Ping timeout: 255 seconds]
winem_ has joined #nixos
jasongrossman has joined #nixos
agander has quit [Read error: Connection reset by peer]
jacereda has quit [Ping timeout: 250 seconds]
regulus_ has joined #nixos
tmaekawa has joined #nixos
tmaekawa has quit [Client Quit]
orivej has joined #nixos
<mdash>
ivegotasthma: you can check out the fork and add it to your NIX_PATH
<mdash>
I'm having fun putting together a git repo with all my machines' nixos, home-manager, and dotfile configs. there's plenty of examples of these on github
<{^_^}>
[nixpkgs] @thoughtpolice opened pull request #56151 → dockerTools: mark store-path-to-layer.sh as executable → https://git.io/fhFln
<mdash>
what I'm wondering is, what do people who like this choose for secrets management? I obviously don't want to put my private keys and passwords into the same storage
jacereda has joined #nixos
alter-schjetne has joined #nixos
<ldlework>
i'm currently just using lastpass to store my secrets which is convienent for me for sharing secrets with family members and using yubikey
timor has quit [Ping timeout: 258 seconds]
<ldlework>
people without fail bring up some breech they had or some problem with their browser extension or something everytime I mention them, so i'll save someone the trouble by doing that
<georgyo>
@mdash greatly depends on scale. If it is just a few machine create a template file with blanks for your needed variables. And then on the actual machines, hand populate
<ldlework>
georgyo: what do you recommend for larger deployments?
schjetne has quit [Ping timeout: 246 seconds]
steell has joined #nixos
<georgyo>
There are git tools that can mark files as secret, and automatically encrypt/decrypt on commit/checkout. However you still need to get the key everywhere
<mdash>
ldlework: yeah, i'm using bitwarden_rs for this currently
<mdash>
I am not sure if i'm quite ready to put private keys in it
<mdash>
my next challenge is that I'd like to move my syncthing configuration into nix expressions
<mdash>
but that means deciding on how to generate some xml config files
<ldlework>
mdash: are you prepping for the Nix Challenge?
<mdash>
ldlework: more like "acting out my compulsions"
<alj[m]>
Anyone in here interested in porting Qubes to NixOS with me?
<mdash>
also I got a new work laptop and it seemed like an excellent opportunity to reorganize my config stuff
<thoughtpolice>
Although it's more of an abstract guideline, for generating complex outputs, it's almost always preferable to write an actual program that will generate the output, then write a Nix expression around that.
<mdash>
thoughtpolice: that was my plan, just some little python script
<thoughtpolice>
Normally this is just some data munging, i.e. you can use builtins.toJSON to export a JSON value of some attrset, write it to a file, and pipe it somewhere
<mdash>
yep
<mdash>
there's builtins.toXML but that would only help if I was aching to write the data munging part in XSLT...
WhittlesJr has joined #nixos
_kwstas has joined #nixos
_kwstas has quit [Remote host closed the connection]
_kwstas has joined #nixos
_kwstas has quit [Remote host closed the connection]
_kwstas has joined #nixos
_kwstas has quit [Client Quit]
<edef>
alj[m]: define "port"
_kwstas has joined #nixos
_kwstas has quit [Remote host closed the connection]
steell has quit [Ping timeout: 255 seconds]
dermetfan has joined #nixos
alex`` has quit [Quit: WeeChat 2.3]
waleee has joined #nixos
schjetne has joined #nixos
ninjin_ has quit [Remote host closed the connection]
ninjin_ has joined #nixos
alter-schjetne has quit [Ping timeout: 246 seconds]
<alj[m]>
Qubes runs only on fedora. The Qubes packages would have to be compiled for nixos and probably a lot of troubleshooting would have to be made.
jacereda has quit [Ping timeout: 268 seconds]
<alj[m]>
I don't imagine running Qubes on top of NixOS will be easily achievable.
<edef>
yeah
<edef>
qubus also does like, puppet or something for setting up machines
<alj[m]>
It uses saltstack for setting up machines
<qyliss^work>
I aim to have a Qubes-like system on top of NixOS, but I don't plan on using any Qubes tools, really
<alj[m]>
My vision is to define a whole Qubes system with nixos
<qyliss^work>
Nothing that Qubes has is really necessary to achieve that goal.
<qyliss^work>
Assuming you're fine with, eg, libvirt over Xen, which I am.
nixy has quit [Remote host closed the connection]
nixy has joined #nixos
<alj[m]>
Okay I'm Interested... qyliss^work how exactly are you gonna do it?
<qyliss^work>
With a lot of work :P
<qyliss^work>
This has been slowly burning away and I've been experimenting for... half a year? at this point. I don't have much useful, because I've done a lot of other Nixpkgs work as well, but I think it's definitely doable.
<alj[m]>
Are you planning on defining VMs for libvirt in nix expressions?
<qyliss^work>
There's no reason you can't have some nice Nix functions for defining VMs, disposable or not, and there's existing software (Xpra, for example) for compositing windows togother
<alj[m]>
Hmm... Do you have a gihub for this?
<alj[m]>
*github
<qyliss^work>
There's not enough yet for that. (And when there is it'll be hosted somewhere other than GitHub.)
<alj[m]>
Are you gonna use x11 or Wayland?
vmandela has joined #nixos
kreisys has joined #nixos
<qyliss^work>
My POC used X. I don't know enough about Wayland to understand the possibilities there. (Although I do use Wayland mostly nowadays.)
<qyliss^work>
Shall we continue this conversation in #nixos-chat?
<qyliss^work>
Or in PM?
<alj[m]>
Pm if you like
<qyliss^work>
I'll at least let you know if it ever gets anywhere
ij has joined #nixos
agander has joined #nixos
hedning has quit [Quit: hedning]
knupfer has joined #nixos
rauno has quit [Remote host closed the connection]
steell has joined #nixos
ij has quit [Ping timeout: 272 seconds]
<alj[m]>
I'll start working on it as well some time in the next few weeks. I'll let you know if I can make some progress as well :)
<nyanloutre[m]>
about this PR, I also suggested an improvement for the `fetch-kde-qt.sh` script. I managed to make it work with only downloading metadata.
reinhardt has quit [Quit: Leaving]
<nyanloutre[m]>
and about testing I didn't find anything else than the plasma test (which test only some kde applications like dolphin and konsole)
Ariakenom has quit [Read error: Connection reset by peer]
fusion809 has quit [Remote host closed the connection]
neminis has joined #nixos
infinii has quit [Ping timeout: 245 seconds]
ddellacosta has joined #nixos
<aanderse>
i have a nixos server running a web application that is configured to send mail via a relay mail server. sometimes the web app is trying to send to a bad address (a user account has been removed, etc...). the relay mail server is trying to send an email back to my nixos server saying "address doesn't exist" type thing.
<aanderse>
nixos is setup to use ssmtp to send the mail... so nothing is listening
<aanderse>
i don't care about the email that the relay mail server is sending back (safe to ignore)
<aanderse>
so how do i setup something on nixos to listen for incoming mail and eat it
<aanderse>
looking to avoid a full postfix, etc... , if possible
<symphorien>
you could setup the web app to use a valid from adress
<aanderse>
well i want to ignore the message
<aanderse>
not clutter up another inbox
yokyolu has joined #nixos
averell has quit [Quit: .]
<aanderse>
i found some links online about setting up either a xinetd or systemd service to listen on port 25 and eat... but i thought maybe there was a nicer way built into nixos
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @orivej to master « kmplot: init at 18.12.1 »: https://git.io/fhFBS
countingsort has quit [Quit: leaving]
<zeec123>
nyanloutre: thanks, I was missing the parantheses
jacereda has joined #nixos
<WhittlesJr>
So I spent a few days trying to package dokku(http://dokku.viewdocs.io/dokku/) and its dependencies, and finally gave up. (It's a massive PITA and would take tons more patching). What should I do with the work I've done? There are several dependencies that I believe I've packaged successfully, and dokku itself could potentially be made to work someday.
<qyliss^work>
Submit the working packages as PRs
<therealwaphire[m>
finally installed nixos with zfs on a hetzner without kvm
<therealwaphire[m>
something I noticed is that GRUB doesn't support `dnodesize=auto` on ZFS
rcshm_ has joined #nixos
<WhittlesJr>
qyliss^work: What about dokku itself though? What if someday someone more masochistic than I wants to take up the torch and finish it?
<qyliss^work>
You could post it on Discourse, or even here - search engines tend to index the logs
<qyliss^work>
I wouldn't submit it to nixpkgs though.
<WhittlesJr>
OK, I'll do that! Also, I've since found tsuru (https://docs.tsuru.io/stable/index.html) which does the same thing as dokku but was 1 billion percent easier to package and worked flawlessly out of the box. So NixOS will finally have a working PaaS implementation.
Tucky has quit [Quit: WeeChat 2.2]
vmandela has quit [Remote host closed the connection]
vmandela has joined #nixos
Phillemann has joined #nixos
<Phillemann>
I have a Python package that I'd like to package. Its setup.py creates an .so file and then there are .py files that load this file. I'd like to hard-code the .so file path into the .py file, but how do I retrieve it? The derivation hasn't been built yet.
simukis has joined #nixos
<musicmatze>
Hi. I'm getting the following error in one of my expressions I'm working on, can someone help? -> error: string '/nix/store/pc951k2jw7fh5yjr3p3496kddc6mhfg5-i3.conf' cannot refer to other paths, at...
justanotheruser has quit [Ping timeout: 250 seconds]
<simpson>
Phillemann: Interesting. Is this using the standard way of building extension modules, or do they need the extension module in order to build more stuff?
<Phillemann>
simpson: It's using distutils to create an extension and then access this extension in Python. If that answers your question.
ikitat has quit [Ping timeout: 255 seconds]
<simpson>
Phillemann: Yeah. Does it need the extension before the check phase? What have you tried so far?
<Phillemann>
There is no check phase.
<musicmatze>
Ah, okay, ... I had a strange layer of indirection in there
<reallymemorable>
is there a plan to upgrade Nix pkg manager for OSX Mojave?
<Phillemann>
simpson: I haven't tried anything because I'm not sure what to patch in inside the .py files.
steell has quit [Ping timeout: 246 seconds]
<Phillemann>
I mean, I could create a wrapper script in my application that somehow substitutes LD_LIBRARY_PATH.
<Phillemann>
I could also write two derivations I guess, since the first one really doesn't need any Python actually. And then use that inside the "outer" Python derivation? Yeah, that actually sounds well.
<simpson>
Phillemann: Oh! It's being opened with ctypes or similar, rather than being imported as an extension module? You could compute the name during patch phase from $out.
<clever>
Phillemann: thats how nearly all python things handle it, but i dont like that, because people keep missing it and random_python_library#42 then fails when imported
<Phillemann>
simpson: Exactly, the Python code uses ctypes.
<clever>
Phillemann: embeding the absolute path of the .so would be better, not sure why more arent doing it
<Phillemann>
clever: Maybe people have exactly my problem - how to "compute" (?) the .so path.
<clever>
Phillemann: at build-time, it will be somewhere under $out (an env var)
<Phillemann>
Hehe. I could literally "find" the thing. :D
<Yaniel>
anyone working on a blender 2.8 package yet?
<clever>
Phillemann: i usually just run ls on the result symlink, find what sub-path its at, and then hard-code the nix to use $out/lib/libfoo.so
__Sander__ has quit [Quit: Konversation terminated!]
<prooftechnique>
I'm managing an ini config file in my Nix config that references ${pkgs.cacert}/etc/ssl/certs/ca-bundle.crt, but when the file is assembled the reference is quoted, which the program doesn't like. Is there a way to unquote such references?
samrose_ has quit [Ping timeout: 244 seconds]
<clever>
ProofTechnique: can you pastebin your nix expression?
<clever>
Phillemann: and you cant use normal env vars like ${out} why?
<Phillemann>
clever: I guess I can, I just thought I didn't have to if the Makefiles looks standard enough.
<clever>
i basically ignore all other distros when packaging things now
<qyliss^work>
stdenv expects ./configure to take a --prefix
<clever>
i just write things for nix, since nix works on every distro
zeec123 has quit [Quit: Page closed]
<qyliss^work>
if there's no./configure / cmake / etc, usually you do installFlags = [ "prefix=$(out)" ]
<qyliss^work>
Or whatever is required for that Makefile
koba1917 has joined #nixos
winem_ has quit [Ping timeout: 246 seconds]
<Phillemann>
qyliss^work: Ah, that I have to manually do, okay.
<Phillemann>
No problem.
reallymemorable has quit [Quit: leaving]
ikitat has joined #nixos
<thoughtpolice>
Phillemann: You may also want to use makeFlags instead of installFlags, if the build phase itself (i.e. just 'make') will depend on the resulting PREFIX
<thoughtpolice>
that is, when you want 'make PREFIX=... && make install PREFIX=...', not just 'make && make install PREFIX=...'
<thoughtpolice>
Important distinction
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
mvnetbiz747971 has quit [Ping timeout: 245 seconds]
shlevy has joined #nixos
callahad has joined #nixos
<{^_^}>
[nixpkgs] @danbst merged pull request #56146 → postgresql: fix regression in NixOS after update to 11.2 → https://git.io/fhFnW
<{^_^}>
[nixpkgs] @danbst pushed to master « postgresql: fix regression in NixOS after update to 11.2 (#56146) »: https://git.io/fhF0o
arjen-jonathan has quit [Ping timeout: 268 seconds]
<evanm>
I am on nix 19.03 and I am trying to build a reflex project...getting an error dyld: Library not loaded: /usr/lib/system/libsystem_network.dylib
<infinisil>
(nix-info *without* -m is made to be pasted into irc directly btw, -m is to have nicer formatting in github issues)
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos
<evanm>
infisil: Apologies
reallymemorable has joined #nixos
<evanm>
And thank you to all of you who work on Nixos. It is a pretty incredible piece of open source work
_kwstas has joined #nixos
<illegalprime>
if a derivation fails to build how do I know what .nix file in nixpkgs it corresponds to? glibc is failing and i can't reproduce it in nix-shell or know what file to edit
_kwstas has quit [Remote host closed the connection]
<rcshm_>
hi, i was able to install and run apache kafka as a service. i did check that the processes for kafka and zookeeper are running. however when i do the console test, nothing is being printed out on the console. i try to check on the /tmp/kafka-logs but it is empty.
<clever>
illegalprime: you can directly run `nix-shell /nix/store/foo.drv` to open a given derivation in the shell, but that still wont help you know which one to edit
<rcshm_>
how should i debug what what wrong in this case? thanks for any help.
<infinisil>
evanm: I'd maybe try nixpkgs master, and the nix master (or something near master for those)
<infinisil>
evanm: And if you can't get it to work, opening an issue is always a good idea
<srhb>
evanm: Are you actually using pure nixpkgs or reflex-platform?
<illegalprime>
clever: so this is weird, when I try to build pkgs.pkgsCross.aarch64-multiplatform.glibc, nix-build trys to build all of its dependencies, which includes _another_, earlier glibc. the first glibc is the one that's failing. would that help me narrow down the file?
rauno has joined #nixos
<illegalprime>
forgot to note: both glibc's are aarch64 so the first one isn't a bootstrapping glibc (I don't think)
<evanm>
infinisil: thank you I will try to move to master and see what happens.
<clever>
illegalprime: trying the same build here...
<clever>
illegalprime: its fetching 90% of it from the cache, and only building a single glibc
<evanm>
srhb: I think you may be on to my problem
dbmikus has joined #nixos
shabius has quit [Quit: Leaving]
<evanm>
I am not trying to compile my own new project. It is an old reflex-platform project. Since the build script uses reflex-platofrm/work-on I may be pinned to some old version and not know it.
<clever>
evanm: yeah, i believe you need to use a new nixpkgs, to fix the new darwin problems
o1lo01ol1o has quit [Remote host closed the connection]
<adamt>
rcshm_: Are you writing anything to the topic you're reading from?
<rcshm_>
yes
sigmundv__ has quit [Ping timeout: 244 seconds]
<adamt>
also, we should probably bump the kafka pkg to the one release a few days ago
<rcshm_>
i just follow the test for producer and consumer scripts.
<adamt>
rcshm_: Do you have a link to the test?
<rcshm_>
i just put in apacheKafka
<evanm>
clever: thanks. I am on 19.03 what nix-channel should I be on?
<adamt>
rcshm_: be sure to keep the console consumer window open while writing to the topic, since the default (used to be?) for it to only show messages that was sent after it started
<rcshm_>
i am just following this adamt.
hedning has joined #nixos
<rcshm_>
how do i install the new kafka, adamt:?
<adamt>
rcshm_: Sorry, that part was just me thinking aloud and trying to prepare for the 19.03 release
<srhb>
rcshm_: I don't think you need to worry about upgrading just yet, it's likely some other issue. :)
<rcshm_>
thanks srhb:.
<adamt>
rcshm_: during step 3, are you writing anything into the terminal after running the command?
<rcshm_>
yes
<adamt>
and hitting enter? (sorry, have to ask)
<srhb>
evanm: Which version of reflex-platform are you on? (and fwiw, 19.03 isn't out yet, at most you're on unstable/master, which is sometimes called 19.03pre)
<rcshm_>
i have a simple topic created "test".
<rcshm_>
topic created successfully.
<illegalprime>
going to try it on nixos
<rcshm_>
no error. but nothing was printed out on the consumer window.
<{^_^}>
[nixpkgs] @oxij opened pull request #56158 → splice.nix: hybrid implementation of Gentoo-like use-flags → https://git.io/fhFEt
o1lo01ol1o has joined #nixos
shabius has joined #nixos
<clever>
illegalprime: pkgs.pkgsCross.aarch64-multiplatform.glibc builds perfectly on a recent nixos-unstable
zupo has joined #nixos
<evanm>
srhb: of course you're right ....I am on nixpkgs-19.03pre169894.44f78998bb
o1lo01ol1o has quit [Remote host closed the connection]
<srhb>
evanm: And reflex-platform?
o1lo01ol_ has joined #nixos
<illegalprime>
clever: oh dang maybe i rebased on an old branch, is nixos-unstable just nixpkgs's master?
<illegalprime>
or what should I rebase on to make a PR?
<clever>
illegalprime: yeah, nixos-unstable is just the latest rev of master to pass a set of tests
<rcshm_>
adamt: is there a way to check the log what is going on? i just don't know where the log is.
<illegalprime>
clever++
<evanm>
srhb: not 100% sure. I took this from my logs.....ca2dc8a7768abbcdf3edd3ede9c359144c84dd3f
<{^_^}>
clever's karma got increased to 95
<srhb>
rcshm_: journalctl -u apache-kafka -f
<srhb>
rcshm_: er, probably without -f to start with :)
<rcshm_>
let. me try that.
<adamt>
rcshm_: Sorry for the delay, i just had to test of the console producer and consumer worked (but my cluster is ancient (v0.10), and I'm a few weeks away from updating to 2.x, so my knowledge is kinda dated)
<adamt>
rcshm_: Yeah, I would just look in the output from journalctl. I'm not sure what to look for specifically.
<rcshm_>
i did run the command and i have got this Feb 21 16:35:55 nixos java[5868]: [2019-02-21 16:35:55,284] INFO Client environment:zookeeper.........
<rcshm_>
so it is running.
<adamt>
rcshm_: If you want, we can poke inside zookeeper to figure out whether anything was written to the topic
<rcshm_>
how do i do that?
<adamt>
rcshm_: Give me a few minutes, and I'll check where the partition offsets are written
<betaboon>
i have a long string with '' ... ''. how do i properly escape singlequotes so that i can write '' within that string ?
<adamt>
rcshm_: In the mean time, go install zk-shell, it's a pretty good zookeeper client for debugging things
<rcshm_>
thanks adamt:
<rcshm_>
let me install that.
<adamt>
(it's packaged in nixos, so it should be easy)
DavidEGrayson has joined #nixos
<srhb>
rcshm_: Random thought, did you permit a replication factor of 1? (assuming you only started a single instance)
<rcshm_>
i have it but i have got to learn how it works.
<rcshm_>
i checked it with your command journalctl -u apache-kafka and it is running.
<srhb>
rcshm_: Try adding extraProperties = "offsets.topic.replication.factor=1"; to the kafka attrset
<rcshm_>
let me try that.
Guanin has quit [Ping timeout: 258 seconds]
<adamt>
srhb: I think that'll just configure a default replication factor (unless the nixos module -- which I don't know -- does something weird, or things in kafka changed more than i hoped for)
<srhb>
adamt: I believe it's for kafka-internal topics as well
<srhb>
adamt: And I think it might refuse to do _anything_ if it can't reach that value.
<rcshm_>
services.apache-kafka.offsets does not exist
<rcshm_>
where should i put this?
<srhb>
rcshm_: Look above again, the attribute is called extraProperties :)
<rcshm_>
oh ok. my bad.
<srhb>
adamt: From the development server.properties file: The replication factor for the group metadata internal topics "__consumer_offsets" and "__transaction_state" [...]
<srhb>
adamt: __consumer_offsets sounds relevant
<adamt>
srhb: Yeah, and I see the default value is 3
<srhb>
Indeed. So my guess is that the offset cannot advance.
<adamt>
That's new, but then again, kafka 0.10 was a long time ago ^^
<srhb>
True :D
<srhb>
Did you migrate them to NixOS?
xkapastel has quit [Quit: Connection closed for inactivity]
<adamt>
srhb: Nah, have a test cluster installation planned for next week, and a bunch of hardware getting racked up
danbst has joined #nixos
<srhb>
nice!
vmandela has quit [Ping timeout: 240 seconds]
<adamt>
The ambitious plan is to connect a bunch of new servers on NixOS, running 2.x, but speaking the old wire protocol to the existing cluster. Then migrate all topics, relocate old hardware, and upgrade from Debian to NixOS, before letting the old dogs rejoin the cluster. :P
<srhb>
adamt: Seems sensible :)
<adamt>
And change wire protocol somewhere in between
<adamt>
srhb: I'm pretty sure it'll be easier than the ceph stuff the other guys are having "fun" with
<srhb>
adamt: lejonet and I (mostly lejonet) are helping getting kravs ceph stuff past the finish line for 19.03
<rcshm_>
admamt: srhb:, it seems to do the trick.
<srhb>
rcshm_: Cool!
<rcshm_>
now it shows Connection to node 1001 could not be established. Broker may not be available. (org.apache.kafka.clients.NetworkClient) and the messages show up.
<rcshm_>
after restart.
<adamt>
rcshm_: Good, sorry to have dragged you in a wrong direction
<rcshm_>
thanks.
<srhb>
rcshm_: The config file you used on MacOS set that value to 1, which is unsafe, but practical for testing purposes...
<rcshm_>
oh not at all.
<srhb>
rcshm_: erm, not sure what that means :)
<rcshm_>
adamt: this is a great help.
<lejonet>
srhb: did you have a look at my part of the branch, anything that seems odd or so? :P
<adamt>
I would have suggested looking in the "consumers" directory on zookeeper, to ensure that your console producer popped up when connecting it
<rcshm_>
at least the messages show up. srhb:
<srhb>
lejonet: I did not, just had dinner, can you resend the link? :)
<rcshm_>
:) let's celebrate for now and i continue.
<lejonet>
srhb: sure :)
<adamt>
and I'm still looking for were kafka hides the partition offsets that reveal whether data is actually written to the topic. :P
<adamt>
srhb: I can say this much. johanot is not having fun with whatever was deployed on our production clusters, due to state getting lost during a normal upgrade, and having a rough time getting the OSD's back up again. :P
<srhb>
adamt: Ouch ouch ouch...
<srhb>
adamt: Is the production cluster on NixOS?
<adamt>
srhb / lejonet: Thanks for assisting krav with the PR, btw. :-)
<adamt>
srhb: Yeah.
<srhb>
:(
<srhb>
I wonder what happened. Hope I can get a full story at some point
<adamt>
I'm not sure about the details, but I think johanot will act the hero tomorrow
<srhb>
Good!
<lejonet>
adamt: np, I've got interest in it too :)
<adamt>
srhb: I'm pretty sure it's not as epic as when we crashed kube-router, though. :P
<lejonet>
adamt: you mean by throwing the grenade? :P
<adamt>
lejonet: He was mumbling something about missing symlinks, so I hope not. :P
<lejonet>
Bring forth the holy hand grenade of FHS!
<reallymemorable>
what web browsers do you guys recommend besides firefox for nixos? having issues with a GUI I'm testing in firefox and google-chrome did not seem to work when I added it to pkgs.
<adamt>
reallymemorable: chromium?
<srhb>
reallymemorable: chromium is another alternative. google-chrome should work too, but
<reallymemorable>
[15095:15095:0221/131422.574550:ERROR:zygote_host_impl_linux.cc(89)] Running as root without --no-sandbox is not supported. See https://crbug.com/638180.
<reallymemorable>
(google-chrome:16116): Gtk-WARNING **: 13:15:38.068: cannot open display: :0
<reallymemorable>
thats what i get when i run from a user
<srhb>
reallymemorable: Are you logge dinto your graphical session as root?
<adamt>
lejonet: I'm not really sure about what this FHS stuff is, but I'll just give you a "lol" anyway to not look clueless :P
<srhb>
If so, don't do that either.
<reallymemorable>
what do you mean
<srhb>
reallymemorable: I'm guessing you're in KDE or Gnome or somesuch?
<reallymemorable>
yes
<reallymemorable>
KDE
argent0 has joined #nixos
<srhb>
reallymemorable: Did you log into KDE as root?
<reallymemorable>
yes
<srhb>
Don't do that.
<reallymemorable>
ok will switch users
<argent0>
hi, is there a way to unload nix (other than removing the loading code in bash_profile)?
<srhb>
Or rather: I would strongly advise you not to do that. You should only become root whne it's necessary to modify the system in ways your user cannot
<srhb>
reallymemorable: This is a security measure.
<reallymemorable>
ok
* adamt
waves goodbye
<srhb>
adamt: o/
<srhb>
argent0: Manually mangling your env vars to exclude it from PATH?
<srhb>
argent0: It sounds like a lot of trouble though.
<argent0>
srhb: maybe I should manualy load nix then. Thanks
<srhb>
argent0: Maybe have a different bash_profile that does not load nix, clear your env, then source that alternative bash_profile
rcshm has quit [Remote host closed the connection]
<ottidmes>
srhb: in my defense, google-chrome was my first Nix package ever, and I believe all the other distros followed that convention, hence the -stable suffix
dbmikus_ has joined #nixos
<srhb>
ottidmes: I'm sorry, I didn't mean that as any kind of serious criticism.
<ottidmes>
srhb: well, I guess in hindsight an additional symlink to google-chrome would have been nice (like how most libs have serveral symlinks), since you are unlikely to install two versions of the same browser
<srhb>
ottidmes: Maybe yes :)
Ariakenom has quit [Ping timeout: 268 seconds]
eadwu has quit [Ping timeout: 264 seconds]
random_yanek has quit [Ping timeout: 255 seconds]
ubert has quit [Quit: Leaving]
softinio has quit [Quit: Connection closed for inactivity]
reinhardt has joined #nixos
hoijui has joined #nixos
requisitebits has quit [Remote host closed the connection]
reinhardt has quit [Client Quit]
<judson_>
Is there an existing general tool for building Nix-side package mirrors a la rubygems, node etc?
<srhb>
judson_: Can you explain what those are?
<srhb>
I'm not sure what the comparison means :)
<srhb>
Better yet: What should the thing do?
<ottidmes>
judson_: maybe look at what cachix does?
<judson_>
i.e. I know there's bundix and pypi2nix, and the prefetch tools, but the problem of "prefetch this thing and template a Nix expression with attributes for a fetchFrom* function" seems general enough?
<judson_>
I'm just looking at a quick hack I wrote for fisher plugins, and wishing there were One True Way, you know?
callahad has quit [Quit: Ping timeout (120 seconds)]
<illegalprime>
clever: it worked! I think pkgs.pkgsCross.aarch64-multiplatform.glibc is just broken in master or it was only recently fixed in the last day
<srhb>
lejonet: Looks good. We can put this on kravs branch or wait until he's back :)
<judson_>
Wow, that does way more than I remembered. I think it might, yes.
<lejonet>
srhb: github very bluntly told me I'm not allowed to put it in his branch myself :( xD
<srhb>
I think I can.
eadwu has joined #nixos
random_yanek has joined #nixos
<lejonet>
"steal" the branch from my repo and try? :)
<ottidmes>
judson_: it should work with any fetcher, and almost all packages in Nixpkgs, at the moment I am working on nixpkg-update that will build upon nix-prefetch to allow automatically updating packages that nix-prefetch can prefetch
evanm has quit [Ping timeout: 256 seconds]
<danbst>
eyJhb: shamless plug - look at pkgs/servers/sql/postgresql/default.nix
<judson_>
Cool. I've got a function that needs to wrap fetchFromGitHub - it looks like I can use an arbitrary fetcher?
<lejonet>
srhb: I could do the insane thing and push ceph and ceph-client to my cluster and test at least those parts
xkapastel has joined #nixos
<srhb>
lejonet: What could go wrong!
ij has joined #nixos
<lejonet>
srhb: indeed, what could possibly go wrong? :D
<ottidmes>
judson_: any, even the builtins, you can even write your fetcher within the call to nix-prefetch itself on the fly
<lejonet>
I've got an overlay setup for my vm host anyway, so quite easy to push ceph-client to it that way :)
<srhb>
lejonet: If you're trusted, you can just nix copy it :-P
<judson_>
Right on. Okay, a shell `while read` and nix-prefetch probably is what I want. TYVM!
<lejonet>
srhb: define trusted ;)
<srhb>
lejonet: nix.trustedUsers
<lejonet>
srhb: I can easily make myself trusted xD
callahad has quit [Quit: Ping timeout (120 seconds)]
<eyJhb>
danbst: Not that many version, so I am going with charles-3.x.nix, and charles-4.x.nix
<srhb>
lejonet: A chance to play with nix copy if you haven't before.. :)
callahad has joined #nixos
<lejonet>
srhb: indeed, I will do that :D
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
srl295 has joined #nixos
<danbst>
eyJhb: it scales for your case as well =) but if there are too many differences between expressions, then indeed, better to split files
<lejonet>
srhb: oh noes, the borg failed to merge
sigmundv__ has joined #nixos
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #56127 → [18.09] elasticsearch: Add zlib to buildInputs for unfree → https://git.io/fhFqe
<srhb>
lejonet: Because we didn't rebase yet
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to release-18.09: https://git.io/fhFuT
freeman42x has joined #nixos
<srhb>
lejonet: (generic.nix)
<lejonet>
Ah, fair enough
<srhb>
I'll leave that and squashing for krav I think.
<lejonet>
Mhm, agreed
<{^_^}>
[nixpkgs] @eyJhb opened pull request #56159 → charles: split into versions (major releases) → https://git.io/fhFuk
<eyJhb>
danbst: not really any differences currently, but might be in the future.. It just doesn't seem like a reason to
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
ircuser5678 has joined #nixos
<ircuser5678>
Hi. I've made an upgrade from 17.09 to 18.09 and while the system works fine I have a small problem with dhcpcd. The message during bootup is "A startjob is running for dhcp client..." and it slows down the boot by about 20 seconds. The thing is, I'm using a static ip address setup on this machine. With 17.09 I didn't had this problem and I made the necessary changes regarding "network." changes from 18.03.
<{^_^}>
[nixpkgs] @peti pushed 0 commits to haskell-updates: https://git.io/fhFuc
<srhb>
ircuser5678: I think we might need to see some config, that issue doesn't ring a bell to me at least.
<srhb>
(I mean, you could just disable it, but I don't think it should be necessary in your setup as described)
rauno has quit [Remote host closed the connection]
<ircuser5678>
This is the network config of that machine
<srhb>
ircuser5678: I wonder if the predictableInterfaceNaming change is at cause.. let's see..
<ircuser5678>
No, at least not if I just comment it out.
<srhb>
Hm, nothing else springs to mind.
lejonet has quit [Ping timeout: 246 seconds]
<srhb>
ircuser5678: Oh, but if you comment it out the interface has a different name too, right?
<srhb>
But yeah, it's probably not related.
<ircuser5678>
Looking at it via journalctl -b stated that eth0 was coming up, so I'd say: no
<ircuser5678>
Why is dhcpcd started anyway if I disable it via dhcpcd.enable = false;?
Ariakenom has joined #nixos
<judson_>
Someone just gave a talk about kubenix, I think? Does anyone have the link?
<{^_^}>
[nixpkgs] @Twey opened pull request #56160 → tcptrace: init at 6.6.7 → https://git.io/fhFur
agander has quit [Ping timeout: 255 seconds]
hyperfekt has joined #nixos
<hyperfekt>
hey y'all. anyone know how to copy a functions formal arguments? i'm trying to wrap a function but apparently the call site uses builtins.functionArgs to pass the arguments only when required instead of by default
lejonet has joined #nixos
vidbina has joined #nixos
mbrock has joined #nixos
sondr3 has joined #nixos
<lejonet>
srhb: okay, so thats what happens when you forget that you actually had done networking changes when pushing stuff to my vm host... xD
ircuser5678 has left #nixos ["Closing Window"]
dbmikus_ has quit [Ping timeout: 246 seconds]
zupo has joined #nixos
Makaveli7 has quit [Ping timeout: 264 seconds]
<srhb>
lejonet: Uh-oh :P
<lejonet>
srhb: I'm still quite annoyed that nixops isn't smart enough to not nuke-from-orbit the interface that its connecting to the servers through... :P
<lejonet>
time to do it again, set the wrong mode on the bond :(
<ottidmes>
hyperfekt: builtins.functionArgs, or lib.functionArgs?
<lejonet>
There we go, now everything is as it should, NOW I can screw around with nix copy :O
JosW has joined #nixos
orivej has quit [Ping timeout: 268 seconds]
<clever>
lejonet: of note, root is trusted by default, so you can always just `nix copy --to ssh://root@laptop /nix/store/foo` and ignore fixing the cfg
<lejonet>
clever: already fixed the config, by twice nuking the server xD
Makaveli7 has joined #nixos
ikitat has joined #nixos
sigmundv__ has quit [Remote host closed the connection]
<lejonet>
srhb: so nows the question, I've sent the 13.2.4 closure to the machine, how do I get it to actually use it?
<{^_^}>
[nixpkgs] @jacereda opened pull request #56162 → purescript: fix build compiling with ghc 8.4.4 → https://git.io/fhFz1
eadwu has joined #nixos
justanotheruser has joined #nixos
ixxie has quit [Ping timeout: 268 seconds]
shibboleth has joined #nixos
vidbina has quit [Ping timeout: 255 seconds]
<srhb>
lejonet: Oh, I thought you just wanted to poke around with the client
<srhb>
lejonet: You can't easily replace your system setup with that closure.
<srhb>
lejonet: Sorry if I misled you :P
<lejonet>
srhb: oh, well, play around with it I can do on my desktop ^^ (VPN ftw!) :D
<srhb>
Ah, right :)
stigo has quit [Quit: WeeChat 2.2]
<lejonet>
srhb: don't worry, I just thought you had a l33t hax or something that you can incorporate it into the system setup or so :P
ij has joined #nixos
<lejonet>
I installed it in my user session, it actually reports its version correctly! and ofc can connect to the 12.2.0 cluster
adamt has joined #nixos
adamt has quit [Changing host]
adamt has joined #nixos
random_yanek has joined #nixos
stigo has joined #nixos
<srhb>
Sounds promising!
<{^_^}>
[nixpkgs] @hlolli opened pull request #56163 → csound-manual: init at 6.12.0 → https://git.io/fhFzF
o1lo01ol_ has quit [Remote host closed the connection]
<lejonet>
it would be nice if I could poke just the ceph-13 branch into the overlay and test the libceph stuff through qemu and libvirt too, but dunno how I'd take the all-packages thingy into a overlay
<adamt>
Still talking about ceph? Good. :P
<lejonet>
Shot, he's back, lets hide!
<{^_^}>
[nixpkgs] @svanderburg pushed to master « xcodeenv: do not rename bundleId to lowercase »: https://git.io/fhFzx
<srhb>
sondr3: Meh. Maybe there's some impure dependency..
<sondr3>
srhb: oh, I mean Spotify works, just not my Top recommendations lol
<reallymemorable>
thank you
cthachuk has quit [Read error: Connection reset by peer]
cthachuk has joined #nixos
johanot has joined #nixos
<srhb>
sondr3: oh :P
<srhb>
sondr3: Maybe you need to hear some new music. I recommend Nekrogoblikon.
<ottidmes>
reallymemorable: the only way to override the packages within nixpkgs without modifying nixpkgs directly used to be the overridePackages attribute in your nixpkgs config, but since then overlays have basically replaced it (although overridePackages still works), since overlays are more composable (i.e. its relatively easy to share an overlay with someone as opposed to overridePackages)
<sondr3>
srhb: what on earth
<adamt>
Last time I heard you talk about music you mentioned dwarf metal, so something that sounds like dead goblins isn't really a surprise :P
<srhb>
:3
<hyperfekt>
ottidmes: sorry, got a bit distracted. i suspect the lib version, forgot most builtins are duplicated in the library. does it make a difference for what i'm trying to do?
<adamt>
sondr3: My gut feeling is that if you prefer to listen to top 10 pop music, you probably shouldn't take advice from srhb on this topic. :P
<manveru>
sondr3: if you check the web player, it hasn't changed either
<manveru>
i have the same bloody recommendations for ages now :(
SyrupThinker has quit [Remote host closed the connection]
drakonis has joined #nixos
<srhb>
adamt: This is probably solid advice..
<{^_^}>
[nixpkgs] @svanderburg pushed to master « xcodeenv: fix generation of ad-hoc wireless installation page »: https://git.io/fhFgr
<sondr3>
manveru: oh, that's really annoying, but thankfully not just me having issues
<srhb>
sondr3: Where do you see this option? I can check mine..
<sondr3>
and there should be a heading called Top recommendations
<sondr3>
or that ^
<srhb>
Ah, never noticed that. I'll have to keep my eye on it and see if mine works..
<hyperfekt>
i find that spotify recommendations only become good if you actively use them and listen to a bunch of stuff on the platform. it's an investment, but one that really pays off
<sondr3>
hyperfekt: yeah, but right now only Discover Weekly updates and I want more things to discover :(
<sondr3>
without having to do much because I'm too lazy to actively discover it
<sondr3>
lol
<manveru>
speaking of broken software, zoom-us is broken too :(
<manveru>
also the `This application failed to start because it could not find or load the Qt platform plugin "xcb"` thing
<hyperfekt>
sondr3: try experimenting with radios. you could make radios based on playlists you like, or a playlist that is equivalent to your library, or based on past discover weeklies, etc
jacereda has joined #nixos
<sondr3>
hyperfekt: but that require's more effort than just clicking on album art
<sondr3>
but yeah, I could
<srhb>
manveru: Did you actually install it?
<srhb>
manveru: iirc it had some problems running in a shell
SyrupThinker has quit [Remote host closed the connection]
<hyperfekt>
sondr3: there are also some sites online that will generate long playlists for you based on others i think
<sondr3>
hyperfekt: yap, tried a bunch of them, I just really enjoy how easy it is with Spotify and it not updating is annoying
<sondr3>
even though Spotify is really, really hit or miss
SyrupThinker has joined #nixos
rotty has quit [Quit: WeeChat 2.4-dev]
<manveru>
srhb: yes, i have it installed
<srhb>
manveru: Meh :/ Works for me. But it is a rather misbehaved piece of software..
<averell>
an ofborg derivation linter would be really cool for those of us making 3 mistakes per line.
<manveru>
srhb: still the best thing for pairing atm...
<srhb>
manveru: Oh for sure.
Glider_IRC_ has joined #nixos
<srhb>
manveru: It's annoying on a package-level. It's pretty good at what it does. When it works.
<manveru>
and we use it all the time for work, just stopped working when i upgraded to unstable
SyrupThinker has quit [Client Quit]
agander has quit [Ping timeout: 246 seconds]
<srhb>
manveru: I feel like this is probably some kind of qt impurity..
<{^_^}>
[nixpkgs] @zimbatm opened pull request #56164 → shab: init at 1.0.0 → https://git.io/fhFgy
<srhb>
manveru: Is it possible that there's some remnant of your stable system in your profile?
bara has quit [Ping timeout: 240 seconds]
<adamt>
manveru: Any chance that you installed it with nix-env?
<LnL>
etu: but with ./nixos instead of the NIX_PATH lookup
<johanot>
srhb regarding our ceph env: all I can say is that /var/lib/ceph/osd-* were all empty, suddenly. which.. makes the osds quite unhappy and unwilling to start. still wondering how/why that state disappeared
<srhb>
johanot: Were they mounts?
<lejonet>
Was just gonna ask, did you forget to mount something
<srhb>
Yeah, feels that way.
<johanot>
tmpfs's
<srhb>
um!
<lejonet>
O.o
<srhb>
Oh, is that a bluestore thing?
<lejonet>
you use bluestore?
<srhb>
lol...
<srhb>
I'll just go to sleep ;)
<lejonet>
bluestore is... an odd bird
<johanot>
mmmh bluestore... heard that word before, makes me feel hungry, but I have no idea what it is?
<lejonet>
johanot: format of the OSD storage, either you do filestore (traditional FS mounted) or their new bluestore which is a weird thing that uses a raw block device like a savage
<srhb>
johanot: I bet ceph-volume lvm activate --all will make them come back
<etu>
adamt: I'm doing some changes to the iso, I'm not sure what that config option does
<etu>
oh, I see from LnL's example
vk3wtf has joined #nixos
<srhb>
johanot: I don't recall which of my module changes you use, but I think I also made systemd units for that..
<lejonet>
srhb is the root of all evil? :O
<srhb>
Probably called ceph-volume-something...
<srhb>
Problably >_>
<lejonet>
Sweet
<lejonet>
Then I know who to blame for those things then xD
<srhb>
:| I never released this to the world!
<srhb>
I just let johanot and adamt deal with it.
<adamt>
etu: If you want, we have a pretty.. uhm... involved setup i can share with you, where we build a live ISO based on one configuration, and containing another configuration that is then installed when run. :P
<adamt>
Excuse me, I don't deal with ceph, I make up excuses and go work on less scary stuff.
<etu>
adamt: nah, I'm just gonna add sound to the graphical iso, because it's a bit weird to ship a graphical iso live environment without sound
<lejonet>
If you don't startle it... :P
Dedalo has joined #nixos
<srhb>
Yeah, I actually like ceph too.
<srhb>
The module I wrote was just a bit of a hackjob.
Dedalo has quit [Client Quit]
<adamt>
etu: Then you certainly don't need anything as crazy as this. Good for you. :P
<lejonet>
I'm happy that I actually get to run a cluster now, after researching it for like 3 years :P
orivej has joined #nixos
<srhb>
lejonet: Reminds me, why are you not listed as ceph maintainer.
<lejonet>
(went through so many tests setups in different times of cephs life, I must say the build system now is... a looooot better than it has been)
Dedalo has joined #nixos
<srhb>
O:-)
<lejonet>
srhb: erhm... uhm... I just remembered something important I should do... xD
<etu>
srhb: He managed to opt out even though he did a lot of work on it!
<srhb>
I can fix that. :-P
<lejonet>
crap
* etu
as well
<lejonet>
well, I only wrote the inital ceph module, nothing fancy :P
<etu>
lejonet: that's way enough
<lejonet>
etu: true
selfsymmetric-pa has joined #nixos
mkoenig has quit [Ping timeout: 246 seconds]
<srhb>
I think you're the best candidate :)
<selfsymmetric-pa>
I'm having trouble with my NIX_PATH
* johanot
executes srhb's cmd-pastes blindly as root on prod...
<srhb>
selfsymmetric-pa: What is $NIX_PATH ?
_kwstas has joined #nixos
<srhb>
johanot: >_> maybe wait until I think a bit tomorrow :P
<selfsymmetric-pa>
Right now it's `/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs:nixos-config=/home/matt/log/nixos/configuration.nix:/nix/var/nix/profiles/per-user/root/channels`.
mkoenig has joined #nixos
<lejonet>
johanot: you should've waited till tomorrow to do that, so you can do it in the afternoon and then just leave ;)
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<selfsymmetric-pa>
The distinction between nixpkgs and nixos has me confused...
<srhb>
selfsymmetric-pa: Looks right to me. Does the channel actually have contents?
<selfsymmetric-pa>
Good question. How can I check?
<johanot>
srhb: lejonet I've already given up on those osd's.. wanna just pull them out and eat popcorn while stuff rebalances :P
<srhb>
selfsymmetric-pa: The default naming of the channel as "nixos" on NixOS is largely a historical mistake
wfranzini has joined #nixos
hoijui has quit [Remote host closed the connection]
<srhb>
selfsymmetric-pa: You can rename it nixpkgs if you want.
<selfsymmetric-pa>
Okay, why not.
<lejonet>
johanot: if you have enough replicas in the cluster, that is the best way to deal with it :P
<srhb>
selfsymmetric-pa: This entry should still catch it: /nix/var/nix/profiles/per-user/root/channels
<srhb>
selfsymmetric-pa: Because that should contain nixpkgs, which in turn contains nixos
<lejonet>
that is the nice part about the distributed part, if you don't need those particular OSD to fulfill your replica setting, just prune em from the cluster, nuke em and remake em :)
<lejonet>
let PAXOS do the rest
<selfsymmetric-pa>
Maybe I need to bootstrap it...
<srhb>
selfsymmetric-pa: yes, you do need sudo nix-channel --update
<srhb>
selfsymmetric-pa: uhhh, this IS NixOS, right?
<hyperfekt>
Though I feel like there must be a better way to wrap option with my own one.
<johanot>
srhb: ceph-volume lvm list gives me everything, but since /var/lib/ceph/osd-* is empty, the osds can neither find their keys nor their superblocks
<selfsymmetric-pa>
srhb: It is. Something look weird?
<lejonet>
well, the only part that I kind of need to fix is the creation of the ceph-volume@ services is my guess, everything else works in the test I wrote for it
<selfsymmetric-pa>
ar: Cześć!
<ar>
selfsymmetric-pa: bry
<selfsymmetric-pa>
Thank you again, srhb
<johanot>
ceph-volume lvm activate 1 <uuid> did half the job.. /var/lib/ceph/osd-1 got populated, but..
<lejonet>
srhb: ooooh, I just found a --no-systemd switch on prepare *runs away to implement that in the test and hope for the best*
<srhb>
johanot: Can you gist me your module tomorrow?
<selfsymmetric-pa>
Hm. What binary do I need for `PKG_CHECK_MODULES` macros to be recognized in configuration scripts?
<adamt>
johanot: Maybe it's reproducible on the same host by just rebooting it? Or was this one of the hosts with the bad sounding ceph errors on shutdown?
<johanot>
lejonet: meaning that reboot is disasterous? :D
<lejonet>
with bluestore, the ceph-volume lvm activate is the equivalent to mount /path/to/device /mount/point for filestore
<selfsymmetric-pa>
I thought it was `pkgconfig` but that doesn't work...
<lejonet>
johanot: if you don't have services for it, yes :P
<srhb>
lejonet: I'm fairly sure there's services for it.
<lejonet>
srhb: then they might not be enabled if so
<srhb>
Possibly...
<johanot>
i'll test it tomorrow. Zzzz. don't want to reboot now, the host will jump back to some disgusting generation far in the past
<lejonet>
xD
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<lejonet>
srhb: pushed the fix to the ceph-13-ceph-volume-fix branch and might spend some time tomorrow on drafting a plan to handle the ceph-volume@ thingymabob
selfsymmetric-pa has quit [Remote host closed the connection]
<ottidmes>
samrose_: people use the mozilla nixpkgs overlay to use nightly
Glider_IRC__ has joined #nixos
Dagger has quit [Excess Flood]
<samrose_>
ottidmes: I can't for the life of me get mozilla overlay working as part of my own custom packages and overlays
<samrose_>
I actually did get a fork of pkgs/development/compilers/rust to work with a simple crate I created and a specific version of nightly
ym555 has joined #nixos
Dagger has joined #nixos
jtojnar has joined #nixos
Glider_IRC_ has quit [Ping timeout: 245 seconds]
Dedalo has joined #nixos
knupfer has quit [Ping timeout: 250 seconds]
<samrose_>
ottidmes: do you have, or know of any good examples that use mozilla overlay for rust nightly with custom overlays and packages that are for building nixos ISO and channels (not just for my local machine)?
ng0 has joined #nixos
zupo has joined #nixos
<freusque>
hey does someone know how to tag some packages for sandbox build?
<freusque>
The only way I know is to run a nix-build with the option as a preliminatory step
<adamt>
Regarding rust, what does it require to make rustPlatform use a newer rust version than the one in pkgs? Is it as simple as just overlaying the rust version i want, and then it'll automagically work?
wfranzini has quit [Remote host closed the connection]
<ottidmes>
samrose_: I had a good example, but removed it, now I just use it for a simple shell.nix, sorry. Building channels? Seems like a hydra thing. Custom NixOS ISO I do use, if you want me give you what I have let me know
wfranzini has joined #nixos
jackdk has joined #nixos
<noonien>
i have a binary that calls another binary from a hardcoded path, /var/lib/some/path
<noonien>
it's a go binary, and the string is in a table, i don't think i can modify it, is there a workaround i can try?
selfsymmetric-pa has joined #nixos
<selfsymmetric-pa>
,locate bin autoreconf
<{^_^}>
Found in packages: autoconf, autoconf213, autoconf264
thommey has joined #nixos
<jackdk>
noonien: you don't have source for this binary?
<samrose_>
ottidmes: yes I have my own hydra instances and am running through these
<noonien>
jackdk: i do not
<noonien>
it's the binary for nordvpn
<samrose_>
ottidmes: I might get a clue from your custom nixos ISO work if you are using overlay there so I am happy to take a look and grateful for you to share
<samrose_>
adamt: are you just working with nix-shell for rust?
<jackdk>
noonien: possibly a bit beyond my nix-fu, but perhaps some of the techniques in https://nixos.wiki/wiki/FAQ#I.27ve_downloaded_a_binary.2C_but_I_can.27t_run_it.2C_what_can_I_do.3F can get you moving?
<adamt>
samrose_: Locally I'm using nix-shell, and just using the mozilla overlay to pick a new Rust version. But then I'm dragging my project into our own overlay of custom $stuff, and using the rust platform in nix for making a nix pkgs
Ariakenom has quit [Quit: Leaving]
sigmundv has quit [Remote host closed the connection]
<samrose_>
adamt: I am trying to understand the same thing you are now. I don't have a good answer yet
<noonien>
jackdk: thanks, but patchelf doesn't work
<noonien>
i mean, it does, because i need to change the path to ld.so
<noonien>
but it doesn't work for providing /var/lib/nordvpn/openvpn to the binary
<adamt>
samrose_: Ah, re-reading your question I see what you mean
selfsymmetric-pa has quit [Remote host closed the connection]
<samrose_>
adamt: although, I have been able to fork and use older versions of pkgs/development/compilers/rust and getting some progress going in that direction
<noonien>
and it seems it needs more than that
<adamt>
Oh well, I'll have to ponder over that another day.
<samrose_>
adamt: if I can get something working I will post it to nixos.wiki
ayerhart has joined #nixos
zupo_ has joined #nixos
selfsymmetric-pa has joined #nixos
<selfsymmetric-pa>
,locate bin pkgconfig
<adamt>
samrose_: Have you managed to do something clever with cargo deps? Like, getting them individually cached in the nix store, so that you don't have to rebuild it all everytime the pkg is updated?
<{^_^}>
Found in packages: pgf_graphics
<selfsymmetric-pa>
,locate bin pkg-config
<{^_^}>
Found in packages: pkgconfig, pkgconfigUpstream
<samrose_>
adamt: no :) however I had planned on getting whatever I am doing working with carnix, which does achieve that'
MinceR_ has joined #nixos
<selfsymmetric-pa>
Hmm. I have a pkg-config on my path but my autoconf still does not understand `PKG_CHECK_MODULES`.
<adamt>
At least the way i've gotten rustPlatform.buildRustPackage working requires rebuild+test of all cargo deps all the time, Zzzzz
<samrose_>
yes, it seems that way
<selfsymmetric-pa>
I have autoreconfHook installed...
<samrose_>
adamt: well, I mean it is that way
random_yanek has quit [Ping timeout: 272 seconds]
zupo has quit [Ping timeout: 245 seconds]
SyrupThinker has joined #nixos
<samrose_>
adamt: https://github.com/tazjin/carnix supposedly gets around that issue, although I have to first get rustPlatform.buildRustPackage to succeed in using my nightly version, and then eventually I will achieve buildingout the cargo file deps with carnix
selfsymmetric-pa has quit [Remote host closed the connection]
<adamt>
samrose_: Yeah, it makes sense to do it in that order I htink.
<tazjin>
and there is a link to the original in the readme ;-)
<samrose_>
tazjin: oy! well, will figure it out as I go
<tazjin>
the carnix sources sometimes disappear from
<tazjin>
pijul, so I put up the GitHub mirror
<adamt>
tazjin: Sorry for my non pijul compatible lingu, I honestly isn't even remotely in the project. :P
<adamt>
*interested in
<ottidmes>
samrose_: here are some bits I use (some outdated, I banished <nixpkgs> and <nixcfg> (i.e. the path to my NixOS config) from my config now)
<tazjin>
adamt: I don't know anything about pijul actually; I extracted the sources from crates.io before the carnix repo was public (I'm not the author of carnix)
<tazjin>
that would be pemeunier I think
<samrose_>
I stumbled across https://nest.pijul.com and some old discussions on the original carnix work there today
<{^_^}>
[nixpkgs] @rycee opened pull request #56169 → undefined-medium: init at 1.0 → https://git.io/fhFaG
arjen-jonathan has quit [Ping timeout: 258 seconds]
<immae>
Hi! Can I transform a path of the form "/tmp/foo" into a store path? (in the same way as /tmp/foo would "copy" it to the path, but here the value will be given as a string, from a command line argument of nixops)
tertl3 has quit [Remote host closed the connection]
<ikitat>
/run/current-system/nixpkgs exists on the client side? I could do something similar with overlays?
SyrupThinker has quit [Remote host closed the connection]
zupo has quit [Ping timeout: 250 seconds]
SyrupThinker has joined #nixos
justanotheruser has quit [Ping timeout: 264 seconds]
<clever>
ikitat: yeah, thats created by line 14
johanot has quit [Quit: WeeChat 2.2]
nightmared has quit [Quit: WeeChat 2.4]
<ikitat>
clever: this looks great. I'm a little confused about the context this runs, is this doesn't look like my nixops configuration
<clever>
ikitat: that is added to your nixos config, and line 14 will run as one of the last steps, when creating the storepath that /run/current-system points to
SyrupThinker has quit [Remote host closed the connection]
<ottidmes>
averell: not sure if thats it, but from the PR template "* Tested compilation of all pkgs that depend on this change using `nix-shell -p nox --run "nox-review wip"`", maybe nox/nox-review can do this, not sure, never used it
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fhFVL
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<selfsymmetric-pa>
Apparently `pkgs.emacsPackagesNg` does not have an `override` attribute when I try to override it in my `configuration.nix`. How can I customize it?
<clever>
selfsymmetric-pa: sounds like you may need a newer gcc? can google find out which one?
<selfsymmetric-pa>
Or a newer clang, perhaps...
<asymmetric>
what does nix-shell --pure actually do?
<NemesisD>
is it okay to delete some thing from /nix/store? i want to force it to rebuild
<veverak>
ok, hi!
* veverak
is new to nixos
<veverak>
and I tried to nixos-rebuild switch on my raspberry pi zero
<veverak>
with the armv6 image
<veverak>
fixed-output derivation produced path '/nix/store/6c7kyj10154aggq1igs6rcpyccazw4ar-gcc-7.3.0.tar.xz' with sha256 hash '0ll8vm23gs63bgr33pgy98m143g3bjgcdfangi859rs7n6d89ssk' instead of the expected hash '0p71bij6bfhzyrs8676a8jmpjsfz392s2rg862sdnsk30jpacb43'
<veverak>
how to approach this kind of problems?
eadwu has quit [Quit: WeeChat 2.4]
<clever>
veverak: ls -lh /nix/store/6c7kyj10154aggq1igs6rcpyccazw4ar-gcc-7.3.0.tar.xz
<veverak>
-r--r--r-- 1 root root 588M Jan 1 1970 /nix/store/6c7kyj10154aggq1igs6rcpyccazw4ar-gcc-7.3.0.tar.xz
<veverak>
?
<clever>
veverak: i'm guessing either your DL got corrupt, or upstream changed something
<clever>
veverak: does it fail the same way every time?
<veverak>
yup
<veverak>
but rebooting now
lunabo has quit [Quit: lunabo]
<NemesisD>
how do i force the rebuild of a particular expression?
<NemesisD>
without rebuilding the whole chain above it
<clever>
NemesisD: why do you want to force a rebuild?
<NemesisD>
clever: i rebooted and the build stopped working. tar ran out of disk space while archiving the docker image. i fed it a different TMPDIR (it was using /run/usr/1000?) and it worked. so now i'm trying to troubleshoot a way to not have to do that
<NemesisD>
the actual disk has plenty of space
<clever>
NemesisD: if you change the nix expression, just adding a space or comment inside a string, it will rebuild it
illegalprime has quit [Ping timeout: 246 seconds]
<ar1a>
can i print a list of what a derivation depends on
<ar1a>
wanting to know what packages are pulled for my cabal2nix
<clever>
ar1a: that works on both outputs, and .drv files
<NemesisD>
clever: how does the size of the tmpdir its using get controlled? df -h /run/user/1000 is only 2.4g
<clever>
NemesisD: that tmpdir is managed by systemd, and is a percentage of your ram, i try to always keep the build temp on /tmp, which is part of the / fs
<ar1a>
doesn't seem to work well with home-manager. just shows everything home-manager has installed :(