<samueldr>
uuuugh, on a `nixos-rebuild switch`, without upgrade, I'm getting `Failed to start local-fs.target: Unit ***.mount is masked.` where *** is one of my mount points, differing (unordered) every time
ericsagnes has quit [Ping timeout: 276 seconds]
hakujin has quit [Ping timeout: 248 seconds]
<colemickens>
alright, well, even with 'set -x' in vm-run-stage2, I'm not getting any additional output, so I'm suspicious of 'switch_root'
<colemickens>
plus my `echo` would've shown up anyway if it had started executing....
<clever>
colemickens: right before the switch_root, do "ls -ltrh /fs/nix/store"
<ghostyy>
hi i have a kind of unique and wierd problem with nixos, ui hope someone here knows how to solve it
<ghostyy>
my disk is failing and a bunch of files in my nix store have holes in them because of it
astiles has quit [Remote host closed the connection]
<clever>
ghostyy: `nix-store --verify --check-contents --repair` will test the drive and repair any damage, but you should probably replace the disk asap if its truely failing
<ghostyy>
so i want to somehow reinstall all of the stuff in there so that its the same versions and etc of everything
astiles has joined #nixos
<ghostyy>
lol yeah im on that too
<ghostyy>
thanks a lot, thats a huge help
<colemickens>
clever: it's... hanging?
<clever>
colemickens: ah, the problem is with 9plan then
<colemickens>
something related to 9p?
<colemickens>
:(
<clever>
its using 9plan to mount the host fs into the guest directly
<colemickens>
yeah, I'd seen a bit of it
<samueldr>
well, nixos-rebuild boot it is then, looks like some other folks here got the same issue, but nothing looks consistent and reproducible :/
<colemickens>
This might be my fault though.
<colemickens>
I'm running a different kernel version, I guess my host is on a different channel than I thought. Let me fix this.
sir_guy_carleton has quit [Remote host closed the connection]
<clever>
colemickens: the "daemon" side of the 9plan connection is within qemu, and the "client" side is in the guest kernel, both come from the same nixpkgs (your fork)
<colemickens>
well, wait though, I nixos-rebuild'd with my custom nixpkgs, so I would've expected the kernels to be the same then...
<Lisanna>
are there any stdenv shorthands for fixing shared library links on binaries with patchelf --set-rpath?
<colemickens>
yeah.
<colemickens>
ok
<colemickens>
hm.
<clever>
colemickens: when the client tries to read something, it will relay the request to the daemon (qemu), which then uses standard posix stuff (readdir, open, read, write) to perform the same request against the host
astiles has quit [Ping timeout: 248 seconds]
<clever>
so the host kernel shouldnt matter
<clever>
it may even work with a darwin or windows host
<colemickens>
I follow you, that's very slick.
<colemickens>
I'm trying to imagine how being on an Azure VM might affect plan9, but can't really think of anything, for the same sort of reason.
<clever>
Lisanna: if you run nix-build against one of these, it will generate a bash script, and if you run the bash script on an ELF, it will patch it
<colemickens>
or maybe it's broken on nixos-unstable-small currently, I suppose?
<clever>
Lisanna: that lets you quickly iterate thru a list of libraries, and also re-patch something that tries to auto-update and unpatch itself
<clever>
colemickens: hydra wouldnt let nixos-unstable update if such a thing was broken
<colemickens>
since I guess this would affect other images that people might care about more than azure? :)
<Lisanna>
clever okay, thanks for the snippet. I think I can build something more general-purpose out of this
<Lisanna>
I was thinking something like the opposite to how stdenv fixupPhase will strip the rpaths from binaries that are unnecessary
jackdk has joined #nixos
<Lisanna>
i.e., take an $out, iterate over all the elfs in it, and append all of the rpaths in buildInputs or whatever to each elf's existing rpath
<Lisanna>
then fixupPhase comes behind you and cleans up
telent has quit [Ping timeout: 264 seconds]
dan_b has quit [Ping timeout: 264 seconds]
<clever>
colemickens: mostly that hydra runs some basic nixos tests, that happen to involve 9plan
<clever>
Lisanna: ah, then you can pretty much just delete the bash heredoc on lines 3,4,6,7 and your done
<colemickens>
ok. well, I'm out of my depth, so happy to try any suggestions you might have
jlu has quit [Quit: WeeChat 2.0.1]
<clever>
colemickens: i cant think of anything obvious, id have to start patching the guest kernel or the host qemu and throwing printf's everywhere
<Lisanna>
clever sure, plus a for each loop + find
<Lisanna>
hey, I've noticed that stdenv seems to already do this to some extent...
<Lisanna>
like, libpthread.so.0 gets fixed up
<Lisanna>
and this is on a binary that has never seen nix before
<Lisanna>
I just blindly copy it from src to $out in a stdenv, and let fixupPhase run
<dhess>
Ugh, my Hydra has been stuck for the past 12+ hours. I've restarted the evaulator, rebooted the Hydra machine, rebooted the builders. Nothing has helped. Anyone have any ideas?
<Lisanna>
libpthread.so.0, librt.so.1, libdl.so.2, libc.so.6, and heck even the interpreter all point to /nix/store paths according to ldd
<clever>
dhess: how exactly is it stuck?
<dhess>
clever: jobs are just queueing up.
<dhess>
Bumping doesn't help.
<clever>
dhess: are jobs visible in the machine status?
<clever>
dhess: what features do the slaves have?
<Lisanna>
dhess can you do other builds on those machines outside of hydra?
<dhess>
No jobs in the machine status
<dhess>
the slaves have all the features
<dhess>
Lisanna: yes, that works fine.
<clever>
dhess: is the hydra publicly visible?
<dhess>
clever: no.
<dhess>
nothing in the logs that indicates issues, either.
<clever>
dhess: hydra will never log about a missing feature
<Lisanna>
oh, that thing I mentioned is because Nix's ldd is being extra helpful
<dhess>
this Hydra has been working more or less fine for months
<dhess>
clever: unless some new essential feature has been added in the past couple of days, that's not it
<{^_^}>
[nixpkgs] @jwiegley pushed to master « configuration-ghc-8.4.x.nix: Remove reference to non-existent bv_0_5 »: https://git.io/vplzc
<clever>
dhess: open the details for a build that isnt buinding, find its .drv url, and then run `nix-store -r --dry-run` on it, can you pastebin the output?
<colemickens>
which is clearly a nixos user, based on the paths mentioned.
<dhess>
clever: ok let me try that
<dhess>
clever: btw are you running 2017-11-21 or -dev ?
<dhess>
I am using nix = nixUnstable in the Hydra config
<clever>
ghostyy: id check to see if you can just delete some of those then, what does `nix-store --query --roots /nix/store/2cqprl9pz1x6zl1nmjwnciqpvh1a02hi-libX11-1.6.5` say?
<ghostyy>
oh a lot of those are snapshots... hmm
<Moredread>
is there a reason why I need to build rust from source currently on the 18.03 channel?
<colemickens>
clever: I'm happy with a bigger hammer too, if that means having the VM pull from cache.nixos.org instead of the 9p mount.
<clever>
ghostyy: ah, that could also explain things
<ghostyy>
that just prints /nix/var/nix/profiles/default-27-link
<clever>
ghostyy: nix fixed the version on the fs, but the snapshot is immutable, so it cant be fixed
iqubic has joined #nixos
<colemickens>
I'm guessing that's going to just mean me going in and making that change myself.
<clever>
ghostyy: try destroying those snapshots and see if the error clears up
<dhess>
/home/dhess/.nix-profile/bin/nix-store
<clever>
dhess: yeah, thats the problem, you have a 2nd copy of nix in nix-env, removing it should fix the issue
<clever>
nix should never be in nix-env when using nixos
<dhess>
ok but I believe that's only in my env
<dhess>
which user evaluates the .drv for Hydra?
<clever>
dhess: hydra-evaluator
<dhess>
I will run that nix-store command as that user
<dhess>
ok
<clever>
and the .drv is already evaluated
<ghostyy>
clever, yeah, all of the errors from the snapshots are now "gone" (in zpool status they show up as: <0x8c>:<0x1974>) but theres still some left over
<clever>
ghostyy: what about "zpool get freeing" ?
<ghostyy>
its set to 0
<ghostyy>
whatever that means
<clever>
that means its not actively trying to free blocks
<Lisanna>
oh man, there's a whole world here that I've been ignoring / ignorant of... setup hooks
<dhess>
clever: do you mean hydra-queue-runner? there is no 'hydra-evaluator' on my system
<clever>
sometimes after a large snapshot destroy, freeing has a value of 5gig, and your free space slowly climbs
<ghostyy>
ahh
<clever>
dhess: yeah, got them mixed up a bit, it also doesnt matter what user you run the command as
<ghostyy>
im mostly concerned about the corrupt files in my store that it claims are still there, im not sure if theyre actuially in use or what...
<dhess>
clever: well running as hydra-queue-runner returns these derivations will be built:
<clever>
dhess: that one better not be asking for any features!
<clever>
dhess: ok, what about `journalctl -f -u hydra-queue-runner` ?
<dhess>
it isn't, I picked the simplest one
<Lisanna>
haha, but so far it looks like absolutely nothing in nixpkgs is using autoPatchelfHook
<clever>
ghostyy: i think we should take this to #zfs
xcmw has joined #nixos
sigmundv__ has quit [Ping timeout: 255 seconds]
<dhess>
clever: ugh that was it. But that's weird, why on earth didn't a reboot fix that?
<dhess>
clever: thanks
<clever>
dhess: what was in that log?
<ghostyy>
clever, ah, sure
<dhess>
clever: nothing! That's how I knew it was the problem :)
<dhess>
not even a start-up error
<dhess>
and the log dated back to prior to the reboot as well
<clever>
dhess: what about systemctl status hydra-queue-runner?
<dhess>
very strange.
<dhess>
clever: well now it's full of build stuff as I just started it
<clever>
ah
<clever>
so the service simply didnt auto-start for an odd reason
<dhess>
anyway thanks, that was dumb of me. I have thought to check the status of that in the past, but like you, I also got confused and checked/restarted hydra-evaluator rather than hydra-queue-runner
<dhess>
It's just been working so well lately that I forgot about it. And it was weird that the reboot didn't start it
<dhess>
I'll look for errors later, maybe something broke in a recent nixpkgs.
<Lisanna>
oh damn, do you just need to add a setup-hook to nativeBuildInputs or whatever and then it just works?
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/vplaS
<{^_^}>
→ 564a6288 by @eqyiel: maintainers: update email address
<{^_^}>
→ 70670342 by @matthewbauer: Merge pull request #39584 from eqyiel/eqyiel-update-email-address
Supersonic has quit [Disconnected by services]
Supersonic112 has joined #nixos
Arcaelyx has joined #nixos
Supersonic112 is now known as Supersonic
<Lisanna>
aszlig hey, thank you for writing autoPatchelfHook !! this is going to save me a *lot* of time with all the proprietary software I have to deal with
<{^_^}>
→ 7f3832bf by @NeQuissimus: linux: 4.16.4 -> 4.16.5
<{^_^}>
→ 8efe41df by @NeQuissimus: linux: 4.14.36 -> 4.14.37
<Lisanna>
now I have to decide if it's going to be less work to try to resurrect the nix expressions for 2.25 from the grave or try to hunt down the source code for the proprietary software I'm using and write a builder for it
<MichaelRaskin>
Lisanna: if it is about using a stale proprietary blob, you could just use a pinned checkout of Nixpkgs. I hope it is not some network software anyway…
pxc has joined #nixos
<Lisanna>
MichaelRaskin pinning a checkout of nixpkgs might be a better option...
hamishmack has joined #nixos
<{^_^}>
[nixpkgs] @CMCDragonkai opened pull request #39586 → labelImg: init at 1.6.0 → https://git.io/vplwO
pxc has quit [Ping timeout: 240 seconds]
<mightybyte>
Anyone know how to change the size of the root partition of the Amazon EC2 AMI for NixOS?
<mightybyte>
It starts really small no matter how large a volume I allocate to it.
<Lisanna>
MichaelRaskin oh god, the whole world is rebuilding
<Lisanna>
but it'll work :) thanks for the suggestion, that's how I'm going to solve this problem from now on
<MichaelRaskin>
I thought you would only install the blob and its dependency from the checkout
<MichaelRaskin>
But maybe it depends on gtk
<ghostyy>
is there an easy way to move a whole nixos installation from one disk to another?
<runciter>
wpa_cli status says i'm disconnected from my network, but after `sudo ip link set dev <interface> up` my computer's actually connected over wifi. what's going on?
<Lisanna>
MichaelRaskin oh, I perhaps naïvely just used stdenv from the older nixpkgs, since I need an older glibc
<Lisanna>
it's stdenv + autoPatchelfHook
<MichaelRaskin>
Did you override a new checkout's stdenv with old checkout's one?
<Lisanna>
not overriding, just using
<Lisanna>
but it's from an older nixpkgs, so I don't have anything cached
<MichaelRaskin>
Because just using and old checkout shouldn't result in many rebuilds…
<Lisanna>
and apparently cache.nixos.org doesn't either
<MichaelRaskin>
That's really strange.
<Lisanna>
I just picked the commit immediately before the glibc 2.25 --> 2.26 change
tyrus82 has joined #nixos
<Lisanna>
should I have picked a stable release instead?
raynold has quit [Quit: Connection closed for inactivity]
<Lisanna>
stuff like perl is getting rebuilt
<drakonis>
just rebase to the latest
<drakonis>
lol
<Lisanna>
drakonis I can't, I need an older nixpkgs for an older glibc version
<MichaelRaskin>
I have no precise idea how you managed to get so many rebuilds, and no energy to debug
<drakonis>
hmm, what's the motivation?
<Lisanna>
lol
<Lisanna>
drakonis proprietary software
<drakonis>
why not create an derivation for an older glibc version and plug that into the old software?
<MichaelRaskin>
I guess that stuff should be isolated in a jail or something anyway just because stale proprietary blobs, sweet format parsing bugs!
<drakonis>
what proprietary software
<drakonis>
yes do it
<drakonis>
containers/jails
<Lisanna>
MichaelRaskin haha yeah
<drakonis>
if you call it jails you'll make the freebsd people get mad :v
<drakonis>
apparently its their claim to glory in 1999
<MichaelRaskin>
Well, nsjail has the jail in the name, and its pretty close functionality-wise
<Lisanna>
drakonis the glibc nix builder is very complicated and I don't want to copy all those files into my local project
<MichaelRaskin>
It is true that they had well-working jails before OpenVZ work even started
<MichaelRaskin>
(also true that jails are useful)
<drakonis>
yes, but that time has passed
<MichaelRaskin>
Well, they still do have working jails
<drakonis>
their stuff isn't generic enough anyways
<drakonis>
it works but you can't do anything more with it lol
<MichaelRaskin>
Yeah, also doesn't allow enough privilege escalations to be interesting
<drakonis>
ha
<MichaelRaskin>
(maybe, just maybe, limited scope and 20 years of use have something to do with that…)
<drakonis>
yes
<drakonis>
(do we take it to #nixos-chat ?)
<MichaelRaskin>
It doesn't look there is anything we are displacing here
<drakonis>
eh
<drakonis>
20 years of use? it barely gets changes
<MichaelRaskin>
Lisanna has a lot to build before asking why the result doesn't work!
jrolfs_ has quit [Ping timeout: 256 seconds]
<drakonis>
the limited scope is why it didn't gain any footing
jrolfs_ has joined #nixos
<Lisanna>
MichaelRaskin :)
<MichaelRaskin>
Well, it is used for partitioning servers
<MichaelRaskin>
It's just that they have to be BSD servers. Which are a niche thing nowadays that networking performance landscape has changed.
<drakonis>
linux networking performance has also vastly improved since
<drakonis>
as long you're on the latest kernel of course
<MichaelRaskin>
Which is the core part of the change in landscape
<drakonis>
ah yes
<drakonis>
also the need for containerization, be it with cgroups+namespaces or virtual machines or both
<MichaelRaskin>
Well, jails are not that bad just for that.
jrolfs_ has quit [Ping timeout: 248 seconds]
<MichaelRaskin>
Also, VMs are not a problem.
<drakonis>
its that you can't do things like virtualizing networks
<drakonis>
bhyve is terrible
<drakonis>
it is extremely limited compared to kvm, despite being available many years after kvm
hamishmack has quit [Quit: hamishmack]
hamishmack has joined #nixos
<drakonis>
hmm, jails didn't have network virtualization until 2013
<drakonis>
2009 actually
<drakonis>
can't believe it had this stuff before linux yet it didn't flaunt it enough
<mightybyte>
Is it possible to change the location of the nix store?
<drakonis>
yes but you'd have to rebuild everything
<mightybyte>
Could you copy everything there first?
<Lisanna>
...how was the first C compiler compiled in nixpkgs?
<drakonis>
what do you mean the first C compiler?
<Lisanna>
I should say, how *is*?
<Lisanna>
how do you build your C compiler without having a C compiler
<Lisanna>
what hackery has had to happen to make this work in nixpkgs
<MichaelRaskin>
There is a static bootstrap compiler
<Lisanna>
MichaelRaskin ah, okay
tyrus82 has quit [Quit: tyrus82]
<drakonis>
how do you think the first c compiler was born in the first place?
<drakonis>
it was written in assembly
<Lisanna>
28 builds to go, but they're all stuck on gcc-6.4.0, and it's running ./configure, which isn't parallelizable ):
<drakonis>
ask dennis ritchie
<Lisanna>
drakonis I know that :p
<MichaelRaskin>
We… don't reproduce that part in Nixpkgs
<Lisanna>
nixpkgs has that problem perpetually though, since it tries to provide *everything*
<Lisanna>
whereas the gcc team when they build gcc just keep using the old gcc from the last time they built it, in an infinite chain
<MichaelRaskin>
Well, there is now a Nix expression for getting the fresh static bootstrap
hamishmack has quit [Quit: hamishmack]
<Lisanna>
arggghhh I have hundreds of cores but they're all useless right now because goddamn autotools
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<MichaelRaskin>
Enjoy your pain
<Lisanna>
why isn't it pulling from cache.nixos.org
<Lisanna>
is there a one-liner that I can run that tests to make sure nothing weird is afoot
<Lisanna>
I mean, something weird is obviously afoot
<MichaelRaskin>
Well, I would start by pretending that the new checkout doesn't exist
<drakonis>
arrrr
<drakonis>
autotools sucks, meson forever
<MichaelRaskin>
And build the package by using only the old checkout
<MichaelRaskin>
That has better chances to use cache
<Lisanna>
the new checkout is the entire rest of my system though
<MichaelRaskin>
Once you have the package built, you can try to write a wrapper that calls the old checkout correctly that can be just imported
<MichaelRaskin>
Then you just casually add (import /path/to/wrapper.nix) into systemPackages or whatever
<MichaelRaskin>
The point is not to have any arguments and to call the old checkout by absolute path, so that there si no difference if the new checkout is also used outside
adamt has quit [Ping timeout: 256 seconds]
<Lisanna>
I mean, that's what I'm doing in my nix code
<{^_^}>
→ c53c7a02 by @SergeL: libuv: 1.20.0 -> 1.20.2
<drakonis>
prob a secret
<{^_^}>
→ 21195458 by @SergeL: nodejs: init at v10.0.0
<{^_^}>
→ e9841a62 by @adisbladis: Merge pull request #39564 from holidaycheck/nodejs-10
<Lisanna>
drakonis lol
<drakonis>
lol
colemickens has joined #nixos
<Lisanna>
I wouldn't really call it a secret
<drakonis>
what do you do?
alex`` has joined #nixos
<Lisanna>
HPC stuff
<drakonis>
oic
<Lisanna>
i.e. stuff 99% of developers don't care about :p
<jcrben>
for my non-root user to browse the web, I had to disable the firewall - first tried allowedTCPPorts = [ 80 443] but that didn't work
<jcrben>
is this explained somewhere? would like to know what the recommendation is - I wouldn't mind having the firewall up
<Lisanna>
jcrben generally firewalls restrict the incoming connection port, not the outgoing port, so adding 80 or 443 to allowedTCPPorts wouldn't help
<drakonis>
HPC stuff is cool though
<drakonis>
hold on
<drakonis>
brb
carolingio has quit [Ping timeout: 240 seconds]
drakonis has quit [Remote host closed the connection]
<Lisanna>
jcrben in general you shouldn't need to set that
<jcrben>
any idea what might be different in my case?
<Lisanna>
I'd try to find out why your packet responses are going out on a different interface than they're coming in on
phreedom has quit [Read error: Connection reset by peer]
phreedom has joined #nixos
<jcrben>
ok, thanks for the tip
unlmtd1 has quit [Client Quit]
<Lisanna>
"If using asymmetric routing or other complicated routing, set this option to loose mode or disable it and setup your own counter-measures. This option can be either true (or "strict"), "loose" (only drop the packet if the source address is not reachable via any interface) or false. Defaults to the value of kernelHasRPFilter. (needs kernel 3.3+) "
<jcrben>
kinda surprised that something weird like this could sneak into a declarative system (fresh install of 18.03)
etu has quit [Ping timeout: 248 seconds]
reinzelmann has quit [Ping timeout: 240 seconds]
ma27 has joined #nixos
jgt has quit [Ping timeout: 248 seconds]
robstr has joined #nixos
<Lisanna>
jcrben it probably has more to do with your physical network config or something outside of your machine than anything in nixos itself, since you seem to be the only one experiencing this
<Lisanna>
I mean, the default is to have it enabled
<Lisanna>
so 99% of NixOS users will have it enabled
tanonym has joined #nixos
<Lisanna>
ugh, I'm going to have to make a custom stdenv just to get this darn thing to work
<jcrben>
ok, that's probably true. I think I might've manually fiddled with stuff using ip earlier too. dunno that I'm going to dive deep with wireshark or strace anytime soon tho
<tanonym>
Upgraded my system to 18.03 and DE to Enlightenment 0.22.1, but the permissions error is back for cpufreq. Message wants me to run a chown and chmod command to fix the issue. Since this is handled by a wrapper in NixOS, not sure how to locate the misconfigured wrapper or how to fix it.
<unlmtd>
im trying to understand `nixpkgs.pkgs` would be a way to override nixpkgs, so I could use with the `fetchNixpkgs` function, right?
ma27 has quit [Ping timeout: 276 seconds]
reinzelmann has joined #nixos
hamishmack has quit [Quit: hamishmack]
<Lisanna>
unlmtd it's a way to overide the version of nixpkgs that is used to build your NixOS system
<unlmtd>
thats exactly what I want
<unlmtd>
Im trying something with it now
<unlmtd>
Lisanna: thanks
<Lisanna>
...maybe I should be using fetchNixpkgs
<tanonym>
nixpkgs.weechat would override the version that nixos-18.03 would normally install, for example?
* tanonym
studies Section 7 of Nixpkgs Manual. Hmm. Mostly Greek to me still. :D
etu has joined #nixos
<Lisanna>
tanonym no
<Lisanna>
tanonym nixpkgs.pkg is a special NixOS option
<Lisanna>
generally you shouldn't mess with it
<Lisanna>
there are other ways to override specific packages
<tanonym>
Yeah, the manual lists a bunch of them.
MP2E has quit [Remote host closed the connection]
<tanonym>
You wouldn't happen to know what's missing in my install: each time I try to do something like nix-env --help I get an error less: invalid option -R. Man nix-env just gives me a no man page message.
<tanonym>
I think my HD is dying. Takes multiple attempts to get it to boot into the system. Usually hangs at stage one trying to mount the zfs.
<Lisanna>
kuri0 I would say probably? the answer is probably somewhere in os-specific/linux/kernel/hardened-config.nix
<tanonym>
So, is nix-env -iA the preferred method to install? Seems to be much faster than nix-env -i now.
reinzelmann has joined #nixos
fragamus has quit [Ping timeout: 240 seconds]
<Lisanna>
omg I hate this piece of software
<Lisanna>
needs old glibc, which means *all* of its dependencies also need old glibc
<Lisanna>
...but it also needs NEW zeromq
<etu>
:DD
<Lisanna>
newer than nixpkgs
<Lisanna>
uggghhhhhhhhh
ThatDocsLady has joined #nixos
<etu>
Lisanna: Sounds like nix is one of few package systems that could handle that without using containers.
ericsagnes has quit [Ping timeout: 240 seconds]
<tanonym>
Lisanna would you happen to know how to fix broken setuid-wrappers? The one for the cpufreq in Enlightenment 0.22.1 isn't set right because I keep getting a message telling me to run a chown and chmod on the cpufreq module.
<Lisanna>
this is a constant battle between trying to decide whether it would be easier to repackage a binary distribution of a piece of internal software or go wage a full-scale war against the terrible official build system for the software's source code
<Lisanna>
tanonym nope, sorry ):
<tanonym>
Well, was worth a shot.
<tanonym>
Can't find much of anything that helps. Just expressions but no instructions about how to make use of them.
<Lisanna>
I wonder if I can override the callPackage that was used to callPackage a package
<Lisanna>
let myPkg = callPackage ./blah.nix; in myPkg.overrideCallPackage myCallPackage
<Lisanna>
something like that
<Lisanna>
otherwise this is gonna get ugly
<Lisanna>
and I'll have to get out my hammer
<tanonym>
Well, guess I'll just use MATE or KDE as my DE instead until I can figure out how the hell to fix setuid-wrappers and make them behave. It would be nice to know how to fix those when they have issues.
<tanonym>
No luck finding anyone who really groks them yet here or much of anything in the github...
* tanonym
thwacks head into wall.
hamishmack has joined #nixos
<Lisanna>
tanonym it's a small community
coot has joined #nixos
<tanonym>
I know. Been trying to find anything in mailinglists or documentation to work out what I'm supposed to do when the wrapper doesn't set the permissions and ownership right.
knupfer has joined #nixos
<tanonym>
I did find an expression for setuid but not quite clear how I'd make use of it...
<tanonym>
I wonder if the permissions in /run/bin are not set right for my user?
tanonym has quit [Quit: WeeChat 2.0]
<Lisanna>
plus I'm pretty sure all the work on this is virtually 100% volunteer
xAFFE has joined #nixos
<kuri0>
Does NixOS developers accept performance patches , compiler optimizations , etc ? Or is that not the goal ?
day|flip has quit [Remote host closed the connection]
markus1189 has quit [Ping timeout: 240 seconds]
markus1189 has joined #nixos
<Lisanna>
omfg
<Lisanna>
I repackaged the darn thing with old glibc
<Lisanna>
but, if there's something that lots of people like, and it's in the form of a patch, then I don't see any reason why it shouldn't be included as an alternate version of a particular package
<{^_^}>
→ 489b0a5f by @woffs: shotcut: 17.11 -> 18.03.06
<{^_^}>
→ c12caf34 by @woffs: shotcut: add myself as maintainer
<{^_^}>
→ cd12c338 by @Mic92: Merge pull request #39569 from woffs/shotcut-18.03
regulus_ has joined #nixos
<Myrl-saki>
So uhh
<Myrl-saki>
I'm stupid. I didn't copy the whole outputs.
<Lisanna>
Myrl-saki glad to hear that nix isn't fundamentally broken ^^
<steveeJ>
Lisanna: what if you invoke the binary with the dynamic linker manually?
<Myrl-saki>
Lisanna: That was so fucking embarrassing. :(
stew has joined #nixos
Ariakenom has joined #nixos
<Myrl-saki>
Lisanna: I'm fundamentally broken after me bitching and realizing what happened.
<Lisanna>
Myrl-saki it's fine <3
<Lisanna>
I bitch in here all day about dumb shit I do
hiroshi has quit [Ping timeout: 268 seconds]
<Lisanna>
steveeJ interesting suggestion... the garbage collector blew it away though so I can't test... working on building the darn thing from source right now
<Lisanna>
steveeJ I'll try that if building from source doesn't work out
<Lisanna>
etu I take it PHP build day was a success?
<etu>
Lisanna: Yeah :)
<Lisanna>
etu glad at least someone is having success building stuff tonight
<etu>
For me it's morning :)
<Lisanna>
I'm reading through the build system code for my thing and I'm sweating more and more with each new revelation
<{^_^}>
[nixpkgs] @Mic92 opened pull request #39592 → masterpdfeditor: use autoPatchelfHook → https://git.io/vplSK
<Lisanna>
imagine all the effort of the nixpkgs build-support, but implemented with ad-hoc handwritten makefiles
<Lisanna>
and shell scripts that live on dedicated servers
pxc has quit [Ping timeout: 260 seconds]
<steveeJ>
how can I use python wheel? it seems to be broken because the wheel store is read-only
blankhart has quit [Ping timeout: 240 seconds]
jacob_ has joined #nixos
hiroshi has joined #nixos
<robstr>
Morning, right now I'm using `nixops` to deploy my application. Since uploading the images is realtive slow I'm looking for alternatives. Is this a usecase for hydra (building my application and upload to aws) ?
<{^_^}>
→ 5a07abc2 by @dudebout: pill 20: fix a few typos
kreetx has joined #nixos
<{^_^}>
→ b706fc95 by @dudebout: pill 20: make it buildable
<{^_^}>
→ e9f5967d by @Mic92: Merge pull request #67 from dudebout/pill-20-buildable
<{^_^}>
[nix-pills] @Mic92 closed pull request #66 → Pill 20: Use `command -v` instead of `which` → https://git.io/vxAwd
stew has quit [Quit: Lost terminal]
<ertes-w>
with function from nixpkgs.lib do i need again to set buildEnv priorities?
lukego has joined #nixos
iMatejC has quit [Remote host closed the connection]
tyrus82 has quit [Quit: tyrus82]
<lukego>
I'm looking for a simple way to import a binary file into the Nix store without its store path being impacted by its file name. Something that will work in older versions of Nix. Any ideas?
ma27 has joined #nixos
iMatejC has joined #nixos
<lukego>
Basically I want to be able to import the same file many times, with different filenames (mktemp), and to always map onto the same store path if the contents are the same.
vaninwagen has joined #nixos
<tanonym>
Got the ZFS setup for nixos done. I think I need to add a line to exclude the pool/nix from getting snapshotted.
<Myrl-saki>
Okay
<lukego>
I *nearly* solved it with 'writeScript "foo" (readFile ./x)' but readFile didn't like binary data.
<Myrl-saki>
Stupid question
<Myrl-saki>
How do I override the stdenv for that package only?
<lukego>
The function 'builtins.path' seems to do the job but doesn't seem to be present in my Nix version (maybe that's new?)
<Myrl-saki>
Like, not including its children.
<ertes-w>
there are lib.hiPrio and lib.lowPrio, but i remember there was one to set an explicit priority
<tanonym>
Looked at the file and location and there's nothing boolean at the inidcated location.
ericsagnes has quit [Ping timeout: 240 seconds]
ylwghst has joined #nixos
<lukego>
(Found a lousy workaround. I'm generating the temp filename based on sha256 hash of contents rather than random entropy. It'll do for now.)
<Dezgeg>
maybe generate a temp directory instead and put it under a constant name under there?
dvim has quit [Quit: WeeChat 2.1]
tanonym has quit [Quit: WeeChat 2.0]
<Izorkin>
How to deploy with nixops custom physical server? Need first install to physical server nixos?
dvim has joined #nixos
<lukego>
Dezgeg: Yes, that is probably the best solution. I'm resisting that because it requires me to learn a bit more about the programming language that is generating the temp files :) like how to delete directories that might not be empty.
<{^_^}>
[nixpkgs] @peti pushed to master « multi-ghc-travis: update to latest git version »: https://git.io/vplhs
simukis has joined #nixos
Ariakenom has quit [Ping timeout: 240 seconds]
<Dezgeg>
heh
<srhb>
Izorkin: That's essentially the easiest way, yes.
<srhb>
Izorkin: (In fact I know of no other)
asuryawanshi has quit [Remote host closed the connection]
<srhb>
Anything else would probably involve nixops evolving into a pxe server of sorts :-)
asuryawanshi has joined #nixos
knupfer has joined #nixos
iyzsong has joined #nixos
<Izorkin>
srhb: how to add to nixops ssh pub key from physical server?
tanonym has joined #nixos
<srhb>
Izorkin: I think the easiest way is to have an initial key and use NIXOPS_SSHOPTS with whatever you need to connect the first time and then let NixOps do the provisioning of the root key itself.
<tanonym>
Would a ZFS install of NixOS need a separate boot partition?
<Mic92_>
tanonym: if you go the grub route, then you need a small one for grub. If you go for EFI with systemd-boot, you will need a slightly large one for the kernel.
NightTrain has joined #nixos
<Mic92_>
I prefer the latter.
<Mic92_>
also because grub cannot handle zfs's native encryption
<tanonym>
I don't think this machine supports EFI.
<tanonym>
Needs an MBR file partition table.
<tanonym>
GPT bricks it.
<srhb>
Mic92_: Hmm, how does it not handle zfs encryption? I thought I was using that. Maybe I have something weird going on. :P
<avn>
tanonym: Also note, that grub unable to read kernels from /nix/store on zfs (if hardlinks used)
<{^_^}>
→ c8b9ac9b by @endgame: haskellPackages.base-compat-batteries: fix bad dependency
<{^_^}>
→ ab5dcddd by @peti: Merge pull request #39456 from endgame/base-compat-batteries-deps
reinzelmann has joined #nixos
<srhb>
Mic92_: Ah, no, apparently I'm just mixing my machines up. Easy to forget when config just magically keeps working for years... :-) (My old work PC I handed in was indeed native zfs encryption BUT systemd-boot)
<tanonym>
I am assuming that EFI won't work if my system uses a MBR file partition table?
<Mic92_>
tanonym: no
<tanonym>
Guess I'm stuck using grub then.
<tanonym>
So 200M should be adequate for a boot partition.
leat has quit [Ping timeout: 255 seconds]
<Myrl-saki>
What's the difference between buildInputs and nativeBuildINputs?
<tanonym>
Also once I have my ZFS set up do I need to type anything to exclude pool/nix from autosnapshotting or is that after I do the install?
<srhb>
I don't think there's any autosnapshotting by default.
<tanonym>
There's a line you can add to configuration.nix
<Mic92_>
tanonym: when snapshotting is enabled set com.sun:auto-snapshot=true for every dataset to snapshot
<tanonym>
So that would be pool/root, pool/home but set to false for pool/nix or else garbage-collect will get borked
<tanonym>
Ah, so after the line is added to configuration.nix, I have to manually set auto-snapshot=true for the ones I want to do that.
<srhb>
Myrl-saki: Okay, I always have to look this up and still feel sort of shaky on it: nativeBuildInputs could better be called depsBuildHost, meaning, these are the dependencies required to build the derivation on the builder architecture (regardless of the target architecture) whereas buildInputs is more like depsHostTarget (build dependencies of the target architecture)
<tanonym>
Going to make sure I can do this install first before I even think about trying to set up disk encryption... :D
<tanonym>
Do I need to worry about swap? I have 4GB RAM on this machine...
<srhb>
tanonym: I'd say yes.
<srhb>
If you ever end up building something non-trivial 4GB is very little.
<tanonym>
How much swap? 8GB or more?
<srhb>
If you use a zvol for it you should be easily able to change it, I think?
<srhb>
I don't know if there's a gotcha there.
<tanonym>
srhb, you mean a zpool/swap allocation?
<srhb>
Yeah
Ariakenom has joined #nixos
<srhb>
Then you can swapoff, zfs set voldsize=whateverGiB zpool/swap and swapon
<srhb>
volsize*
hakujin has joined #nixos
<tanonym>
Huh, that is very very handy.
<srhb>
Er, that needs a mkswap in there.
<srhb>
But like I said I don't know if there's a gotcha with running swap on a zvol.
<tanonym>
How would I mkswap on a ZFS volume? That usually creates a linux_swap FS on a partition.
freeman42x]NixOS has joined #nixos
rihards has joined #nixos
<srhb>
I don't think there's anything special with it? mkswap /dev/zpool/swap ?
<srhb>
I might be out of my zfs depth though :-) Perhaps someone else can chime in.
<tanonym>
Looks like you were on the right track, srhb
<tanonym>
That pretty much describes what you were suggesting :)
<srhb>
Ah cool. :)
hakujin has quit [Ping timeout: 276 seconds]
<tanonym>
Looks a bit involved to switch the size, but would do the trick if I need a larger swap to build something.
ericsagnes has joined #nixos
<srhb>
Is it more complicated than I described?
<srhb>
As long as you ensure you don't need the space while doing it :)
<rizary>
i have use node2nix to develop my backend project. It seems that the default nix return derivation { tarbal = ...; package = ...; shell = ...;}, how can i import it so that i can use like `cp -rf ${backend}/lib/node_modules/backend/* $out/prod/`
<{^_^}>
[nixpkgs] @domenkozar pushed commit from @rvl to master « nixos/buildkite-agent: fix variable expansion in hook scripts »: https://git.io/vp8fY
<Myrl-saki>
srhb: Too bad, you scared me away from it. :x
seville has joined #nixos
<srhb>
:'(
<srhb>
Mic92_: Rust seems really good if those kind of metrics are needed. Would be great if we could achieve something similar, but I have no idea what it would take. It's probably horrendously difficult. :P
<Myrl-saki>
srhb: About how many files would it be dangerous to do this?
<srhb>
Myrl-saki: I honestly don't know.
<{^_^}>
[nixpkgs] @domenkozar pushed commit from @rvl to release-18.03 « nixos/buildkite-agent: fix variable expansion in hook scripts »: https://git.io/vp8fW
<srhb>
Myrl-saki: You could probably autogenerate a bunch of small different derivations and start sticking them in the store. See what happens.
<srhb>
(And by bunch I mean hundreds of thousands, probably...)
<Mic92_>
srhb: rewriting in different languages rarely pays off
<Myrl-saki>
srhb: So low thousands should be fine, I believe?
<srhb>
Mic92_: Oh I mean improving the actual build producs of ghc. I don't care much for programming in rust itself. :)
<srhb>
Myrl-saki: Sure. We already easily handle low thousands of store entries.
<srhb>
Even high thousands.
<Mic92_>
srhb: rust is probably way worth then the average c/c++ projecct.
<tanonym>
xmonad is supposedly very similar to i3 but based on Haskell. You also need to have the Haskell compiler installed.
<srhb>
Mic92_: I believe that! :)
<Mic92_>
since nobody uses dynamic linking
<lukego>
This might sound far-fetched but... Is there any way that unix timers (setitimer) would beahve strange inside a nix-build sandbox?
tyrus82_ has joined #nixos
Fare has joined #nixos
<domenkozar>
lukego: inside vm or just sandboxed build?
<srhb>
Mic92_: Rust seems to explode the nix expressions though. Is there a way to share dependencies better between projects, or is that just the way the ecosystem works?
<tanonym>
I would like to get the us international keyboard layout, but us_intl doesn't seem to work in configuration.nix
<srhb>
Seems like what happens with the go vendored stuff.
<lukego>
Just a sandbox build. and it would have to be something that changed in a recent nix.
Guanin has joined #nixos
<lukego>
Just being paranoid... Looking into why a profiler doesn't show any samples when I run it under a nix sandbox.
<Mic92_>
srhb: yes. this is a problem. Ideally we could have a single package set. Currently carnix generates per project and these feature flags in cargo blows up the size.
<Mic92_>
they can be shared.
<srhb>
Oh, if that's just the only problem it doesn't sound too bad :)
<srhb>
As long as it's not a fundamental limitation.
tyrus82 has quit [Read error: Connection reset by peer]
<lukego>
Oh I reproduced it outside the sandbox now. Sorry, no nix issue :)
<domenkozar>
:)
<Myrl-saki>
srhb: nix-make actually seems perfect. What's wrong with it?
<Myrl-saki>
I mean
<domenkozar>
I think we should put together a sandboxing reference with what is different than one might expect in linux
<Myrl-saki>
Anyways, I spoke too soon. There seems to be duplication of .h files, I guess that can be fixed with nix-optimise.
<srhb>
Myrl-saki: :P
<domenkozar>
Mic92_: 2 months of contributions
pxc has joined #nixos
tyrus82_ has quit [Read error: Connection reset by peer]
tyrus82 has joined #nixos
<tanonym>
Can I change compression types on per-pool basis? The instructions I'm following have me setting compression to lz4 when creating the pool.
<srhb>
tanonym: iirc you can even change it on a per volume basis.
tyrus82 has quit [Remote host closed the connection]
<domenkozar>
Mic92_: srhb: I'll give him/her commit access, but I'll ask to add name+surname to maintainer list
tyrus82 has joined #nixos
<tanonym>
Okay, that's good because they suggest using a zle compression for the swap volume, or else change the ashift to 9 from 12 and then you can use lz4.
<domenkozar>
we used to have policy that people shouldn't be anonymous and while it wasn't fully enforced, I think we shouldn't loosen it
<srhb>
domenkozar: I have no objection with that, the maintainer list is great.
Lisanna has quit [Remote host closed the connection]
s4sha has joined #nixos
<s4sha>
hi
<srhb>
s4sha: Hello!
NightTrain has quit [Quit: Lost terminal]
<s4sha>
I opened https://github.com/NixOS/nixpkgs/issues/39479 2 days ago and I've been struggling with environments since ^^ Still reading the Nix Pill #7 but I know I'm gonna have a lot of questions so I joined here in advance
* tanonym
crosses fingers and hopes the install doesn't blow up in my face.
<tanonym>
Well, looks like it's working.
<srhb>
s4sha: Great place to be for advice, welcome. :)
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
leotaku_ has quit [Ping timeout: 268 seconds]
<s4sha>
srhb: thanks. I already have one question ! I read everywhere that keys in a derivation became environment variables passed during the build
<srhb>
s4sha: Right.
peacememories has joined #nixos
<s4sha>
now I have erlang complaining about locale, but when I try and set LC_TYPE, LANG and such in the set I pass to stden.mkDerivation, I get errors during the build
<s4sha>
warning me that I can't set those
<s4sha>
why is that ? Do I misunderstand the error messages ?
peacememories has quit [Client Quit]
<srhb>
s4sha: Maybe you could show the exact error messages, I haven't seen that before. Sounds like an attempt to prevent impurity by relying on the host LANG etc.
adamt has joined #nixos
<s4sha>
yeah, I was meaning to do just that but then I ran nix-build -K -A pleroma to get said errors and it started copying an incredible number of derivations from cache.nixos.org
<s4sha>
this is quite annoying ^^
adamt is now known as Guest58197
<srhb>
Presumably those are the dependencies you need.
coot has quit [Quit: coot]
leat has joined #nixos
<s4sha>
I hope so, but I thought my system was up to date, and I already tried to build like 2h ago, how could there be so many new packages to download ? it's pulling xf86vidmodproto ?!
<srhb>
s4sha: Did you update your channel?
<srhb>
If you didn't garbage collect, you must have changed something else. Or one of your derivations is pulling in an unpinned nixpkgs.
<s4sha>
yes, it's updated to 18.03 since yesterday evening, but you said something very important I think : I did garbage collect, maybe this morning, I would have thought it was before I tried to build but I could be wrong
<s4sha>
if a build fails, then none of its dependencies are kept ?
<Mic92_>
How do I get qt3d out of qt for buildinputs?
<srhb>
s4sha: Yes, they're all kept.
<srhb>
s4sha: Nothing disappears automatically from the store.
<{^_^}>
[nixpkgs] @zimbatm pushed to master « terraform: update all plugins »: https://git.io/vp8kM
<s4sha>
but not marked as used and hence can be garbage collected ?
<srhb>
s4sha: If there's no gc root holding onto them, then yes, they can be garbage collected.
sigmundv__ has quit [Ping timeout: 256 seconds]
<s4sha>
a slow connection is the best master : it forgives none of your mistakes
<srhb>
s4sha: Haha, that is a good point.
<srhb>
s4sha: fwiw I don't garbage collect ever, more or less. :P
<s4sha>
well that's an important lesson learnt, thank you
<srhb>
s4sha: Unless I do a huge base bump of my nixpkgs so I need everything rebuilt again.
<srhb>
Local cache is great, learn to love it. :)
* srhb
has a 79 GiB /nix/store currently
<Mic92_>
srhb: you are cheating with your zfs compression
<srhb>
That's true. ;)
<s4sha>
I created my VM disk only 20G big
<srhb>
s4sha: I'd definitely grow that.
<s4sha>
yeah ! it's nearly done ! my derivation will soon fail properly !
<srhb>
:)
<srhb>
s4sha: fwiw I don't see any error if I do: nix-build -E 'with import <nixpkgs> {}; stdenv.mkDerivation { name = "test"; var = "hi"; LANG = "test"; LC_CTYPE = "test"; src = ""; installPhase = "mkdir -p $out && touch $out/test"; }'
leat has quit [Ping timeout: 240 seconds]
<srhb>
So it does not appear to be a mkDerivation limitation
<s4sha>
I can't build your derivation because «variable $src or $srcs should point to the source» ?! but there are no sources, that's why you set it empty, right ?
reinzelmann has quit [Quit: Leaving]
azahi has joined #nixos
<s4sha>
so, I finally go the errors I mentioned above. It goes : /nix/store/xn5gv3lpfy91yvfy9b0i7klfcxh9xskz-bash-4.4-p19/bin/bash: warning: setlocale: LC_ALL: cannot change locale (en_US.UTF-8)
<srhb>
s4sha: Oh, so the error is coming from setlocale.
<s4sha>
hmmm so it's not related to the way nix handles environment variables at all ?
<srhb>
No. :)
ylwghst has quit [Ping timeout: 240 seconds]
<srhb>
And I don't think the warning is critical unless the build refuses to continue.
yegortimoshenko has joined #nixos
<s4sha>
no, it's not what breaks the build but I'm very doubtful I can actually fix what's wrong without even being able to set variables the way I need to
Guest58197 has quit [Changing host]
Guest58197 has joined #nixos
Guest58197 is now known as adamt
<s4sha>
it's written in erlang and using mix, so it's failing at trying to mkdir somewhere in the fake $HOME because I don't even know how to make it put it stuff where it should : )
<s4sha>
and I have no idea where that should be in a build environment and not in a user's home
<srhb>
There are some mix packages in-tree already, perhaps you can look at them for inspiration.
<sphalerite>
ertes-w: not really, ncurses is the implementation of all the terminfo stuff
ma27 has quit [Quit: WeeChat 2.0]
ma27 has joined #nixos
darlan has joined #nixos
darlan has quit [Client Quit]
JosW has quit [Quit: Konversation terminated!]
dbmikus has joined #nixos
<ertes-w>
true
<{^_^}>
[nixpkgs] @xeji opened pull request #39602 → maintainer-list.nix: add real name for xeji → https://git.io/vp8q0
alexteves_ has joined #nixos
azahi has quit [Changing host]
azahi has joined #nixos
ma27 has quit [Quit: WeeChat 2.0]
ma27 has joined #nixos
orivej has joined #nixos
rosa has quit [Quit: rosa]
<{^_^}>
[nixpkgs] @domenkozar merged pull request #39602 → maintainer-list.nix: add real name for xeji → https://git.io/vp8q0
<{^_^}>
[nixpkgs] @domenkozar pushed commit from @xeji to master « maintainer-list.nix: add real name for xeji »: https://git.io/vp8mt
ma27 has quit [Quit: WeeChat 2.0]
ma27 has joined #nixos
asuryawanshi has quit [Ping timeout: 248 seconds]
szicari has joined #nixos
<{^_^}>
[nixpkgs] @Scriptkiddi opened pull request #39603 → gams: init at 25.02 → https://git.io/vp8YE
<robstr>
Hey, How can I clonse private repositories from github on hydra ? Can someone help here
Ariakenom has quit [Ping timeout: 240 seconds]
orivej has quit [Ping timeout: 256 seconds]
coot has joined #nixos
Ariakenom has joined #nixos
joko has quit [Ping timeout: 248 seconds]
<aminechikhaoui>
robstr: use ssh ?
joko has joined #nixos
<robstr>
aminechikhaoui: I'm using nixops to deploy a hydra server and my repository is private. Should I create a new keypair, copy the private on the hydra server ? Is there a better way ?
<etu>
robstr: I'd say create a new one and add it as a deploykey that is readonly on that repo
iqubic has quit [Remote host closed the connection]
ma27 has quit [Ping timeout: 255 seconds]
blankhart has joined #nixos
ma27 has joined #nixos
ThatPako has joined #nixos
Fare has joined #nixos
<ThatPako>
Any idea why a python project throws `No module named gi` even though I added pygobject3 and gnome3.gtk to my propagatedBuildInputs, and wrapGAppsHook to my nativeBuildInputs?
tertle||eltret has quit [Quit: Connection closed for inactivity]
<goibhniu>
ThatPako: that sounds familiar, I'd check other packages that use pygobject3 to see if there are some tricks you can use.
<ThatPako>
goibhniu: I have another one sitting right here on my drive, which works. and there is no difference, except for the source and a yml dependency
<jtojnar>
ThatPako: buildPythonApplication is weird
<ThatPako>
oof
furon has joined #nixos
<goibhniu>
there might be an environment variable you can set that helps, or something like that ... I'd check a few
<jtojnar>
ThatPako: sometimes using pythonPath instead of propagatedBuildInputs works
<ThatPako>
jtojnar: hm... now I'm getting `Namespace Gtk not available`
<jtojnar>
ThatPako: add gobjectIntrospection
<jtojnar>
ThatPako: its setup hook will populate GI_TYPELIB_PATH
<ThatPako>
reee, now it complains about the Rsvg namespace
<jtojnar>
ThatPako: that is in librsvg
<ThatPako>
Seems like adding librsvg to my pythonPath won't work :/
tanonym has quit [Quit: Leaving]
<etu>
robstr: No idea. I've never interacted with Hydra myself.
<robstr>
etu: ty
<ThatPako>
jtojnar: is it possible to enable introspection for librsvg?
rosa has joined #nixos
ma27 has quit [Ping timeout: 240 seconds]
ma27 has joined #nixos
drakonis_ has quit [Quit: Leaving]
rosa has quit [Client Quit]
<{^_^}>
[nixpkgs] @svanderburg pushed to staging « nodejs: use xcodebuild patch for Node.js 10.x so that modules with native dependencies on darwin will build properly »: https://git.io/vp8cD
<{^_^}>
[nixpkgs] @ookhoi opened pull request #39605 → netcat-openbsd-1.187-1-1.190-1 - set version from 1.187 to 1.190: → https://git.io/vp8cj
bennofs has joined #nixos
astiles has joined #nixos
rosa has joined #nixos
grafoo has joined #nixos
hhomar has joined #nixos
<grafoo>
has anyone used libvirt/virt-manager on a "foreign distribution"?
astiles has quit [Remote host closed the connection]
astiles has joined #nixos
Ariakenom has quit [Ping timeout: 256 seconds]
<LnL>
robstr: if you use inputs it will use the ssh keys from the hydra (or hydra-queue-runner) user
mkoenig has joined #nixos
<Taneb>
Is there a tool that can take a list of paths to the nix store (or something similar) and give me a graph of how they depend on each other?
<LnL>
nix-store -q --tree <path>
<LnL>
or --graph, that creates a graphviz file you can render
mkoenig_ has quit [Ping timeout: 264 seconds]
<jtojnar>
ThatPako: yes, that needs to be in buildInputs
<jtojnar>
only python packages should go to pythonPath
<jtojnar>
grafoo: hmm, we do not really have the global data (services, polkit rules) handled well on NixOS, not sure how to do that on normal dstributions
pxc has quit [Ping timeout: 240 seconds]
<LnL>
you can't use fetchgit for private stuff since it's a fixed output drv and it won't have access to keys etc
rosa has quit [Quit: rosa]
<LnL>
jobset inputs are updated by the hydra service, this happens outside of nix builds
rosa has joined #nixos
ThatPako has quit [Quit: Lost terminal]
<robstr>
LnL: I see. Ok but what is the problem right now ? I need to validate github.com ?
<LnL>
robstr: the hydra users' ~/.known_hosts doesn't contain the github host key so it's failing
<LnL>
either configure programs.ssh.knownHosts with the result of ssh-keyscan github.com
ma27 has quit [Ping timeout: 276 seconds]
<LnL>
or login as the hydra user and ssh to github, but you'd have to do that again if you ever reinstall/move the server
<LnL>
robstr: this will show you an example of how to configure the option: nixos-option programs.ssh.knownHosts
ma27 has joined #nixos
mkoenig has quit [Remote host closed the connection]
<yegortimoshenko>
jtojnar: i've switched to gnome on wayland :-)
<yegortimoshenko>
i'll upstream a lot more stuff when i have some time
mkoenig has joined #nixos
logzet has joined #nixos
<robstr>
LnL: great, but still looks like my setup is incorrect, maybe because nixops is deploying the keys directly ... I will destroy it and recreate it
<jtojnar>
cool yegortimoshenko
ma27 has quit [Ping timeout: 240 seconds]
<jtojnar>
do you have a branch?
<M-bbigras>
I cloned nixpkgs and updated a package. I know I can do `nixos-rebuild switch -I nixpkgs=/path/to/my/nixpkgs` to install it but it seems to reinstall everything or something like that. Is there a way to only install the package I modified and its dependencies?
ma27 has joined #nixos
<clever>
M-bbigras: first checkout a nixpkgs revision that is the same as your nixpkgs channel
<clever>
[root@amd-nixos:~]# nix-instantiate '<nixpkgs>' --eval -A lib.nixpkgsVersion
<{^_^}>
→ fc7ae1fe by @Synthetica9: atom: 1.26.0 -> 1.26.1
<{^_^}>
→ d89128bb by @Mic92: Merge pull request #39600 from Synthetica9/atom-1-26-1
katona has joined #nixos
<katona>
I would like to package an app that loads its data relative to the executable (in src), so I figured I will just put a symlink to the executable in bin, but where should I put the rest? On archlinux it is inside opt
<sphalerite>
(emacs, nix-buffer) So I've got a dir-locals.nix that allows me, with nix-buffer, to call `cargo run` using M-!. It doesn't work with cargo mode yet though. What might be missing?
fragamus has joined #nixos
<katona>
jtojnar: thank you.
lonokhov has quit [Quit: WeeChat 2.0]
<dx_>
need some tip: My 'Dell Inspiron Mini 10' when i close the LID goes to suspend-mode. When i open the LID, it wakes-up because I can ssh into it. But the console is black, but i can type keys i see because i used tmux. I'm not using X11. However, all is fine when i type 'systemctl suspend' it wakes up fine
<dx_>
sphalerite: i tried having or not having 'acpid' .. still same behaviour
<nliadm>
if I'm writing an overlay, should I be just using super.callPackage straight? or should I be using something like `callPackage = super.callPackageWith (self)`
<sphalerite>
dx_: is there anything in the journal on suspend that looks different from when you use systemctl vs closing the lid?
<sphalerite>
nliadm: just use super.callPackage
leat has joined #nixos
<sphalerite>
nliadm: the fixpoint magic makes the callPackage use args from the composition of all the overlays
<nliadm>
ah, so if I override a package in my overlay, and also callPackage something that uses that package, it'll get resolved correctly
<nliadm>
ok
<nliadm>
that's what I'm worried about
<nliadm>
I don't have to specifically pass my version in
jrolfs__ has joined #nixos
vaninwagen has joined #nixos
<dx_>
sphalerite: `journalctl -u suspend -f` shows nothing when i `systemctl suspend`
<sphalerite>
dx_: yeah you'll want to show the logs for everything, not just suspend
<dx_>
sphalerite: nothing also shows in `journalctl -u acpid -f`
<sphalerite>
just look at `journalctl -e` after each suspend type and compare them
rosa has quit [Quit: rosa]
rosa has joined #nixos
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Lisanna has joined #nixos
jrolfs__ has quit [Quit: WeeChat 2.0]
dx_ has quit [Ping timeout: 256 seconds]
chisui has joined #nixos
<jtojnar>
yegortimoshenko: how are you running it?
astiles has joined #nixos
<jtojnar>
I cannot even run the greeter in Wayland
<yegortimoshenko>
jtojnar: just `XDG_SESSION_TYPE=wayland dbus-run-session gnome-session` as user and it runs
civodul has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
plakband has quit [Client Quit]
vaninwagen has quit [Ping timeout: 248 seconds]
ryantrinkle has quit [Ping timeout: 264 seconds]
<GlennS>
Can anyone explain the relationship between generations of NixOS in the EFI boot menu and generations in nix-env? I'm trying to find the right way to delete some specific NixOS generations to free up space in /boot
<GlennS>
but I'm not sure if I just go deleting things in /boot whether that will make a big mess
zybell has joined #nixos
judson has quit [Remote host closed the connection]
pxc has joined #nixos
simukis has quit [Read error: Connection reset by peer]
leat has quit [Ping timeout: 240 seconds]
rosa has quit [Quit: rosa]
rosa has joined #nixos
<robstr>
I'm trying to get hydra running from my nixops deployment with this configuration https://gist.github.com/rsoeldner/234325108947b8e9cc7eba319edb37a6 but after adding a simple build (fetch from github) and _evaluate this project_ nothing happens, Evaluation pending since is growing ... How can i debug this ? which step did i miss ?
<{^_^}>
→ a5f6cdfd by @Profpatsch: lib/debug: add traceValFn, traceValSeqFn, traceValSeqNFn
<{^_^}>
→ a455637d by @Profpatsch: lib/debug: remove the deprecated strict function
<{^_^}>
→ a7fdd10b by @Profpatsch: lib/debug: deprecate & modernize showVal
Maxdamantus has quit [Ping timeout: 240 seconds]
Maxdamantus has joined #nixos
<Profpatsch>
heh
pxc has quit [Ping timeout: 264 seconds]
rosa has quit [Client Quit]
Myrl-saki has quit [Read error: Connection reset by peer]
peacememories has quit [Read error: Connection reset by peer]
peacememories has joined #nixos
rosa has joined #nixos
humanoyd has quit [Ping timeout: 240 seconds]
humanoyd has joined #nixos
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @oxij to master « nixos: alsa: better docstring »: https://git.io/vp8w2
Myrl-saki has joined #nixos
grafoo has quit [Quit: WeeChat 1.4]
<GlennS>
Found the answer to my question:
rosa has quit [Client Quit]
<GlennS>
the NixOS generations are profiles found in /nix/var/nix/profiles/system, so you can just run nix-env --some-stuff -p /nix/var/nix/profiles/system
<GlennS>
and then when you next do `nixos-rebuild switch` it works
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @oxij to master « nixos: xserver: add related packages »: https://git.io/vp8w9
rosa has joined #nixos
coot has joined #nixos
<jxf>
Hey friends. When I `nix-env -i ruby`, I'm able to use Ruby, and I see Ruby in `~/.nix-profile/bin`. However, after I install a couple of gems, their binaries show up in `~/.gem/ruby/2.5.0/bin`, which isn't on my $PATH, so I can't use any of the gem-based binaries.
<jxf>
Is this an issue with the Nix packaging, or should I be manually adding this directory to my $PATH?
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Mic92_>
If you want a more robust solution, you can also take a look at bundix
rosa has quit [Quit: rosa]
<Mic92_>
Because then it will be automatically rebuild against system libraries, while .gem might become stall. This problem is not just specific to NixOS.
<GlennS>
So, I'm not quite clear on the answer to this, someone else may give you more useful information. I can tell you that nixos-infect *does* work.
<kreetx>
GlennS: yup. I'm just worried that if I run into any problems I'm on my own :)
rosa has joined #nixos
<GlennS>
One option might be to try it on a VM first.
<kreetx>
not complaining -- just got deploy working on local virtualbox and it was a bit of research..
<GlennS>
but yeah, that is a perfectly reasonable position :)
<kreetx>
I guess I could just try on some of the hosting providers, I need a machine anyway and eventually must get there
Myrl-saki has quit [Ping timeout: 240 seconds]
leat has joined #nixos
asuryawanshi has joined #nixos
<AGTT>
Hi. I recently found NixOS and I am curious of how: "runtime dependencies are found by scanning binaries for the hash parts of Nix store paths" (from https://nixos.org/nix/about.html (#Complete dependencies)). Are binaries told to look up their dependencies when they are run? How? Would you have any pointers? (And how are runtime dependencies defined in this case? Libs/system() calls (e.g.)?) Thanks! :-)
<GlennS>
If you're planning to use a virtual private server, and don't want to have to muck around with the all disk formatting stuff, you might find NixOps helpful. It's pretty easy to get started with.
<kreetx>
AGTT: there is a nix pill that addresses that
<kreetx>
yeah, that's the one. I just read them, they are very good if you're interested in the guts of nix
<AGTT>
...which I am. Thank you. :-)
<jxf>
Q: What shell determines the one that runs when `nix-shell` is invoked? I'm using `fish`, but `nix-shell` starts `bash` and uses my `.bash_profile`.
pxc has joined #nixos
<Fare>
Hi. Is anyone using the android sdk under nixos? When I run it, it insists in "upgrading" itself into ~/Android/ at which point the result doesn't work, expecting a LFS setup.
<kreetx>
GlennS: yes, I'm trying to use that with my local virtualbox and managed to get it working
<GlennS>
cool
<Mic92_>
Fare: are you using the fhs environment for android? provided in nixpkgs?
<kreetx>
GlennS: I guess if I succeed in turning a machine to nixos, then nixos would just work (tm) from then on?
<kreetx>
*nixops would just work ..
<GlennS>
most likely
<GlennS>
unless you somehow messed up the boot loader I guess?
<GlennS>
but yeah, once you've got one generation of NixOS that boots ok, you're pretty safe to muck around and try things out, since you can always roll back to it
rosa has quit [Quit: rosa]
<kreetx>
I would try ec2 but I'm such a cheapskate! :p And also, the ec2 machines have cpu credits..
<GlennS>
oh wait, sorry, you correction
<GlennS>
*your
<Mic92_>
Fare: are you using android-studio?
<GlennS>
Amazon might do 1 free year of micro instance still?
<Mic92_>
GlennS: it does.
<kreetx>
already used up :)
<kreetx>
does google cloud do this too?
asuryawanshi has quit [Remote host closed the connection]
<Mic92_>
kreetx: if you are a company
<jtojnar>
jxf: it just uses bash, I use `nix-shell --run fish`
<Mic92_>
and they will send sales droid after you.
<Mic92_>
*sale droids
asuryawanshi has joined #nixos
rosa has joined #nixos
<jxf>
@jtojnar Just discovered the `NIX_BUILD_SHELL` environment variable, that seemed to do the trick.
<jtojnar>
jxf: oh, cool did not know this
Myrl-saki has joined #nixos
rosa has quit [Client Quit]
pirateking has quit [Quit: Page closed]
<Fare>
Mic92_, I've tried to use various things, to no avail
<Fare>
I was just trying to run android applications on my laptop
dex1983 has quit [Quit: Ping timeout (120 seconds)]
<Fare>
Mic92_, how do I use said FHS environment?
<Mic92_>
Fare: if you start android-studio, it will start this environment. However you can also `cat` the android-studio script and start the same script without passing the android-studio command
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<simpson>
$(nix-shell -p pythonPackages.virtualenv) will get you the basics, but this recipe from the docs may be closer to what you want.
peacememories has joined #nixos
<ldlework>
eh, I'm not on NixOS, but OSX, and I don't want to use nix shell, I just want to understand how I am supposed to know that the package path is "pythonPackages.virtualenv" when trying to install virtualenv
<{^_^}>
[nixpkgs] @yegortimoshenko pushed 4 commits to master: https://git.io/vp8y1
<{^_^}>
→ b901c40a by @yorickvP: oauth2_proxy: update module for extraConfig support
<{^_^}>
→ a037cbd4 by @yorickvP: oauth2_proxy: add keyFile, make some options optional
katona has joined #nixos
johnw has quit [Ping timeout: 256 seconds]
<{^_^}>
→ 048c991e by @yorickvP: oauth2_proxy: use explicit upstream default for setXauthrequest
<toby1851>
newbie question - suppose i'm working on a C program that links against readline... "nix-env -i readline-7.0p3" has installed the libraries, but not afaics the include files
<ldlework>
simpson well thanks for the solution anyhow
<{^_^}>
[nixpkgs] @yegortimoshenko merged pull request #39008 → Youtrack package and service → https://git.io/vpfCm
<{^_^}>
[nixpkgs] @yegortimoshenko pushed 3 commits to master: https://git.io/vp8yQ
<{^_^}>
→ 6c0806bd by @yorickvP: youtrack: init at 2018.1.41051
<{^_^}>
→ 6e33df0f by @yegortimoshenko: Merge pull request #39008 from serokell/youtrack
<{^_^}>
→ 000482ff by @yorickvP: youtrack: add service
<toby1851>
other distros with which i am more familiar would put them in a separate readline-dev (or -devel) package
<ldlework>
well
<ldlework>
nix-env -iA pythonPackages.virtualenv error: attribute ‘pythonPackages’ in selection path ‘pythonPackages.virtualenv’ not found
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<simpson>
toby1851: You can either use nix-shell to set up a development environment, or write a Nix expression for your C program and call Nix to compile your development tree.
ryantrinkle has joined #nixos
<toby1851>
ah! ok, i've heard of nix-shell, but had no idea it might be relevant to this sort of thing... will investigate
<{^_^}>
→ 29c66d9f by R. RyanTM: kid3: 3.6.0 -> 3.6.1
<{^_^}>
→ 4c19cae5 by @xeji: Merge pull request #39544 from r-ryantm/auto-update/kid3
<rihards>
hey, i'm having missing dependencies error when trying to install the Chart-diagrams Haskell package. here are the last bits of output https://pastebin.com/DqcYK8he this can be reproduced with nix-shell -p "haskellPackages.ghcWithPackages (pkgs: [pkgs.Chart-diagrams])"
<rihards>
anyone willing to run the command to werify that this is can be reproduced?
<katona>
Would someone please have a look at this package? I'm not sure if it is ready to merge, I haven't packaged anything for quite sometime now: https://github.com/Mounium/nixpkgs/tree/zeronet
<rihards>
any suggestions on how to aproach solving the issue are welcome as well
<rihards>
I see that haskellPackages.lens currently is at 4.16.1 in my channel (just updated). aren't such incompatibilities caught automatically (by the nix build farm or something..?)
hhomar has quit [Quit: leaving]
ma271 has joined #nixos
ma27 has quit [Quit: WeeChat 2.0]
ma271 has quit [Client Quit]
ma27 has joined #nixos
ma27 has quit [Client Quit]
<simpson>
rihards: nixpkgs Haskell versions come directly from Hackage, and Haskell library versioning is brittle at best. I'll attempt to reproduce, but I likely won't be able to exactly reproduce unless I'm on your commit.
aarvar is now known as typeastronaut
knupfer has quit [Ping timeout: 265 seconds]
<rihards>
well, for what it's worth - I just ran nix-channel --update
pxc has quit [Ping timeout: 268 seconds]
<rihards>
so this happens at the latest commit in the unstable channel
fragamus has joined #nixos
<simpson>
Yep, fails for me too, wanting `lens >=4.0 && <4.16`
<rihards>
ok, thanks. are these kinds of things supposed to be reported somewhere? or this wouldn't get fixed because the Haskell expressions come directly from Hackage as you say?
Huck777 has joined #nixos
ma27 has joined #nixos
<mpickering>
You can report it. You can probably fix it by adding one line to the haskell package configuration file
drakonis has joined #nixos
<simpson>
I'm not sure. My personal response has been to sigh and go write some non-Haskell, but I'm pretty comfortable not liking or using Haskell.
ma27 has quit [Client Quit]
<rihards>
btw, can I pass a path of different nixpgs (cloned from github) to nix-shell? had a quick glance to the man page and the manual bet found nothing
<simpson>
I reread the manual and I think that these kinds of fixes are appreciated both here and also upstream, since a break in unstable Nix like this probably reflects a break in Hackage.
<simpson>
I don't know what's supposed to happen with $(sudo pip) on OSX but it's probably not good.
<rihards>
simpson: thanks! I'll try to pin Chart-diagrams locally then. so I guess there's nothing to report? this should just get fixed automatically when nixpkgs pulls in the latest expressions from hackage, right?
ma27 has quit [Quit: WeeChat 2.0]
<simpson>
rihards: Yes. I'm trying to remember who to ping about that; I think Profpatsch?
<mightybyte>
How come when I do "nix sign-paths -k nix-serve.sec --all" and then "nix verify --all" it says a bunch of packages are untrusted?
<AGTT>
Is it possible to convert an existing environment, for example, Arch Linux, by installing Nix and letting Nix manage all the packages in the system, like NixOS?
<clever>
AGTT: i have done that manually before, and i nixos-in-place automates it, but you are still limited to the existing partition layout
<kreetx>
AGTT: there's also nixos-infect
<clever>
AGTT: i also have a kexec trick that lets you do similar and change the partitioning however you like
<AGTT>
yes, that's fine
<kreetx>
and there's a pull request for nixos-assimilate
<AGTT>
do you mean the fs layout?
<clever>
AGTT: yeah, both the fs and partition layout
alex`` has quit [Ping timeout: 265 seconds]
Fare has joined #nixos
<AGTT>
I wouldn't really want/need to change it, anyway...
<shapr>
Has anyone tried nixos on the novena arm board?
<kreetx>
if I'm not mistaken, the nixos-infect did try to do something with partition table too
<kreetx>
(could be wrong)
<clever>
AGTT: depends on the target, one machine ive wanted to convert has a 1tb / and a 20gig /home
<AGTT>
So, I just booted up ArchLinuxARM on RaspberryPi 3B. Would Nix still work on that?
<clever>
AGTT: you would need an arm build of nix to start with
pxc has joined #nixos
<AGTT>
I guess, I could get it?
<AGTT>
Or, compile it.
<clever>
AGTT: its not currently maintained by hydra, so you would have to compile it, and binutils currently fails to build on arm, so that wont work with current nixpkgs
<AGTT>
:-(
<AGTT>
:'(
<AGTT>
I guess I'll have to debug...
orivej has joined #nixos
<rihards>
Profpatsch: thanks for the info. I'll take a look at the relevant packages there
<M-bbigras>
Is it possible to check the build directory after a package is compiled? I'm wondering which krb5 doesn't have the kinit command.
<clever>
M-bbigras: nix deletes the build directory after the build has finnished, but you can view the log with `nix-store -l /nix/stor/epath` or maybe `nix log`
<symphorien>
M-bbigras: you can add a failing command at the end of the build and add option -K to nix-build, alternatively
<M-bbigras>
clever: thanks
Fare has quit [Ping timeout: 240 seconds]
orivej has quit [Ping timeout: 264 seconds]
astiles has quit [Remote host closed the connection]
astiles has joined #nixos
typeastronaut has quit [Ping timeout: 264 seconds]
<sphalerite>
hm, I'm getting quite a few corrupted paths on my chromebook. I haven't been doing much with it, it's just been sitting around (but powered on) for the past 2 or 3 weeks. Any ideas what might be causing this?
yegortimoshenko has quit [Remote host closed the connection]
<drakonis>
that might help
fragamus has joined #nixos
xy2_ has quit [Ping timeout: 264 seconds]
astiles has quit [Ping timeout: 268 seconds]
<drakonis>
i know it isn't the optimal solution for the user
<sphalerite>
wilornel: you still haven't given enough information here to really give anyone who doesn't just so happen to be having the exact same problem as you the means to help you
<simpson>
wilornel: That sounds like the wrong Python version was expected. You know about Python 2 and Python 3 versioning in general?
rosa has quit [Ping timeout: 276 seconds]
<drakonis>
did you use nix-shell
<drakonis>
by the way, it doesn't seem like imports installed with nix-env work
<sphalerite>
D: why do so many of the broken paths have to be ones that aren't in the binary cache and are expensive to build >_>
<ldlework>
drakonis are you talking to me?
<simpson>
ldlework: This looks like mix'n'match. The ideal is to do either full virtualenv (with Nix-provided Python or Homebrew or whatever) or full Nix.
<drakonis>
no that's wilornel
<ldlework>
simpson I'm trying to do the former
<ldlework>
with nix
<sphalerite>
drakonis: what do you mean by "imports installed with nix-env"?
<drakonis>
the python packages
<drakonis>
so far i've only gotten them to work correctly under nix-shell
<sphalerite>
oh yeah
<ldlework>
I only want nix for python and pip and virtualenv so that pip and manage packages outside of the store inside virtualenvs
<sphalerite>
that's pretty much intentional
<simpson>
ldlework: Do you have a .nix that I can try to reproduce with on NixOS?
<simpson>
ldlework: Oh, and it's choking on those final `pip install -r` invocations. Okay, then yeah, this might be intractable unhappiness in one of your requirements.
<ldlework>
simpson even just running pip without the requirement files causes that error to happen
<ldlework>
pip is busted in this expression
rosa has joined #nixos
<simpson>
Oh. That's an unfortunate bug in the manual then. Sorry.
<ldlework>
perhaps more than just the manual?
<sphalerite>
Does anyone know a filesystem that might be good to use on my chromebook, where random file corruption seems to be a thing that happens, which will work with only 2GB of RAM (hence not zfs)?
<moredhel_[m]>
zfs should work on low-memory systems, it's only if you are wanting to do deduplication (which on a home machine shouldn't be needed...)
<moredhel_[m]>
but, zfs won't save you from data-corruption
<sphalerite>
moredhel_[m]: it will at least detect it and throw I/O errors rather than giving back the wrong data
<{^_^}>
→ 5c7b7c36 by @erikarvstedt: pythonPackages.opentimestamps: 0.2.1 -> 0.3.0
<{^_^}>
→ 4e79f2fb by @erikarvstedt: opentimestamps-client: 0.5.1 -> 0.6.0
<{^_^}>
→ ac0ee0c8 by @dotlambda: Merge pull request #39599 from erikarvstedt/opentimestamps
simukis has quit [Quit: simukis]
<moredhel_[m]>
hmm, possibly... I never saw any memory issues on my laptop, it wasn't 2gb though
<drakonis>
moredhel_[m], no
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<drakonis>
zfs is awful on devices low on memory
<moredhel_[m]>
8GB+ of memory for the best performance. It's perfectly possible to run with 2GB or less (and people do), but you'll need more if using deduplication.
<drakonis>
zfs by design does everything on memory
<sphalerite>
zybell: udos?
<drakonis>
it will chug when dealing with large files
<zybell>
sphalerite: there was a filesystem to make install on fat possible in the old days. If that is not removed from kernel it could solve your problem.
<sphalerite>
zybell: why would I want to do that?
<sphalerite>
I want to detect corruption, not risk even more
<drakonis>
zfs wasn't designed for anything other than servers full of ram
<zybell>
the path is stored in files (with access bits and so) and that means md5sum is available.
<drakonis>
sphalerite, why not ext4?
<moredhel_[m]>
yeah, but it works pretty well on most machines
<moredhel_[m]>
I run it on all my laptops and have never had an issue
<zybell>
(can be run on it)
<sphalerite>
drakonis: because that doesn't detect errors
<{^_^}>
→ 61b96ab9 by @xeji: Merge pull request #39168 from vodkabot/master
<{^_^}>
→ e8c93655 by @vodkabot: twilight: init at 2018-04-19
<moredhel_[m]>
it seems people are very happily running zfs on small machines... I think it's only when running dedup, or when coping hundreds of GB of data, which isn't possible in your case because chromebook
migy has quit [Ping timeout: 240 seconds]
<drakonis>
tweaking though
<moredhel_[m]>
yeah, apparently you don't really need to... but it's easy to do if you think you need it.
migy has joined #nixos
<sphalerite>
oh, forgot another reason I can't use zfs… the nixos kernel does not work very well on here
<drakonis>
galliumos i guess?
lord| has joined #nixos
<drakonis>
also having a good cpu helps with zfs
<drakonis>
not just throwing tons of ram into it
<wilornel>
Alright, I figured out the issue I had with python..
<wilornel>
I keep getting this though... " ValueError: ZIP does not support timestamps before 1980"
<drakonis>
read what it says
<wilornel>
Right, it doesn't say which file though..
<{^_^}>
→ 06c7d4d4 by @Ekleog: opam: propagate `curl` and `unzip` dependencies
<{^_^}>
→ 8c27246e by @Mic92: Merge pull request #39619 from Ekleog/opam-issue-14466
rosa has quit [Quit: rosa]
griff_ has joined #nixos
<ghostyy>
hi, i have one nixos system, i would like to migrate it to another, i know you can just move over the configuration.nix file and switch to it, but what about all of the packages install with nix-env? how do i install all of the packagtes isntalled via nix-env on one computer on another computer?
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @jbgi to master « blobfuse: init at 1.0.1-RC-Preview (#39073) »: https://git.io/vp4Je
<gchristensen>
so I made some nice patches to the nixos docs, to make it easier to hack on, and now I can't get them b/c they're on a laptop which is packed away ....
<samueldr>
how frustratingly away?
vaninwagen has quit [Ping timeout: 260 seconds]
<gchristensen>
deep in my suitcase :)
<samueldr>
in another room or in a high security location on way to a cargo hold
Guest16581 has quit [Ping timeout: 264 seconds]
<gchristensen>
deep in my suitcase cradled by some shoes in my suitcase
<srhb>
I'm always happy when I'm reminded of the existence of that program. :P
<cindylind>
Anyone know how to do systemd.mounts? I'm trying and I think I'm doing it wrong.
leo60228 has joined #nixos
<ghostyy>
im installing nixos on a hard drive, on a computer running nixos
<ghostyy>
i see in ps its running nix-store --export or something like it
<leo60228>
I've installed Docker in my user environment (which I manage declaratively via a user overlay), set virtualization.docker.enable = true, and put it in my system packages, one at a time, in that order.
nD5Xjz has quit [Ping timeout: 240 seconds]
rosa has quit [Quit: rosa]
<ghostyy>
is it going to copy the host's /nix/store to the new nix installation's store?
<Mic92_>
cindylind: usually fileSystems."/mntpoint" should be enough, what do you want to achieve?
olto has quit [Ping timeout: 256 seconds]
<leo60228>
When I set virtualization.docker.enable and rebuilt, I ran out of disk space, so I cleared some up and retried.
<cindylind>
Mic92_: I want to have it mounted before a particular service and with special options.
<ghostyy>
if so i definitely dont want that, since the reason im installing nixos on another disk is because this one is corrupt and theres errors in my nix store that dont seem possible to fix
<leo60228>
It didn't error, but docker still didn't work. I added it to systemPackages, and it downloaded a bunch of stuff (like docker itself...), but it still doesn't work.
<clever>
ghostyy: nix will check the checksums as it does that
<ghostyy>
ah alright
<ghostyy>
man thats pretty cool...
rosa has joined #nixos
<ghostyy>
i just sort of assumed it would do something completely idiotic and write garbage to the new disk but i guess nixos is smarter than that
<leo60228>
I've checked, and it seems like the disk being full made it only partially build the derivation, but still think it fully built (because the hash was already calculated).
<leo60228>
How do I fix this?
<clever>
leo60228: what hash was already calculated?
<leo60228>
I believe that what I want to do is convince Nix that the checksum is invalid on my Generation 20 system configuration, but I'm not sure how to do that.
<leo60228>
not sure which
<leo60228>
i just know that nix thinks that changes are made that have not been made
<leo60228>
not 100% sure how nixos-rebuild works, but at least one derivation it generated seems to be broken
<leo60228>
but it doesn't think it is
<leo60228>
how do I check that?
<clever>
leo60228: how exactly is it broken?
<cindylind>
Mic92_: Thank you :)
<steveeJ>
completely unrelated question but I'm desperate to learn the answer: when setting up an ipsec site2site tunnel, why can't the ipsec gateways not directly talk to other side?
<leo60228>
clever: it thinks that whatever virtualization.docker.enable does was done, but it wasn't done to to running out of disk space
johnw has joined #nixos
<PolarIntersect>
hmm... I keep trying to install nixos on this machine, but grub just complains about "error: unknown filesystem" - which I assume means that the UUID is different when it boots from what it said when I installed
<PolarIntersect>
Does anyone have ideas on how to reconcile this?
<clever>
leo60228: did it fail during the build or activation?, have you rebooted since the failure?
<Mic92_>
steveeJ: I hope you are using strongswan and not racoon-tools
<gchristensen>
shlevy: 'round?
<leo60228>
build
<clever>
leo60228: nix will try again since it failed
<Mic92_>
and both or at least one machine has a routable ip
<clever>
leo60228: just re-run nixos-rebuild switch
<leo60228>
i have rebooted since the failure, because the switch crashed kde
<leo60228>
well, i've already tried
<steveeJ>
Mic92_: I'm using a BSD derivative called OPNsense. I'm not sure what's inside for the IPsec part
<clever>
leo60228: and?
<leo60228>
nothing happened
Imoom has quit [Quit: leaving]
<leo60228>
i added docker to systempackages and it installed it (presumably for the first time, because I haven't updated my channel since before the failure)
<clever>
leo60228: can you pastebin the contents of configuration.nix and the output of `ls -l /etc/systemd/system/` ?
<Mic92_>
steveeJ: on both sides?
<clever>
leo60228: your not supposed to add docker to systemPackages
<leo60228>
i guessed that
<clever>
leo60228: virtualization.docker.enable adds it to systemPackages for you
<leo60228>
but it wasn't working
<clever>
leo60228: can you pastebin the contents of configuration.nix and the output of `ls -l /etc/systemd/system/` ?
<steveeJ>
Mic92_: yes
<leo60228>
ok
<steveeJ>
Mic92_: the tunnel works, but the gateways can't talk to each other, which is unfortunate as I'd like to do DNS forwarding between them
<steveeJ>
Mic92_: well, it works when I set the source to the LAN address of the source gateway that wants to ping
<steveeJ>
Mic92_: that's a weird form of indirection
<clever>
ghostyy: try downloading the iso to /dev/shm and then dd it to a usb stick?, /dev/shm/ will bypass the hdd and just eat ram
<leo60228>
welp
<leo60228>
my nix store is 60GB
<Mic92_>
steveeJ: you should checkout what source ip it chooses, if you do not specify it explicitly
<leo60228>
this is gonna take a while
<Mic92_>
in tcpdump
<leo60228>
at least it's on an ssd
<leo60228>
clever: btw i'
cindylind has joined #nixos
greglearns has quit [Quit: Page closed]
<leo60228>
ve gotten fed up a few times with nix, a few random packages will be invalid because i'm too lazy to properly make an override
<clever>
leo60228: forcibly editing the nix store like that can cause all kinds of weird problems
cindylind_ has quit [Ping timeout: 260 seconds]
<Mic92_>
steveeJ: route get <ip> might be also handy
<clever>
Mic92_: i think you forgot an `ip` in the start:`ip route get 192.168.2.1`
<Mic92_>
clever: no this is for freebsd not linux
rosa has quit [Quit: rosa]
<clever>
ah
rosa has joined #nixos
<leo60228>
clever: path ‘/nix/store/8ly4sghqh667iqhwywwfw8jcqwhxxsv6-user-environment’ was modified! expected hash ‘15f38a085a64a734aba0bb48d39cddc83c90e4a6b05d922c6d9cebb6c6cfdeba’, got ‘97503c2066818605ac1fadeab7d0a5e20bbfa9cd6905c990aa1a04928856b2b8’ is the only error
<clever>
leo60228: something in a nix-env profile is either corrupt or was modified improperly
<clever>
but that wont cause docker/systemd weirdness
<clever>
once those 2 are ran, that "corrupt" path will be gone
<leo60228>
ok
<leo60228>
but why isn't docker working
<clever>
has the --verify finished?
<leo60228>
yes
<leo60228>
that was the only error
<steveeJ>
Mic92_: to clarify: I need to provide "-S IP1" where IP1 is the gateway's from the local internal subnet, not the external facing one
<clever>
leo60228: what is the output of `nix-store -q --deriver /run/current-system` ?
<leo60228>
was going to say at the beginning and forgot, i'm using nixpkgs-unstable
<Mic92_>
steveeJ: mMst of my knowlegde about routing tables is iproute2/linux specific. Linux would show a: `no route to host`, if it has no destination. In the other case it would not ommit the package, but maybe send it to wrong interface.
<leo60228>
wait no i'm not never mind
<leo60228>
i'm wrong
<cindylind>
I still can't get systemd.mounts working. The type it wants is a list of submodules 'config.mod = [ { foo = 1; bar = "one"; } { foo = 2; bar = "two"; } ];'. But when I do something like `systemd.mounts.something = [ {where = "somewhere";} ];` it complains.
<clever>
leo60228: what does `sudo nix-channel --list` say?
<clever>
leo60228: and then pastebin the output of `nix-store -qR /nix/store/vzsfy8hy86zw4qdmb022wniww4src6rf-nixos-system-leonix-18.03.132083.06c576b0525.drv`
<leo60228>
how do I use an unofficial kernel patch in configuration.nix? not on my main system, working on an iso that'll need a patch
<Mic92_>
steveeJ: yes, this what I expected. linux picks the most specific source ip on its own. I am not sure why this not the case for freebsd. But I also don't know what the routing table looks like for this VPN setup.
<clever>
leo60228: what is the output of: nix show-derivation /nix/store/9y0gqcdg6228qgrjvz4qax7m2iw4z0gq-system-units.drv
<clever>
looks standard, nothing that would break things
Ariakenom has quit [Ping timeout: 248 seconds]
<steveeJ>
steshaw[m]: what is the transfer network in this case?
cindylind has quit [Ping timeout: 260 seconds]
<clever>
leo60228: as an experiment, try turning on services.openssh.enable and then pastebin the output of nixos-rebuild switch
<leo60228>
one second, i'm updating the nixos channel to see if that's the problem
<leo60228>
...
<leo60228>
why is this installing 100s of packages
<leo60228>
i'm not on the unstable channel
<clever>
leo60228: something changed that caused a mass-rebuild
<srhb>
leo60228: 18.03 bumped today.
<leo60228>
oh ok
sanscoeur has quit [Ping timeout: 240 seconds]
<leo60228>
btw kde crashed at this point last rebuild if i disconnect
<clever>
leo60228: thats also fairly good, it will basically force nixos to rebuild everything, so anything that might have somehow gone corrupt and un-detected is going to be rebuilt