IRC channel logs


back to list of logs

<bavier>has anyone else here had problems lately resolving mDNS hostnames?
<bavier>my /var/log/messages shows a segfault in nscd
<efraim>I had an embarassing networking issue that was entirely related to an unplugged Ethernet cord
<DoublePlusGood23>The links in my /var/guix/profile/system seem to be broken
<DoublePlusGood23>Any idea how to fix
<DoublePlusGood23>system-1-link/ etc. seem fine
<DoublePlusGood23>scratch that
<DoublePlusGood23>only system-1-link/* seems properly mapped
<rekado_>hello from ICAB!
<efraim>I'm almost there!
<efraim>I don't know if we have anything set up for streaming but if there is good WiFi we can try jitsi
<rekado_>no streaming setup unfortunately
<rekado_>we seem to have blown a fuse just now, so we’re a little delayed
<rct>Hi :-) I wanted to install guixsd on my machine, but I can't find any information about the configuration of filesystems with btrfs subvolumes in config.scm - all the examples I found are about ext4 or lvm. I couldn't find a reference for config.scm and there's no man page for it. Could you please tell me, where I can find a reference or similar documenation?
<mbakke>rct: I have not tested it, but did you try declaring the mounts using the regular (file-system ...) reference? See
<ng0>rct: actually LUKS, not lvm. we don't have lvm support at the moment
<ng0>I can dig into my saved configs of some of the people around here and see if I have an btrfs example
<ng0>grep says no, but there should be something in 'gnu/tests/install.scm' that could help you
<ng0>so it seems like it should just work like any other filesystem
<ng0>I haven't used btrfs in a while
<rct>ng0: Oh sorry, yes I meant LUKS :-D Ok, I'll check that - thank you :-)
<orelozno1>Hello :) As a free software end user, I tried the virtual installation of GuixSD-0.14.3 with libvirt and qemu.
<orelozno1>It was promising but I was forced to uninstall libvirt and no way to reinstall it properly.
<orelozno1>So, I went down on virtualbox (debian testing repositories). My last command line, at this point : # cp /etc/configuration/desktop.scm /mnt/etc/config.scm
<orelozno1>If I understand, it is necessary to modify the copied file in /mnt/config.scm but vi tells me that it is read only.
<orelozno1>So I'm looking up a way to adapt config.scm :)
<efraim>try saving the file with :w!
<orelozno1>efraim : OK, i'm going to test it now, thank you :)
<orelozno1>It seems to be ok ! :w! did'nt seem to work, i've tried :wq and the file is modified :)
<mbakke>Not the best weather for walking to ICAB.
<ng0>uh. I just searched through my local mail archive.. did Fredrik ever solve their brtfs subvolumes issue? are subvolumes in some way, even if hacky, supported right now?
<ng0>or is this still something we need to fix?
<ng0>so… we have a rather basic btrfs integration. bug report for feature completions incoming later today/this week
<dvn>where can I find everyone at ICAB? I'm here, but I just see MySQL people :)
<dvn>where can I find everyone at ICAB? I'm here, but I just see MySQL people :)
<mbakke>dvn: we're in the dining area.
<dvn>ah, ok
<dvn>The large group?
<dvn>mbakke: ^
<mbakke>Did you find us?
<dvn>no... I'm very confused
<dvn>It's like I'm at the wrong place... Are there mysql people where you are?
<mbakke>Haven't seen any! Did it say ICA at the door?
<dvn>ICAB? yes
<dvn>There must be another ICAB building
<mbakke>I'm going out on the street. Look for a confused person with GuixSD hoodie.
<mbakke>Building number 8.
<dvn>Yea, I'm somewhere else then...
<dvn>I'm here:
<mbakke>Yeah, that's another ICAB :/
<dvn>:( I see..
<dvn>So I went to the address on here first:
<dvn>However it was just residential flats...
<dvn>and the linked webpage "" is down
<dvn>searching for icab brussels brought me to where I am now... Which does have a FOSDEM fringe meeting, coincidentally --- for MySQL :P
<mbakke>Yeah the building doesn't look like much - I walked straight past too.
<mbakke>But you were in the right street.
<mbakke>There's still time though :)
<mbakke>And the door actually says ICA here.
<dvn>I see... OK, so see you in about 40 minutes
<mbakke>Great :)
<bavier>wow, quick the list of attendees
<bavier>ACTION squints through morning eyes
<mbakke>bavier: When are you coming? :)
<bavier>I'd love to, but another year
<amz3>^ link to guix / guile stuff ^
<Digit>if any of those items are made with pure hemp, i'd buy em each as fast as i could.
<Digit>s/pure/organic/ i meant
<bavier>I suppose this is a good start for our aarch64 build slave:
<bavier>3087 store items were analyzed:
<bavier> - 280 (9.1%) were identical
<bavier> - 0 (0.0%) differed
<bavier> - 2807 (90.9%) were inconclusive
<efraim>0 different is good
<bavier>ah, I forget I've cherry-picked the gnulib-lock commit out of core-updates on this machine
<bavier>that's probably explains the large number of inconclusive
<efraim>guix build --no-grafts --keep-going --check $(guix package -A | cut -f1,2 --output-delimiter='@') ?
<bavier>I mean, this machine just isn't in sync with hydra, since that commit changes a lot of derivations
<cogrendel>The postgres service is unable to start for me on GuixSD, and I am unsure of how to diagnose my problem.
<cogrendel>I get the error: ERROR: program "/gnu/store/9622izzy7p5rcgls1ikd6f2fpyx45rhy-pg_ctl-wrapper" exited with non-zero code 256
<cogrendel>when trying to start the service
<cogrendel>Does anyone have any advice on how to figure out what's going on?
<buenouanq>post your config.scm
<buenouanq>or the postgres service part of it
<buenouanq>is all I have and it works
<cogrendel>(postgresql-service) is all I have as well
<cogrendel>Here is my full config
<buenouanq>weird things have been going on with services
<buenouanq>I've been meaning to explore it more before I send in any kind of report/email, but I don't have time and maybe should just pass the buck and see what happens
<buenouanq>this is a new one I haven't seen yet though
<buenouanq>so maybe whatever is wrong is getting worse (;゜Д゜)
<cogrendel>As another small issue my timezone seems to be wiped on each `guix system reconfigure`. After it completes I am on UTC time. Has anyone else experienced this?
<ng0>do you run a timeservice?
<ng0>otherwise I'd sugesste either running one or/and catching up with tlsdate with the leap + force option and then writing it to the hardware clock
<cogrendel>I'm not sure what you mean by timeservice, but I run ntp and have my timezone set in my config.scm. Is this insufficient?
<ng0>usually it's enough
<ng0>but your hardware clock might be out of sync
<ng0>see also
<ng0>do you dualboot with Windows? many years back this used to happen to me, that Windows wrote the time to hardware clock on shutdown
<cogrendel>I don't currently, but this computer did dualboot windows before I installed guixsd.
<ng0>what does hwclock --show say?
<ng0>compared to your current timezone and time, is there an offset?
<cogrendel>my hwclock shows: 2018-02-02 12:17:59.059542-0500
<cogrendel>which is the correct time and timezone
<cogrendel>`date` shows: Fri Feb 2 17:19:26 UTC 2018
<cogrendel>To clarify, my timezone is correct when I boot, it just goes to UTC on reconfigure instead of EST.
<ng0>and you did set EST or the appropriate value in the config?
<ng0>for now if you are sure it is correct I'd sync the software clock to the hw clock
<ng0>I mean safe the sw value to the hw
<cogrendel>My timezone is set to EST in the config file. My config:
<ng0>like I wrote, I'd safe the SW value to the HW
<ng0>if it changes again after next reconfigure, you might know more
<cogrendel>Ok, thanks ng0
<CharlieBrown>civodul: Could you help coerce me into reading/hearing French every day?
<CharlieBrown>civodul: kaj uzu chi tiun lingvon anstatau la anglan
<ng0>I know this has been asked before in the past.. but copying the /gnu/store location from one drive to another is as easy as simply guix copy ?
<ng0>I'd move the mountpoint and its content
<ng0>so right now it's located under /dev/sda3 and /dev/sdb1 would be the new /gnu/store
<pkill9>ng0: i moved it to another partition with `rsync -va /gnu /home/guix` and then bind-mount /home/guix to /gnu
<pkill9>e.g. `mount --bind /home/guix /gnu`
<pkill9>though i do it in fstab
<pkill9>haven't had any problems
<CharlieBrown>ng0: Will a simple cloning of the drive not work with GuixSD?
<ng0>the overall reported health of all the spinning rust is not looking so good
<ng0>so I need to know this in advance for the substitutes server
<ng0>almost all are once more on old_page/pre-failure
<CharlieBrown>I can't use GuixSD if disk clones won't work.
<catonano>sneek later tell congrendel I had similar issues with the postgres service in the past. Please write on the mailing list, I will reply there
<sneek>Will do.
<ng0>I think if I wait a couple of more months I could just buy a server instead of maintaining old patched up desktop hardware
<DoublePlusGood23>ng0: ケーラブ sounds like a job for btrfs snapshots
<ng0>well, if btrfs integration in GuixSD would be working..
<ng0>if I would use btrfs, I'd use not just one plain volume
<ng0>*would be feature-full working
<ng0>at least subvolumes aren't well integrated, if at all
<jackhill>ACTION wonders is a future subvolume interface could be similar to an LVM interface.
<ng0>I walked through subvolumes with someone today.. the person will send an bugreport as soon as they can
<icarious>Hi all
<icarious>Is it possible to build and use guix from 100% source only and not the binaries?
<icarious>I mean, does it support recursive (dependencies included in the building process) compilations like Gentoo?
<vagrantc>ACTION thought there were some bootstrap binaries for that
<vagrantc>icarious: but basically
<vagrantc>in fact, i've accidentally triggered source-only builds by setting the substitute servers to an empty variable
<OriansJ>icarious: well the problem is that if you just start with a filesystem with a bunch of source packages on a piece of hardware, you can't actually do anything
<OriansJ>So we include the binaries that we make from source that we include to provide things like the Linux Kernel, GCC, bash and other functionality required to build the programs you wish from source
<OriansJ>The bootstrap binaries can be independently reproduced by you should you so desire from the sources provided
<icarious>OriansJ, I understand. Actually what I was looking for is a prebuilt debootstrapped stage like Gentoo does and then proceed from there without using binaries anymore.
<OriansJ>icarious: that is the default guix bootstrap binaries
<OriansJ>--no-substitutes exists for that very reason
<pkill9>is there an option that functions as opposite to --no-substitutes, that only uses substitutes? e.g. --only-substitutes
<icarious>OriansJ, and is it possible to re-build the entire system (something like emerge -avuDN @world)?
<ng0>think of the bootstrap binaries like a very tiny minimal stage3 (or even stage2? been too long)
<ng0>at this point you should read more about GuixSD and forget about Gentoo terminology
<ng0>-avuDN @world is not a thing.. packages will be updated when their package receipes change
<OriansJ>icarious: well, technically that is what guix system init /mnt/etc/configuration.scm /mnt --no-substitutes does
<icarious>I have to try guix I guess, Merely asking will make me sound like an idiot
<DoublePlusGood23>icarious: I think guix system build /etc/config.scm would do something similar
<icarious>last question, how does the user ensure the sources are automatically verified against signature?
<OriansJ>icarious: there are no dumb questions, just room for greater collective understanding but if you don't feel comfortable sharing here, I understand.
<icarious>Pardon my if I have to quote Gentoo again. They have this thing
<DoublePlusGood23>Anyone willing to help me package Telegram? Lotta dependencies to go through...
<ng0>that#s not even a good validation
<ng0>I've used Gentoo for a while and never found the trustworthy "pulls" trustworthy enough.
<catonano>DoublePlusGood23: I suppose many of the Telegram deps are npm based, aren't they ?
<DoublePlusGood23>catonano: It's actually a native Qt app
<ng0>icarious: all builds are verified against their recorded sha sum
<OriansJ>icarious: if you look at each of the package definitions you'll notice that they include (sha256 (base32 "04lvyyp7is ... that is the checksum of the source code being downloaded
<vagrantc>many packages in guix build reproducibly, which would provide some measure of confidence, but there isn't a good trust path to verify the git repository defining what and how to build packages :/
<vagrantc>pretty much the one deal-breaker for me using guix more than just experimentally
<adamvy>vagrantc: what do you use instead?
<icarious>Wait. guix doesn't support LVM?
<vagrantc>adamvy: debian
<icarious>I use a LVM+LUKS setup
<OriansJ>vagrantc: well we could simply have a signed git repository and do what arch does and have the guix developers sign their work
<vagrantc>OriansJ: i think individual commits may already be signed
<vagrantc>icarious: LVM is a little less interesting since everything is installed in /gnu/store/
<icarious>vagrantc, I just want a 100% disk encryption setup with btrfs. LVM satisfied it till now :P
<vagrantc>ACTION suspects it wouldn't be too hard to add LVM+LUKS support
<vagrantc>it's already got the LVM binaries, just need to add them to the initrd and teach the initrd how to make use of them
<icarious>ya only the inird would be the hurdle
<vagrantc>i've got installs on a luks partition
<catonano>DoublePlusGood23: ah ok. I thought it was a webb app. My suggestion is set up a public branch somewhere and work on it as muchh has you can. Then write on the mailing list on thhe occasional hurdle and the final goal
<vagrantc>so it already knows how to support luks
<catonano>more than someone would lend a hand
<catonano>DoublePlusGood23: it happened to me recently. I was struggling and someone helped me and finished whaht I was trrying to do
<DoublePlusGood23>icarious: Yeah, that's annoying rn. I've gotten btrfs+luks with a encrypted swap partion to work before, but I don't recall how.
<DoublePlusGood23>catonano: That's a good idea. What's the prefered git hosting service? I *could* use github but it doesn't seem like the most ethical choice.
<ng0>you can use whatever.. really doesn't matter as your personal choice
<ng0>as long as it doesn't state "official" it's not attached to the project.. my opinion
<DoublePlusGood23>ng0: fair enough
<catonano>DoublePlusGood23: lie ng0 says. Official projects worry about ethically hhosting their code. Single persons can be more relaxed about that
<DoublePlusGood23>catonano: that makes sense.
<DoublePlusGood23>Notabug seems good enough
<bavier>guix does not mind hosting wip branches in its repository
<bavier>there are quite a few there already
<OriansJ>savanna is always an option, so is gitlab
<ng0>for which you need savannah and a blessing that your code is alright..
<bavier>right :\\
<OriansJ>ng0: not an issue for GPL code
<ng0>the code -> as in Guix code
<OriansJ>stage0 got approval in less than a week
<ng0>I think I'm allowed a bit of pessimism after 3 years.. I'm just working outside of Guix and send my patches
<OriansJ>GNU projects like Guix shouldn't even take anywhere close to that long
<adamvy>icarious: I also have LUKS running without LVM. Was very easy to set up.
<ng0>actually I'm working the way I do because it's easier for me and I can do whatever.. I see issues with getting access, but I don't really care anymore.
<icarious>adamvy, That's possibly only with a single partition right? For me I have an encrypted LVM container and got 3 logical volumes underneath it
<ng0>you create the luks container , open it, and create the partitions inside
<DoublePlusGood23>ng0: can you have partions inside?
<ng0>so /dev/sda3 could be your luks, and there can be volumes inside
<ng0>that's what I just wrote
<vagrantc>supported in guix's initrd?
<ng0>there's some good guides online, in particular the Archlinux listing.. a bit messy, but you can use it
<DoublePlusGood23>ng0: With LVM I have, but without it I though it was only able to format it to a single filesystem
<icarious>ng0, you mean this?
<icarious>This is precisely what I have. But isn't this LVM?
<ng0>I had DoublePlusGood23 lvcreate?
<ng0>damn ..
<ng0>DoublePlusGood23 lvcreate?
<ng0>this way you add logical volumes
<ng0>h.. one moment
<ng0>it's gettting late and I'm mixing thingds up
<icarious>I am confused now. How is using pvcreate / vgcreate / lvcreate "NOT" using LVM?
<ng0>I know that I had configs i nthe past wiuth this
<DoublePlusGood23>ng0: yeah but under GuixSD you can't use LVM
<DoublePlusGood23>as root
<ng0>if you are really curious and bored, walk down the history of my config repo... I don't know how or what I had in the past, I just know it wasn't LVM and it was too long ago to be in the present content of it
<ng0>but it was re-init'ed in august.. so my hope that it's in there is low
<ng0>ah yeah
<DoublePlusGood23>ng0: right now I have one partion, luksFormated, then ext4 formated, with a swapfile
<DoublePlusGood23>btrfs complicates things because I can't use the swap file and without LVM I can't have a second partion under the luksFormated portion. I think I recall making a second luksFormated partion with an ephemerial key of somesort to use as swap before, but I can't quite remember what magic went on.
<icarious>DoublePlusGood23, I wonder if a single btrfs partition with subvolumes could be a replacement for LVM for such purposes?
<DoublePlusGood23>icarious: It's just the problem of having a swap, sadly btrfs wouldn't do it afaik
<ng0>that was exactly an issue I dealt with today.. bug report incoming soon
<icarious>DoublePlusGood23, I know. It doesn't support a swap file.
<ng0>wrt sub volumes
<icarious>ng0, what issue?
<DoublePlusGood23>Now the real solution would be to port zfs :p
<ng0>GuixSD btrfs doesn't understand sub-volumes for now.
<icarious>I so want to see ZFS on GNU+Linux really
<icarious>: /
<icarious>Never used it though. But if btrfs can be so good, I wonder what ZFS would be
<DoublePlusGood23>icarious: It's great on ubuntu. I just finished an Ubuntu with an encrypted ZFS root the other day. I also use it on FreeNAS
<icarious>ACTION is not into non-free distros (unless they can be used freely down to the kernel level)
<icarious>ACTION uses Gentoo with linux-libre + @FREE USE Flag and couple of other manual hacks
<DoublePlusGood23>icarious: :p Maybe Trisquel 8 will support it
<icarious>DoublePlusGood23, in the post-apocalyptic future. Trisquel's dev is slow as heck in recent times
<icarious>Heck, gNS is slower :P
<DoublePlusGood23>icarious: It's almost like they're tracking Debian instead of Ubuntu xp
<icarious>DoublePlusGood23, at the rate they are going, soon they will start tracking Debian old stable :P
<icarious>I have used Trisquel around 2011-2012. Was my first FSDG distro. Then I used / contributed to Parabola for years until 2015 when I switched to Gentoo totally
<bavier>Trisquel is years ahead of the distros I work on at $dayjob
<DoublePlusGood23>icarious: Very true. I had 8 running on my x200 for a little while, but I couldn't stay away from guix for long xp
<icarious>That good eh?
<icarious>I wonder. I should tinker around with guix now
<icarious>even though I will miss the USE Flags (The only thing I think I will miss)
<DoublePlusGood23>icarious: Well it's very hacker-friendly
<icarious>I have heard. lisp-like and all
<DoublePlusGood23>icarious: Still clearly in beta however.
<icarious>emacs of distros as they say
<icarious>i know. well, the problem is I have a very complicated setup for it to be my mainstream. EFI-System Partition + LUKS Container + BTRFS + BTRFS SubVolumes to be exact
<icarious>No /boot crap either
<icarious>But I would try anything (As long as its free), for as long as I can stay away from the systemd stuff
<ng0>welcome to eating your own doggo food 101 with this setup.
<icarious>ng0, lol
<DoublePlusGood23>icarious: Not a fan of systemd
<icarious>DoublePlusGood23, I used to be at one point. I even defended the useless slander in the past. But recently it went beyond my acceptable threshold of intrusion
<icarious>*opposed the slander I meant
<icarious>This was sayonara to systemd for me
<icarious>at that point I ditched systemd in favour of openrc
<ng0>efi, doable already. the rest will benefit from everyone who works on it
<icarious>EFI is crap and messy. Specially the whole 260MB FAT32 ESP Partition non-sense
<DoublePlusGood23>icarious: I can't say much about Shepherd besides I like that it's portable
<icarious>DoublePlusGood23, Sherpherd?
<icarious>Oh ok
<DoublePlusGood23>icarious: GuixSD's init sdaemon
<icarious>guix's init
<icarious>I sort of like openrc now. Debian should have opted for it instead of radically moving to systemd
<DoublePlusGood23>catonano: It looks like npm is not packaged...I was going to take a stab at as well
<DoublePlusGood23>that is a web app
<ng0>that is node you want.. and after that you get a depressing wake up call about packaging npm and nodejs modules ;)
<ng0>we have a branch for this
<ng0>and the progress is slow but steady
<ng0>but reality hits hard
<DoublePlusGood23>ng0: It doesn't want to play nice?
<ng0>asks those involved in getting it to integrate.. and our mailinglists should have threads. tl;dr it's a horror story.
<bavier>here's a writeup:
<bavier>I don't think the situation has changed much since
<ng0>so the first rule of npm is you don't speak of npm. the second rule of npm is you do not speak of npm. and the third rule is you repeat a strange mix of 1 and 2 200 times while you are jogging up and down the Alps.
<ng0>a shortcut we wouldn't use today but you could try is to include binary blobs of the nodejs modules
<OriansJ>bootstrapping is always a horror story, otherwise it wouldn't involve so much work
<DoublePlusGood23>ng0: I'm probably just gonna stick with packaging Telegram. I thought being a web app might've been a simpler job
<icarious>DoublePlusGood23, you are the guy who packages Telegram in f-droid?
<MrCStone>I have recently installed GuixSD and have a couple problems! Is this the place to ask or is just related to the package manager?
<ng0>Guix and GuixSD are here
<ng0>hm.. speaking of bootstrapping.. I've started work on packaging Xen last month.. because, you know $reasons.. I'm just interested in the vanilla version of it, but it seems it could go into Guix at some point? parabola has packages for it if I wasn't too tired looking at it.
<bavier>I would assume if parabola has a package then it would be fine in Guix too
<MrCStone>Great! So three main problems right now! a) Unless I login as root, wifi doesn't start! I just su - and it starts on its own, but cant as a user
<MrCStone>Second, I can't find a way to set the keyboard map on X apart from manually setxkbmap (I tried putting it in profile, both in etc and ~/.profile, and only works if I manually source it)
<icarious>ng0, what about KVM+libvirt in guix? working?
<ng0>try it?
<MrCStone>And 3, guix pull (before upgrading) stops at around 73%
<ng0>I know many things but not everything, icarious :)
<MrCStone>And says something about stack overflow
<ng0>I mean, I'm just at the start of Xen.
<ng0>kvm+libvirt should just work
<DoublePlusGood23>icarious: lol no, I'm just a guy
<icarious>lol ok. I have never used Xen. Found the Archwiki documentation to be traumatic to even try
<ng0>so far I have a compiling Xen but need the rest + testing
<bavier>MrCStone: I've not seen the stack overflow from 3); I'll try it now
<MrCStone>bavier: thanks! Maybe it's just that I'm a bit of a noob (been using linux for some years, started a sysadmin course and decided to give guix a try)
<DoublePlusGood23>MrCStone: Are you using NetworkManager?
<MrCStone>Yep, had to connect using nw-tui though
<bavier>MrCStone: great! glad to have you here
<MrCStone>During install it was all good, but then once installed it didn't
<MrCStone>So I killed all networking services and restarted networking, used nm-tui and it did work
<DoublePlusGood23>MrCStone: That is really strange...
<MrCStone>But the weird thing now is that it works under root just as I login, but if i login with a user it doesnt
<MrCStone>Okay, maybe I should nw-tui with the user?
<DoublePlusGood23>MrCStone: Hmm. What groups are you in
<MrCStone>So that it saves the config somewhere it can access?
<MrCStone>Give me a sec
<MrCStone>I pretty much copied the template changing a couple things
<MrCStone>users netdev audio video wheel
<ng0>if you create a profile (connection) with root, it will s imply be for root with NM.. you need to connect as normal user.
<ng0>*create the connection I meant
<MrCStone>But when I connect with root, and then change to the user
<MrCStone>It keeps on working
<DoublePlusGood23>ng0: oh, that makes sense
<DoublePlusGood23>ng0 to the rescue
<ng0>and to change the connection details then, you need to gksu or whatever to root
<DoublePlusGood23>bavier: Web Dev is hell
<MrCStone>Is okay with normal sudo?
<ng0>what I meant is, if you use for example Mate as DM you will be asked to auth as root