IRC channel logs


back to list of logs

<mitc0185>anyone around?
<davexunit>mitc0185: hello.
<davexunit>around but somewhat busy. first time here? welcome!
<mitc0185>hi davexunit
<mitc0185>it is
<mitc0185>I just sent an email to the gnu-system-discuss list
<mitc0185>was looking for some help configuring my network adapter (wireless)
<mitc0185>iwconfig does not support passwords
<mitc0185>and the network I have access to is configured to use a password
<mitc0185>was wondering if there's a different program I can use that would support that
<davexunit>mitc0185: wpa-supplicant
<davexunit>I don't know anything about configuring it, but I know it can connect to password protected WAPs.
<mitc0185>davexunit: np. I'm happy to figure out how to use it
<davexunit>cool :)
<davexunit>good luck
<mitc0185>davexunit: is wpa-supplicant available by default?
<davexunit>I don't think it comes on the installation image, if that's what you mean.
<davexunit>not in the 0.8 release, anyway. I believe we added it after the release.
<mitc0185>ok, so would I need to compile and install it in order to use it?
<davexunit>no, just 'guix package -i wpa-supplicant'
<davexunit>that will download and install it
<mitc0185>I am in a remote location
<mitc0185>do not have all of my gear
<mitc0185>I will see if I can scare up a network cable though
<davexunit>ah, bummer. :(
<mitc0185>nah it's all part of the fun
<mitc0185>I just read about Guix in my FSF newsletter
<mitc0185>excited to have it up and running already
<mitc0185>and maybe get a chance to help out with something
<davexunit>mitc0185: I'm glad that article inspired you.
*davexunit wrote it
<rekado>in the 0.8 announcement lvm2 is listed as an updated package. The system installation page in the manual says that LVM support is missing, though. Is this still the case?
<jmd>rekado: where was that announcement?
<rekado>the lack of LVM support is mentioned here:
<jmd>I think it means that the lvm tools are available, but installing on an LVM is not currently possible.
<civodul>Hello Guix!
<amirouche>héllo guix :)
<th3kent>hello guix!
<th3kent>am trying to install 0.8, and it's failing with the following:
<th3kent>unionfs invoked oom-killer: gfp_mask=0x...
<th3kent>unionfs cpuset=/ mems_allowed=0
<th3kent>Out of memory: Kill process 2068 (guix) score 526 or sacrifice child
<th3kent>Killed process 2068 (guix) total-vm:829080kB, anon-rss:201792kB, file-rss:0kB
<th3kent>guix system: error: build failed: unexpected end-of-file
<th3kent>2470 operations
<th3kent>any help troubleshooting this will be much appreciated.
<jxself>Out of memory:
<jxself>That seems to be the key part.
<jxself>Add more memory?
<jxself>Also, if you haven't enabled subsitutes then that would be good too.
<th3kent>at a minimum, how much memory is required to run "guix system init ..."? my laptop has 512mb.
<jxself>I'm not sure. I imagine that it depends on what needs building, which would depend on what packages you've elected to include in your config. So it's not a question of how much RAM to run Guix system init but how much RAM to compile those packages ;)
<jxself>Enabling binary substitutes instead of building it all yourself would likely help.
<th3kent>ok. i think substitutes are enabled by default, my installation had a list of files to download, following by a list to build.
<jxself>I don't think they are enabled by default? Or at least didn't used to be. If they are now that's great.
<rgrau_>IIRC they aren't
<rgrau_>I had to get the file and issue the command
<rgrau_>putting that file in the basic distro would be a win
<davexunit>rgrau_: it is in the distro
<davexunit>find /gnu/store -name
<th3kent>ok, lemme try "explicitly" enabling substitutes.
<rgrau_>oh... ok, then another thing I missed :/
<davexunit>I think that hydra should be authorized by default
<davexunit>rgrau_: well, it's not exactly out in the open.
<davexunit>I think the confusion occurs when the install image uses substitutes, and then the installed system doesn't.
<tadni`>civodul: Hey civ. o/
<civodul>Sleep_Walker: i came up with this patch that may be of interest to you:
<civodul>nkar: did you have a chance to try it out?
<nkar>is it the issue about /boot?
<nkar>no, I didn't. and I don't have time for this right now. maybe will try tomorrow morning.
<th3kent>hello again guix. i made some progress ...
<civodul>hello th3kent
<th3kent>i have another partition, which i plannewd to use for swap. so i issued mkswap on the partiton and enabled it. my installation went further, but failed with permission errors on /gnu/store/*-linux-libre-3.17.2/ when copying the pata_* modules i need for my i.d.e drive in my laptop. i added write permissions recursively for user, and the installation complete with no further errors.
<th3kent>hello civodul.
<Sleep_Walker>civodul: I will try
<Sleep_Walker>but that is not problem for me
<Sleep_Walker>I copied the files manually and it should be sufficient
<Sleep_Walker>problem now is, that init exits (with 0 return code) and I got kernel panic
<Sleep_Walker>I need to add some debugging/tracing (equivalent to `sh -x') to see, where the issue is
<Sleep_Walker>but not with my current scheme/guile skills
<th3kent>i have installed 0.8 and the reboot loads the pata_atiixp module. i still end up at the (guile-user)> prompt. the backtrace has "In /gnu/store/*-module-import/gnu/build/linux-boot.scm:\\n385:7 2 (#<procedure 1e943c0 at /gnu/store/...\\nIn srfi/srfi-1.scm:\\n619:17 1 (for-each #<procedure load-linux-module* (file)> ..."
<Sleep_Walker>and it would be nice, if there is some chroot helper to set environment correctly
<civodul>th3kent: could you post all the details to bug-guix?
*civodul is starting to mix up bug reports :-)
<th3kent>ok, will do.
<civodul>Sleep_Walker: do you get a Guile prompt, or just a panic?
<Sleep_Walker>civodul: just panic
<civodul>and what's the message before that?
<civodul>"cannot mount root-fs" or similar?
<Sleep_Walker>IIRC that ext4 partition was mounted
<Sleep_Walker>but I have to retest and recollect
<Sleep_Walker>it would be nice, if it would duplicate output to /dev/kmsg
<Sleep_Walker>(so I can collect it on serial)
<th3kent>i get a guile prompt.
<Sleep_Walker>(like `exec > /dev/kmsg 2>&1')
<th3kent>message before guile prompt is "ERROR: In procedure load-linux-module: Unknown error -1".
<th3kent>oh, i see, question was for Sleep_Walker, not from Sleep_Walker (-:
<Sleep_Walker>th3kent: parallel bug reporting :)
<th3kent>yeah! :-D
<Sleep_Walker>civodul: it happened after all that 'adding < >'
<civodul>Sleep_Walker: maybe you could run it with "console=ttyS0"?
<civodul>Sleep_Walker: adding what?
<Sleep_Walker>during first boot there is sequence steps for different values
<Sleep_Walker>users, groups, apps, profiles
<Sleep_Walker>I'm not sure, it's too quick
<Sleep_Walker>console=ttyS0 will redirect _kernel_ output to serial
<Sleep_Walker>and I need _userspace_ redirection to /dev/kmsg, to be part of the log
<civodul>well, everything that goes to /dev/console is redirected
<civodul>that includes user-space stuff
<civodul>so you did see the creation of user accounts, groups, and all that on first boot?
<Sleep_Walker>panic occured after that
<civodul>ooho, that's rather good news, then ;-)
<Sleep_Walker>I will wipe the partition and retry
<civodul>do you know if there was any "Service foo started" message, from dmd?
<Sleep_Walker>well, it needed only resize of LVM, creation of new primary partition, setup grub chainboot and I got to this panic ;b
<Sleep_Walker>it's too quick
<Sleep_Walker>I'll retry to collect it
<civodul>could you post the changes you made for LVM support?
<Sleep_Walker>I gave up in the end
<Sleep_Walker>I prepared initrd with LVM2 support, but failed to initialize device mapper
<civodul>that sounded quite ambitious ;-)
<civodul>did you used 'mapped-device' for that?
<Sleep_Walker>I didn't use scheme for that
<Sleep_Walker>I copied busybox as well :b
<civodul>that's cheating :-)
<Sleep_Walker>first guix, then hacking in guile
<th3kent>bug report submitted.
<rgrau_>I saw the bugfix in , should I apply the patch directly in the grub.scm that's currently in my profile? or it will get erased when running reconfigure?
<rgrau_>I mean, what's the way to apply guix patches in a running system?
<alezost>rgrau_: you may just "guix pull"
<Sleep_Walker>does this sound familiar?
<civodul>rgrau_: to test it, you would need a checkout for Guix, on top of which you would apply the patch
<civodul>Sleep_Walker: nope
<Sleep_Walker>I couldn't collect it on serial because of having `console=ttyS0 console=tty0'
<Sleep_Walker> this is my log
<Sleep_Walker>but as `system init' died on grub error, it may not be prepared correctly
<Sleep_Walker>ok, the warning I pasted is because I'm putting grub outside MBR and it doesn't have room to fit
<civodul>that's weird: it says "entering a prompt" and then panics without giving a prompt
<Sleep_Walker>I edited install-grub to contain "--force" and going to retry
<Sleep_Walker>with --force it works in my case
<Sleep_Walker>hard to say how to implement generally :)
<civodul>it looks like the 'tcech' user account is part of a group that doesn't exist
<civodul>probably the 'user-account' declaration is incorrect
<tadni`>If I was going to make a patch for davexunit's use-package-modules and use-services-modules -- where would be a good place to throw this?
<civodul>Sleep_Walker: re --force, that's an interesting approach, indeed :-)
<civodul>tadni`: i would put these two macros in the (gnu) module
<Sleep_Walker>civodul: well, it seems that GRUB authors wanted to discourage such use with strong warning...
<Sleep_Walker>but I haven't found way how to do it in a clean way
<Sleep_Walker>they'd like to reserve some disk space in partition before filesystem
<Sleep_Walker>and there is room for that in MBR, but not sda4 (I have that grub loading grub)
<tadni`>Wait, is there a formal module for GNU -- or just throw it into the general /gnu directory somewhere?
<tadni`>I mean, I was thinking of just throwing use-package-modules at the bottom of /gnu/packages.scm and use-service-modules, respecitvely in /gnu/services.scm.
<tadni`>Is this not ideal?
<Sleep_Walker>civodul: I was inspired by example in
<civodul>tadni`: (gnu) is the gnu.scm file
<Sleep_Walker>maybe there should be some '(groups ' section as well?
<civodul>that example is outdated
<tadni`>civodul: Ah, silly me.
*tadni` always has an ant up his bum to do these things, when he's too tired to be reasonably doing these things...
<tadni`>(gnu) is pretty barren.
<civodul>Sleep_Walker: actually, it's not outdated, but it's misleading: there's no guarantee in this example that GID 100 is valid
<tadni`>Should I add a "use-system-modules" too? Would this be handy/worth the effort?
<civodul>Sleep_Walker: so the solution would be to use a group name instead, such as "users"
<civodul>tadni`: i'd say no
<davexunit>hello #guix
<civodul>hello, davexunit!