IRC Logs for #crux-devel Thursday, 2014-02-06

nrxtxjaeger: but you should keep the information somewhere especially on complex ports, e.g. for contacting on problems, initial commit message might be a good place00:03
jaegerIn what situation would you prefer to contact the Packager rather than the maintainer>?00:04
nrxtxdid i got this right that packager is the one originally wrote a package?00:05
jaegeryes, and the Maintainer is the person who keeps it working. If the maintainer is different that means the packager no longer works on that port00:05
nrxtxmaybe a maintainer has a backwards question or someone took over a maonated package from another maintainer00:05
nrxtxjust thought about it, don't know if this case will ever occur00:06
nrxtxbut once you dropped the information completely there is no back00:06
prologicI do kinda agree with jeager here00:07
jaegerIt's something I'd prefer to have in a metadata section of a package database if we had that sort of thing00:07
prologicoften you take over a port that someone else maintained00:07
prologicand fix it, update it, etc00:07
prologicso you end up knowing just a smuch about the port/software that the original maintainer did00:07
prologicre latest test iso at
prologicI"ll test this out as well00:13
prologicjaeger: for some reason the USB image is unreadble01:36
prologicany reason for that?01:36
prologicI've dd'd it to a usb stick and it's unreadable01:36
jaegerhave a different usb drive to test? It was ok for me01:37
jaegeruse any odd dd options?01:38
prologicjust straight dd if= of=01:40
prologicnrxtx's built iso worked fine01:40
jaegeryou had that issue with one of the previous ones too, right?01:50
jaegerI just tested it again, works fine here01:50
jaegerUsing dd if=blah.iso of=/dev/sdc bs=16M01:50
prologicyes I wasn't able to bootup or mount your previous build either01:56
prologicI'll try again with bs=16M01:57
prologicand I'll confirm the md5sum matches01:57
prologic$ sudo dd if=crux-3.0-20140205.iso of=/dev/disk3 bs=16m02:02
prologic18+1 records in02:02
prologic18+1 records out02:02
prologic314572800 bytes transferred in 48.188680 secs (6527940 bytes/sec)02:02
prologicOS X can't read the resulting USB disk02:02
prologicI'll have to try at home on my desktop02:02
prologicobviously I"m at work (OS X at work)02:03
jaegerwonder if disk utility would do any better02:04
prologicnormally dd works just fine on OS X02:21
prologicI'll try at home to confirm02:21
*** mavrick61 has quit IRC03:49
*** mavrick61 has joined #crux-devel03:50
Romster<nrxtx> "Packager" is kind of confusing < how so? it's the person that made the package/port to begin with. maintainer is the current maintainer07:24
Romsterblock size wont mean much other than hdd buffering speed.07:31
Romsteronly time it means anything is on cd-rom/dvd reads bs=204807:32
*** deus_ex has quit IRC08:50
*** deus_ex has joined #crux-devel08:55
*** deus_ex has quit IRC09:07
*** deus_ex has joined #crux-devel09:09
frinnstRomster: yeah but "packager" usually has other meanins in the linux distro world09:44
frinnstsince most distros are shipping binary *packages*09:45
frinnstI'd rather just get rid of the whole packager line altogether09:45
frinnstvanity is a sin anyways :)09:45
Romsteri wouldn't mind a contributor line(s) and a group of users pitch in.10:05
Romsterper a port for those that use that port. even if its only one that has commit access to maintain it.10:05
Romsterthis whole model of port hording is not working for the majority of maintainers.10:06
teK_frinnst: +110:37
*** deus_ex has quit IRC10:43
*** deus_ex has joined #crux-devel10:44
prologicyeah nah11:12
prologicfrinnst, +111:12
prologiclet's nuke the Packager line11:12
*** deus_ex has quit IRC11:30
*** deus_ex has joined #crux-devel11:42
teK_=======> ERROR: Downloading '' failed.11:53
teK_=======> ERROR: Md5sum mismatch found:11:54
teK_MISSING   26ea806689426797f4a9b50f75882ef7  docker.rc11:54
teK_NEW       966e1916d611427c44686ad09145996f  docker.conf11:54
teK_NEW       dcaae2f852c97ba73436c77cda3eb4e3  docker.rc11:54
teK_=======> ERROR: Building '/usr/ports/packages/docker#0.8.0-1.pkg.tar.gz' failed.11:54
prologicPyQt is mine?11:56
prologicoh :)11:57
prologicthought you were talking about contrib11:57
prologicI haven't added anything to contirb yet :)11:57
prologicI'll fix these two up right now11:57
teK_at present, btrfs(progs) is a non-optional requirement. As it is in core that's fine.. you _may_ look into making it optional, imho12:00
teK_another thing:12:01
teK_=======> New files found:12:01
teK_NEW       -rw-r--r--      root/root       etc/docker.conf12:01
teK_=======> Building '/usr/ports/packages/docker#0.8.0-1.pkg.tar.gz' succeeded.12:01
teK_woot? :)12:01
teK_and your kernel-config-script might a) check /proc/config.gz of no arguments are given (instead of sitting there doing nothing forever) and b) might be readjusted to test for iptables/NAT support:12:04
teK_Unable to enable network bridge NAT: iptables failed: iptables -I POSTROUTING -t nat -s ! -d -j MASQUERADE: iptables v1.4.21: can't initialize iptables table `nat': Table does not exist (do you need to insmod?)12:04
Romsterdocker md5sum error12:05
Romsteri'm building it too.12:05
teK_Romster is sloooooow12:05
teK_hi there!12:05
Romsterhey i just got home i was walking the dogs12:05
teK_no excuses!12:06
Romsterso what's with these new email alias?12:06
Romsterand is that all the active maintainers in contrib now?12:06
teK_thought they might come in handy12:06
teK_I think so.12:06
Romsterso the rest are orphans12:07
teK_I think so. You might recheck that by using git log12:07
Romsteryeah be a good diea.12:07
teK_last(1) did not provide any useful data ecause the git shell won't update utmp :\12:08
prologicyeah ok12:08
prologicall good improvements12:08
prologicso yeah make sure you reconfigure your kernel12:08
teK_btw Romster, do you happen to know if CONFIG_IP_NF_IPTABLES includes the NAT table?12:08
prologicor the daemon will not work12:08
Romsteralso why README.rst and not just README?12:08
prologicbecause it renders nicely on bitbucket12:09
prologicwhere I host my ports :)12:09
prologicI will probably also keep a clone12:09
Romsterbut it breaks on prt-get readme foo12:09
prologicI assume prt-get readme still works?12:09
prologicdoes it?12:09
prologicwe should fix prt-get12:09
Romsternot checked i assume it wound't12:09
prologicteach it some new tricks12:09
prologicrendering md/rst12:09
Romsterwould not12:09
teK_haha :D12:09
prologictry it? :)12:09
teK_never gonna happen12:09
teK_i'm too stupid to enable NAT12:10
Romsterprt-get readme docker no results12:10
Romsterprt-get cat docker README.rst works12:10
prologicyeah ok12:11
prologicI'll rename to just README then12:11
Romsterone might add that prt-get readme is rendundant since prt-get cat foo README also does the same thing.12:11
prologicdoesn't amtter12:12
prologicfor now I'll comply12:12
prologicit would be nice to teach prt-get how to render Markdown/reStructuedText though :)12:12
prologicbut that's a story for another day12:12
prologicteK_, Romster try docker again12:12
prologicyou'll have to ports -u prologic obviously :)12:12
prologicteK_, and fixed pyqt/sip as well12:12
RomsterteK_, Depends on: NET [=y] && INET [=y] && NETFILTER [=n]12:12
Romsterso i guess not.12:13
teK_I have Depends on: NET [=y] && INET [=y] && NETFILTER [=y]12:14
Romster../src/ undefined: CDmTask12:15
Romsteri guess i need to fix my kenrel :D12:15
Romsterbe nice to use docker instead of chroots for my port testing here.12:15
prologicI'm actually testing two ports I'm putting into contrib12:17
prologicvia docker :)12:17
prologicdocker buidl -rm .12:17
prologicin my ~/contrib'12:17
prologicteK_, how are you still getting mismatches for thse12:18
nrxtxRomster: you can also use plan lxc without docker12:18
Romster"Docker comes with three main Storage Backens:" <- typo Backends:12:18
Romsterin README.rst12:18
nrxtxdocker is mainly some kind of frontend/distribution system12:19
Romsternrxtx, i was wondering that myself.12:19
prologicI get a clean build here12:19
prologicThanks Romster :)12:19
nrxtxi already use lxc, which works quite nice instead of chroot12:19
Romsterbut i wanna use docker on a host for versionsort.com12:19
Romsternrxtx, got examples?12:19
prologicI know why12:20
nrxtxsimply check "how to use" for command line or automated via python12:20
prologicmy httpup repo is not live12:20
prologicit's only synced once per day12:21
prologicjust a moment12:21
nrxtxRomster: docker just sits on top of this commands12:21
prologictry now guys12:21
prologicsorry about that12:21
nrxtxthey wrote some go bindings for liblxc and use that12:21
Romsternrxtx, i thought lxc was the brains of it all.12:22
Romsterwhy on earth is it written in go though12:22
Romsterfor docker12:22
prologicnrxtx, yeah Docker is just a nicer way of using lxc IHMO12:22
prologicit's process-oriented12:22
prologicwith a really nice UNIX-ish API/CLI12:22
prologicrun, rm, rmi, top, start, stop, kill12:23
nrxtxprologic: and docker provides image/container distribution12:23
prologicit's quite nice12:23
prologica really really nice ecosystem12:23
prologicthat's buitl up quite quickly12:23
prologicincluding a public image registry12:23
nrxtxRomster: go got quite popular in different areas12:23
prologic/exec -o docker run crux echo 'hello world'12:24
prologichello world12:24
prologiceven works from X-Chat :)12:24
RomsterCONFIG_BLK_CGROUP says it's option on the test but the cgroup mount is required. that's confusing.12:25
prologicthis is seriously a nice way to test dependency problems with your own ports12:25
Romsterprologic, i already do it in a chroot.12:25
prologicwhat version kernel are you using Romster?12:25
prologicRemember to use 3.8+12:25
Romsterwhy do you think the majority of mine are working and ones i have tested i report.12:25
prologicProbably recommend 3.13 personally12:25
Romsterah older kernel12:25
prologicyeah I'm saying donig it via Docker is even nice rhtn a chroot12:26
prologicit "just works"12:26
prologicI didn't have to write any complicated bash scripts12:26
prologicor bind mounts12:26
prologicetc: )12:26
Romsterbut for a desktop machine even jsut lxc be enough12:26
nrxtxRomster: that's the way i use it, plain lxc12:26
prologicdon't diss it till you try it :)12:26
prologicyou can't brea: docker run crux echo 'hello crux'12:27
prologiceven if you *could*12:27
prologicyou'd still ahve to go build the lxc container yourself12:27
prologicbefore you even started :)12:27
Romsteryeah does look more effort than using docker12:27
prologicI officially maintain the "crux" Docker image in the Docker standard library (as it's called)12:28
Romsterseen that12:28
Romsterircnotifyer why do a ports -u to only prt-get depinst python12:28
prologic$ docker run -p 8000:8000 dc7c18f9cd78 irc.freenode.net12:28
prologicUsing circuits-dev serial12:28
prologiccircuits.web/ ready! Listening on:
prologic10.0.0.1 - - [06/Feb/2014:11:49:58] "POST /freenode-proxy-checker.txt HTTP/1.0" 404 629 "" ""12:28
Romsterwhy not a ports -u opt12:28
prologic^^^ running my ircnotifier bot :)12:28
prologicthat does bitbucket/mercurial commit notifications of many of my proejcts12:29
Romsterheck i'm suprised you don't have python in the base image.12:29
Romsterconsidering everything you do is in python.12:29
prologicyes I will probably update the ircnotifier Dockerfile12:29
prologicto only update the opt repo12:29
prologicyou're quite right12:29
prologicbut really12:29
prologicI should build a prologic/crux-python image12:29
nrxtxprologic: you can load the templates with lxc12:29
prologicand based ircnotifier's off that12:29
Romsteri'm just nit picking here though so don't mind me.12:30
prologicRomster, I wanted to keep the "crux" base image12:30
prologicin the std. lib clean12:30
prologici.e: as if it were instaleld from iso12:30
prologic(which it is)12:30
prologicand only "core" ports12:30
Romsterand i would love to see how you made that base archive.12:30
Romsteri'd like to make a crux 3.1 testing one.12:30
prologicit's all there12:30
Romsterrootfs.tar.xz is on ?12:31
prologicit gets built by build.sh12:31
prologicit relies on the ISO12:31
prologicso rebuild a new ISO12:31 makes rootfs.tar.xz12:31
prologic./ /path/to/crux-3.1.iso12:32
prologicwell yeah12:32 just wraps it all up12:32
prologicdownlaods the latest crux-3.0.iso12:32
Romsterhmm thought you may prune core a bit more.12:32
Romsterthough it's here i can make a 3.1 from it.12:33
Romstercurrently got chroots everywhere.12:33
Romstereven a i686 one.12:33
Romsterdocker could replace them all :D12:33
prologicabout to push this to contrib12:34
prologicall okay? :)12:34
Romsterx in docker with DISPLAY works to?12:34
Romsteri can't fully test it until i redo my kernel12:34
prologicI'm going to prune it a bit more12:34
prologicbut probably not include it in the Docker std. lib12:34
prologicjust in my own private prologic/ repo12:34
nrxtxRomster: if you want plain lxc just install lxc and have a look into /usr/share/lxc/templates12:34
prologicas ucrux, ncrux and pcrux12:34
prologicmicro, nano, pico12:34
Romsternice idea12:35
prologicbut it will be built by a script12:35
prologicso they can be reliably rebuilt12:35
prologicI already did this for my openvz images12:35
prologicso I already have the tools in place to do this12:35
Romsteri had another idea (yes i know you are getting sick of my ideas by now)12:35
prologicYes X/Xorg will run in a Docker container12:36
prologicdocker -privelged ...12:36
prologicdocker run -privelged ...12:36
prologicsorry 2nd one :)12:36
Romstera standared command line interface to configure crux with a constantant interface like "crux set|add foo filename"12:36
prologicsed is good enough :)12:36
Romsterinstead of all the seds and echos everywhere crux could have a standard way to configure base stuff.12:37
prologicwell maybe12:37
prologicbut it's hardly necessary12:37
prologichard work for little gain12:37
Romsterthen anyone can save/reuse a script of settings to configure crux to there settings.12:37
Romsteri guess so it was an /idea/ i got.12:37
Romsteri never say they are good ideas.12:38
prologicoh no it's a good idea12:38
prologicjust not worth it :)12:38
prologiceffort vs. cost vs. time12:38
Romsterbut with a crux comamnd line setting program it can also have a testsuite to make sure it's doing as it's meant too.12:38
prologicsed does the job12:38
prologicecho as well12:38
prologicdo you guys have a convention for commit messages to contrib/12:39
prologicor don't care as much?12:39
Romsteri have my own format.12:39
prologicso do i :)12:39
Romsterbut it's usually foo: some message12:39
Romsterlook though the git log12:39
nrxtxRomster: if you learn go you can automate everything from pkgfile build to test running it :D12:40
prologicin case anyone wanted to try/use this as well12:41
prologicI build my contrib port12:41
prologicupdate it, or whatever12:41
prologicand run docker build -rm .12:41
prologicdocker run <imageid> depinst blah12:41
Romsternice to know12:42
Romsterstill messign on my sql and now more filter rules it's around 500 lines of rules so far.12:42
prologicSo Docker ports?12:44
prologicall good, etc?12:44
prologicOr you guys still configuring kernsls and testing? :)12:44
Romsteri haven't started to configure kernel i got too much stuff open to bother rebooting now.12:44
Romsterweekend i'll do it12:45
Romsterand i got springs rear shhock assorbers and exhorst and extractors to fit to my car this weekend as well and it's gona be too hot saturday12:45
Romsterso it'll be a sunday and onwards probably get it done by wednesday and i'll just use the work van until i get that done.12:46
Romsterand to top it off my windscreen cracked 2 days ago... so another job when i can afford that.12:46
prologicteK_, what did you say about improving the script?12:47
prologicuse /proc/config.gz if no args are given12:47
prologicand it exists12:47
prologicwhat else?12:47
Romsterover worked under paied fixing up a shitbox wagon and trying to do all the programming in the world.12:47
teK_prologic: [[ if $# -lt 1 ]]; then echo usage: $0 kernel.config ; exit 1 ; fi12:48
prologicyeah ok :)12:49
Romster something like in there for finding the .config i guess prologic12:49
Romsteri personally use /proc/config./gz for most things.12:50
Romsteri personally use /proc/config.gz for most things.12:50
Romsterbut not everyone enables that which is silly.12:50
teK_and the if outside the [[ ]] ;)12:50
Romsterlol nice one teK_12:51
prologicjust upeated and push changes to that12:55 is much nicer now12:55
Romsterand making it u+x12:55
Romsterdoes git save permissions or even ports -u?12:56
prologicit should12:56
Romsteranoying when i have to u+x allt eh sh scripts and pre-post to run them manually.12:56
prologicdoes here12:57
teK_it doesnt.....12:57
Romsternot good12:57
teK_rm -rf /usr/ports/prologic ; umask 077 ; ports -u prologic12:57
teK_-> everything is 6/70012:57
prologicit does here12:58
Romsterguess that's a security thing to avoid world writable files or something i dunno can't be bothered to test.12:59
Romsteri need to go afk and relax.12:59
teK_have fun12:59
teK_ports is an alias to sudo ports..13:04
prologicoh shit13:07
prologicdid I commit that with root ownerships?13:07
prologicnope looks okay here13:08
prologictry ports -u prologic again?13:08
teK_it's got nothing todo  with your repo..13:09
prologicsorry then what are you showing me?13:09
teK_the misery in my life =)13:11
prologicI don't get what you're getting at :)13:11
prologic$ docker ps13:12
prologicCONTAINER ID        IMAGE                         COMMAND                CREATED             STATUS              PORTS                    NAMES13:12
prologic775febfafe28        prologic/ircnotifier:latest   /ircnotifier/start i   10 minutes ago      Up 10 minutes>8000/tcp   sleepy_fermi13:12
prologiclove it13:12
nrxtxcan docker also as possible with lxc directly overlay the current used rootfs?13:39
nrxtxis interesting point for testing ports13:39
prologicI'm using it to test ports and build the iso at present13:40
prologicbut I"m also starting to more and more seriously use it in production13:40
prologicfor running up services, web applications, etc13:40
nrxtxprologic: mounting the current rootfs as read only into a docker container?13:43
nrxtxnot creating a new one13:44
prologicyeah I guess you could do it13:44
nrxtxhaving a disposable write layer on top13:44
prologicnot sure :)13:44
prologicoh wait13:44
prologicthey are disposable anyway13:44
prologicby default13:44
prologicDocker uses a layered file system13:44
prologicone of aufs, devmapper or btrfs13:44
prologicthe container is created as a read-only layer on top of the base image13:45
prologicchanges to the image do not persist until you docker commit13:45
prologicbut there is this concept of volumnes too13:45
prologicyou can mount host paths as volumnes in a container13:45
prologicand specify paths in a container to be volumnes13:45
prologicand reference them from other containers13:46
Romsteri see thinp provisioning so i guess lvm can be used to with thinp napshot to write changes to a cow iamge13:55
*** Roomster has joined #crux-devel14:02
*** Roomster has quit IRC14:03
*** Roomster has joined #crux-devel14:03
nrxtxprologic: the concept is there but can be the base image the current rootfs of the "host" ?14:05
prologiccan the base image of a container be the rootfs of the host?14:12
prologicin theory yes14:12
prologicthough I'm not sure why you'd want to run your "host" as a container14:13
prologicyou could write a Docker file like this14:13
prologic(and then run it)14:13
prologicFROM scratch14:13
prologicADD . /14:13
prologicand put this at /Dockerfile on your root filesystem14:13
prologicdocker build -rm .14:13
prologicdocker run -i -t image-id /bin/bash14:14
prologicand it would be exactly what your desktop is I guess14:14
prologicbut why? :)14:14
prologicwhy not just use my crux image :)14:14
nrxtxprologic: i do not want to create a base image, just use the current filesystem as is mounted read only, no copying or something else14:39
*** Roomster has quit IRC14:39
prologicI wonder if14:43
prologicdocker run -v / scratch /bin/bash14:43
prologicworks :)14:43
prologic$ docker run -v / scratch /bin/bash14:44
prologic2014/02/07 00:44:04 Invalid volume: path can't be '/'14:44
nrxtxprologic: it would be the fastes way of testing result of updating single ports :)14:50
jaegerRomster: block size can cause issues on really shitty USB controllers but usually that's if you go larger, not smaller15:05
prologicnrxtx, actually no it wouldn't20:14
prologicI think you're missing something important here :)20:14
prologicbind mounting / (of of the host) to a new container20:14
prologicisn't all that useful for testing port dependencies20:14
prologicbecause you're not necessarily in a "clean" environment20:14
prologicwhere you may not necessarily pickup missing depneiceis20:14
prologicor footprint disrepencies20:14
nrxtxprologic: ok thjat is a good point20:39
nrxtxsaw the docker release 0.8?20:40
nrxtxah yes already up to date :)20:47
Romsterjaeger> Romster: block size can cause issues on really shitty USB controllers but usually that's if you go larger, not smaller <- that's good to know.21:57
jaegerI haven't seen it on USB controllers made in the last few years at least... might be a thing of the past22:00
jaegerI just do 16M as it seems a decent performance point22:00
jaegerRomster: regarding FS#982 that is intentional22:27
jaegerBecause you are supposed to accept a license or similar before downloading them22:28
jaegerheh... booted the ISO in hyper-v23:25
jaegerwacky... if you load ata_piix in a hyper-v machine it bows out of the way for hv_storsvc23:44
jaegerer, hv_storvsc23:44
jaegerwhich explains why /dev/sda appeared after boot finished the first time23:45

Generated by 2.11.0 by Marius Gedminas - find it at!