*** tilman has quit IRC | 01:03 | |
*** tilman has joined #crux | 01:05 | |
*** kls has quit IRC | 01:12 | |
*** doomicide has quit IRC | 02:09 | |
*** xeirrr has joined #crux | 02:20 | |
*** xeirrr has left #crux ("Leaving") | 02:41 | |
Romster | except prt-get sorts ports dependencies before sysup | 03:00 |
---|---|---|
Romster | what it doesn't do is inject new dependencies | 03:01 |
Romster | in such cases its better to just use prt-get update -fr `revdep` | 03:01 |
*** mavrick61 has quit IRC | 03:26 | |
*** mavrick61 has joined #crux | 03:27 | |
*** Romster has quit IRC | 04:35 | |
*** Romster has joined #crux | 04:35 | |
*** diverse has quit IRC | 04:42 | |
rmull | I have a gigantic pile of disks on a gentoo fileserver - two md raid arrays that are joined with lvm. If I remove gentoo and install crux on this machine, is it probable that the md arrays and the lvm volume will remain intact as if nothing happened? | 04:53 |
rmull | The arrays and everything are pretty old, from the 2.6 days | 04:53 |
Romster | it is lvm2? and what mdadm metadata version? | 04:57 |
Romster | i have multiple mdadm arrays in the same VG on lvm2 here. | 04:58 |
Romster | rmull, ^ | 04:58 |
rmull | Yes, lvm2. Not sure the md version | 05:00 |
rmull | I'm sure it'll "just work," I'll tackle it sometime soon | 05:00 |
rmull | thanks Romster | 05:01 |
Romster | mdadm -D /dev/mdx | 05:02 |
Romster | or --detail | 05:02 |
Romster | if its superblock 0.9 1.1 or 1.2 your good. | 05:03 |
Romster | just save the contents of /etc/lvm/ just in case and mdadm.conf | 05:05 |
*** Feksclaus has quit IRC | 05:21 | |
*** Romster has quit IRC | 05:26 | |
*** Romster has joined #crux | 05:26 | |
jaeger | rmull: yes, it should theoretically work. Haven't tried that myself, though | 05:35 |
*** Romster has quit IRC | 05:38 | |
*** Romster has joined #crux | 05:38 | |
Romster | providing its at the versions i mentioned it'll assemble and activate. | 05:56 |
Romster | jaeger, hmm i've been meaning to have lvm2 cryptosetup and mdadm on the crux iso too. | 05:56 |
Romster | so one can assemble unlock and activate to get to files needed for a setup or root on lvm/luks/mdadm | 05:57 |
*** lnds has joined #crux | 07:26 | |
*** polygraphed has quit IRC | 07:40 | |
Romster | prologic, what causes this https://gist.github.com/therealromster/db9675fe2441edacbcb2 | 08:12 |
Romster | happend after i did a kernel update 3.17.3 -> 3.17.7 and i used make silentoldconfig | 08:13 |
Romster | https://github.com/tianon/docker-overlay/issues/35 | 08:26 |
Romster | that's what i'm getting | 08:27 |
Romster | POSIX_MQUEUE | 08:27 |
Romster | only added to the check script 7 days ago | 08:28 |
Romster | prologic, https://github.com/docker/docker/pull/9823 | 08:28 |
*** Romster has quit IRC | 08:42 | |
*** Romster has joined #crux | 09:00 | |
*** lnds has quit IRC | 09:31 | |
prologic | Romster, new kernel option dependency? | 09:34 |
prologic | must be for upcoming new release? | 09:34 |
prologic | Romster, weird I don't get that here | 09:35 |
prologic | are you running a newer Docker to me? | 09:35 |
Romster | something with new docker and libcontainer | 09:36 |
Romster | no same as in contrib. | 09:37 |
prologic | oh | 09:37 |
Romster | i had to enable that MQUEUE | 09:37 |
prologic | yeah that's why it's curious | 09:37 |
Romster | now it works | 09:37 |
prologic | I don't get that issue here | 09:37 |
prologic | hmm | 09:37 |
Romster | you probably have it on already | 09:37 |
prologic | lemme see if I have MQUEUE enabled by coincidence | 09:37 |
Romster | i didn't | 09:37 |
prologic | $ zgrep MQUEUE /proc/config.gz | 09:37 |
prologic | # CONFIG_POSIX_MQUEUE is not set | 09:37 |
prologic | nope | 09:38 |
Romster | wtf | 09:38 |
prologic | $ dki crux crux | 09:38 |
prologic | CRUX version 3.1 | 09:38 |
Romster | well you seen my gist | 09:38 |
prologic | works fine here | 09:38 |
prologic | I know :) | 09:38 |
Romster | docker version 1.4.0 ? | 09:38 |
prologic | it's not that I don't believe you | 09:38 |
prologic | it's just that I don't believe you :) | 09:38 |
prologic | ahh no | 09:38 |
prologic | 1.3.2 here | 09:38 |
prologic | wtf | 09:38 |
Romster | i'm totlly sysuped btw | 09:38 |
Romster | there you go it's needed on 1.4.0 | 09:38 |
prologic | I haven't upgraded my local docker yet | 09:38 |
prologic | haha | 09:38 |
prologic | :) | 09:38 |
prologic | ahh | 09:39 |
prologic | cool | 09:39 |
prologic | thanks for the heads up :) | 09:39 |
Romster | no problem | 09:39 |
prologic | been without internet for 3 days | 09:39 |
prologic | churning to internode | 09:39 |
Romster | oh | 09:39 |
prologic | not quite there yet but soon hopefully | 09:39 |
prologic | will hopefully have a /29 IPv4 and /56 IPv6 for home | 09:39 |
Romster | uped my base size and learned because a iamge is snapped shotted off the base image if you want that image bigger they all need to be removed off that. and downloaded again on new snapshots. | 09:40 |
Romster | blowing away all of docker containers is just stupid. | 09:40 |
Romster | jsut to increase that limit as i read on a blog | 09:41 |
Romster | really that many ipv4 at home? | 09:41 |
Romster | jsut go with ipv6? | 09:41 |
Romster | so your hosting at home on adsl2+ or something | 09:42 |
Romster | i need some docker CI magic so i can build crux ports. | 09:43 |
Romster | i never worked on such a beast. | 09:43 |
prologic | that's what I'm being given :) | 09:51 |
prologic | the smallest block they can give me really | 09:51 |
prologic | I plan to for some things | 09:51 |
prologic | mostly for the family, etc | 09:51 |
prologic | yeah I'm a little behind on some of my proejcts :) | 09:52 |
prologic | been busy building cgod :) | 09:52 |
prologic | https://bitbucket.org/prologic/cgod | 09:52 |
Romster | pointless side project i thought you were working on docker circuits web thing... | 09:53 |
Romster | you trying to revive gropher even more. | 09:54 |
prologic | pointless? hmm | 09:57 |
prologic | I hardly think so :) | 09:57 |
prologic | I find it useful in many ways really | 09:57 |
prologic | but yeah still working on docker, circuits, autodock and others :) | 09:57 |
prologic | you know how OSS goes :) | 09:57 |
Romster | yes all too well | 09:59 |
Romster | eh gropher feels like bbs | 10:00 |
prologic | hmm | 10:00 |
Romster | next you'll want a dial in bbs | 10:00 |
prologic | I think bbs is different | 10:00 |
prologic | but many bbs's were implemented on top of Gopher I think | 10:00 |
prologic | not sure :) | 10:00 |
prologic | nah | 10:00 |
prologic | in any case | 10:01 |
prologic | I'm finding Gopher convenient | 10:01 |
prologic | I don't have to mess around with a stupid annoying UI | 10:01 |
prologic | UI(s) are tedious at best to get right | 10:01 |
Romster | i wouldn't mind setting up a mesh network but not much point when people around me have nfi | 10:01 |
Romster | so dillo/lynx isn't much use? | 10:02 |
prologic | sure they are | 10:02 |
prologic | in fact I find elinks quite a useful browser | 10:02 |
prologic | I don't like lynx much myself | 10:02 |
prologic | bit clunky | 10:02 |
prologic | install Overbite for FF and check out my Gopherspace :) | 10:03 |
prologic | gopher://daisy.shortcircuit.net.au/ | 10:03 |
Romster | had already | 10:03 |
Romster | now if only hexchat knew how to link gopher:// | 10:04 |
prologic | just as an example; one thing that's really nice and handy (sure I wrote scrapers that do the trick) is that I can view xkcd comics and read the little blurb without the annoyance of having to do the whole dance; "Click xkcd entry in RSS; Change URL to m.xkcd.com; Click image; Read blurb" | 10:05 |
prologic | because normally the blurb is a tooltip which I can't view because of my zooming into the screen, etc so I only see/read a small portion of the blurb and it cuts off and if you move it goes away (nature of tooltips) | 10:05 |
Romster | ah yeah tooltips suck on zooming in | 10:06 |
prologic | yeap sure do | 10:06 |
Romster | so do most GUIs like say on fontsize | 10:06 |
prologic | that's just one example of the convenience I'm talking about | 10:06 |
prologic | but also the fact that I can present resources adn content without worring about annoying CSS/HTML/JS and rendering issues (not to mention browser incompatibilities) is wonderful :) | 10:07 |
Romster | http://xkcd.com/1466/ haha sites down i'll keep refreshing and dos it some more in the process | 10:08 |
Romster | ok so it might be good for craping data and for nearly blind people | 10:09 |
Romster | only if sites were made in such a way as meada type worked | 10:09 |
prologic | well see now here's the thing | 10:09 |
prologic | IHMO the Web is broken :) | 10:10 |
Romster | yes you keep saying that | 10:10 |
prologic | Yes I'm a Web Developer :) | 10:10 |
prologic | It was originally designed to share resources | 10:10 |
prologic | but it got a bit abused over the years :) | 10:10 |
prologic | and quickly commercialized | 10:10 |
prologic | but what ya gonna do :) | 10:10 |
prologic | you should see the new umart website | 10:10 |
Romster | improve on the technology | 10:10 |
prologic | it's awful | 10:10 |
prologic | it doesn't even work | 10:10 |
prologic | the technology was fine | 10:11 |
prologic | it's the people that's the problem :) | 10:11 |
prologic | haha | 10:11 |
teK_ | gg..gopher? | 10:11 |
prologic | i.e: standard JS implementation | 10:11 |
prologic | standard rendering standards | 10:11 |
teK_ | folks.. :p | 10:11 |
prologic | blah balhblah | 10:11 |
prologic | hi teK_ :) | 10:11 |
teK_ | hello! | 10:11 |
Romster | oh its doing DOM stuff in javascript eh? | 10:11 |
Romster | hey tek | 10:11 |
prologic | anyway look at IRC | 10:12 |
prologic | it's older then Gopher and still going strong | 10:12 |
prologic | :) | 10:12 |
prologic | so *meh* | 10:12 |
Romster | yeah but IRC is simple | 10:12 |
prologic | and every tom dick and harry tried to bastardize it | 10:12 |
prologic | MSN *couch*, ICQ, etc | 10:12 |
Romster | its even used on web based chats | 10:12 |
prologic | and now most seem to use some kind of XMPP | 10:13 |
prologic | precisely | 10:13 |
prologic | it's the simplest things that last :) | 10:13 |
prologic | IHMO | 10:13 |
prologic | we have to do *so much* to make "websites" performant these aysd | 10:13 |
prologic | what with CDN(s), uglifying JS/CSS | 10:13 |
prologic | compression, caching | 10:13 |
Romster | the only thing broken on the web is the programmers that do brain dead stupid stuff. the tools are fine. | 10:13 |
prologic | etc etc | 10:13 |
prologic | and it keeps getting worse | 10:14 |
Romster | how can you stuff up irc? | 10:14 |
*** xeirrr has joined #crux | 10:14 | |
prologic | well :) | 10:14 |
prologic | you can't change people | 10:14 |
prologic | :) | 10:14 |
prologic | oh MSN tried :) | 10:14 |
prologic | MSN was an extension to IRC | 10:14 |
Romster | so restrict what they can do... ie gropher | 10:14 |
prologic | at least the underyling protocol behind it was | 10:14 |
prologic | see IHMO Gopher is all about resources | 10:14 |
prologic | nothing more nothing less | 10:14 |
prologic | it has the simplest UI/presentation | 10:15 |
prologic | that is consistent across all Gopherspaces | 10:15 |
prologic | no matter where you go | 10:15 |
prologic | it's awesome :) | 10:15 |
Romster | only reason MSN got shut down was they got skype | 10:15 |
Romster | so a gropher crux port browser is next << | 10:16 |
prologic | well I ported gopher | 10:16 |
prologic | the original UWN client | 10:17 |
prologic | but I didn't really like it :) | 10:17 |
prologic | but lynx and elinks are there already | 10:17 |
prologic | lynx supports it without recompiling | 10:17 |
prologic | so does curl | 10:17 |
prologic | elinks just needs --enable-gopher | 10:17 |
prologic | tilman, no:) | 10:17 |
prologic | far from it :) | 10:17 |
tilman | (i hate web 2.x shit as well) | 10:17 |
prologic | backend web developer having revenge :) | 10:17 |
prologic | there's a reason I stick to "backend" :) | 10:17 |
tilman | i was made to write an "app" with flask last year | 10:18 |
prologic | of any kind | 10:18 |
prologic | oh lovely | 10:18 |
tilman | kinda liked that lib | 10:18 |
prologic | how'd that experience go? | 10:18 |
tilman | pretty well | 10:18 |
prologic | not that Flask isn't all that bad btw | 10:18 |
tilman | the sql code i wrote for that sucks ass though | 10:18 |
prologic | ahh yes | 10:18 |
prologic | rdbms | 10:18 |
prologic | bane of my existence | 10:19 |
tilman | :)) | 10:19 |
Romster | databases | 10:19 |
prologic | normally I just use files | 10:19 |
tilman | ! | 10:19 |
prologic | databases are typically overkill | 10:19 |
Romster | -_- | 10:19 |
prologic | Romster, you'll like cgod btw | 10:19 |
prologic | if you run up your own Gopherspace | 10:19 |
prologic | it has a very nice CGi scripting engine | 10:19 |
prologic | you can write CGI(s) in any lang you want | 10:19 |
prologic | but that's not the nice part | 10:20 |
Romster | can't i do that now on apache and anything else. | 10:20 |
prologic | the nice part is the interface to the session manager for storing persistent data | 10:20 |
tilman | lol, curl supports gopher by default | 10:20 |
tilman | what the shit | 10:20 |
prologic | e.g: gopher://daisy.shortcircuit.net.au/1/~prologic/sandbox/contact | 10:20 |
prologic | tilman, yeap :) | 10:20 |
prologic | always has | 10:20 |
tilman | shouldn't gopher://daisy.shortcircuit.net.au/README.md work? | 10:21 |
prologic | no | 10:21 |
prologic | 0/README.txt | 10:21 |
tilman | '0' is a gopherspace? | 10:21 |
teK_ | you are seriously sick. | 10:21 |
prologic | the selector (client-side) is normally <itemtype>/<path> | 10:21 |
tilman | 0/dosgames | 10:21 |
tilman | *cough* | 10:21 |
prologic | 0 for text | 10:21 |
prologic | item types | 10:21 |
tilman | kk | 10:21 |
tilman | so, i have a question | 10:21 |
prologic | teK_, :) | 10:22 |
tilman | are you still using SMTP? | 10:22 |
tilman | or did you switch to UUCP too? | 10:22 |
prologic | teK_, I am sick | 10:22 |
prologic | with the cold/flu :( | 10:22 |
teK_ | s/still/yet/, tilman | 10:22 |
teK_ | prologic: get better soon :-) | 10:22 |
prologic | tilman, I still use Google Apps :) | 10:22 |
prologic | thanks | 10:22 |
prologic | trying :( | 10:22 |
tilman | -Fair question :) Here's why I'm using Gopher in 2014. | 10:22 |
teK_ | stop gophering all day! | 10:22 |
tilman | +Fair question :) Here's why I'm using Gopher in 2015. | 10:22 |
prologic | got some drugs that'll help hopefully | 10:22 |
tilman | there, have a patch for README.md | 10:22 |
prologic | tilman, lol | 10:22 |
tilman | happy new year :D | 10:22 |
prologic | haha | 10:23 |
Romster | http://xkcd.com/409/ that suits prologic | 10:24 |
Romster | damn time wasting xkcd | 10:25 |
*** lnds has joined #crux | 11:20 | |
*** xeirrr has left #crux ("Leaving") | 11:45 | |
Romster | prologic, how on earth do you bind mount a host directory inside a docker container? | 12:23 |
Romster | all the examples on the net use -v /usr/foo:/usr/foo but -v is for version | 12:24 |
Romster | i realize its not portable doing this but for my own use | 12:24 |
prologic | docker run -v hostpath:containerpath ... | 12:28 |
Romster | -v=[]: Create a bind mount with: [host-dir]:[container-dir]:[rw|ro]. | 12:31 |
prologic | that's the one | 12:31 |
Romster | isee that but -v is --version | 12:31 |
prologic | no | 12:31 |
Romster | docker start -i -v /var/cache/ccache:/var/cache/ccache 94a49153176c | 12:31 |
Romster | flag provided but not defined: -v | 12:31 |
prologic | not for the run command | 12:31 |
prologic | you cannot do it via start | 12:31 |
prologic | after the fact | 12:31 |
prologic | docker run -v | 12:31 |
Romster | oh flags are specific to each command | 12:32 |
prologic | correct | 12:32 |
Romster | -_- | 12:32 |
prologic | sub-commands ftw | 12:32 |
Romster | no wonder i'm going crazy | 12:32 |
prologic | yup :) | 12:32 |
Romster | whos crazy idea was that | 12:32 |
prologic | pretty normal actually | 12:32 |
prologic | i.e: git :) | 12:32 |
prologic | think about it | 12:32 |
Romster | and ok how do i start an existing image with that mount | 12:33 |
prologic | you mean an existing stpped container? | 12:33 |
prologic | you can't afaik because you can't "bind mount" after the fact | 12:33 |
prologic | you'll have to create a new container I think | 12:33 |
Romster | never really noticed it iwth git... probably because i'm familiar with git. | 12:33 |
Romster | seriously | 12:33 |
Romster | this shits me | 12:34 |
prologic | what would you loose? | 12:34 |
Romster | first container size... now this | 12:34 |
prologic | docker commit :) | 12:34 |
prologic | commit your container as an image | 12:34 |
prologic | and re-run it new | 12:34 |
Romster | not really done with it to commit but ok i'll try that. | 12:34 |
prologic | doesn't matter | 12:35 |
Romster | does not help no man page so browse the net | 12:36 |
Romster | fatal: Not a git repository (or any parent up to mount point /home) | 12:39 |
Romster | Stopping at filesystem boundary (GIT_DISCOVERY_ACROSS_FILESYSTEM not set). | 12:39 |
Romster | oh right so i have to then set that up first... | 12:39 |
Romster | simple god damn task... | 12:40 |
prologic | wut? | 12:40 |
prologic | no | 12:40 |
prologic | docker commit | 12:40 |
prologic | not git commit | 12:40 |
prologic | read the docs :) | 12:40 |
Romster | oh for fucks sake | 12:43 |
Romster | what am i doing | 12:43 |
Romster | i meant to type docker but i thought of commit as git sigh | 12:44 |
Romster | did it without even thinking | 12:44 |
prologic | more coffee :) | 12:45 |
Romster | sometimes i think i should quit computers, at my rate. | 12:45 |
Romster | docker commit 94a49153176c crux-with-incomplete-build | 12:47 |
Romster | e77aa24be505f99a4372d565bef7583d116be2dfe71ce2ff0a2dad95a942e655 | 12:47 |
Romster | ok so it's not in docker ps --all | 12:47 |
Romster | as a named commit. nor is it's hash | 12:47 |
Romster | ok so it's a docker images | 12:48 |
Romster | docker run -i -t -v /var/cache/ccache:/var/cache/ccache crux-with-incomplete-build | 12:49 |
Romster | and success | 12:49 |
Romster | thanks prologic | 12:49 |
prologic | no | 12:53 |
prologic | but it's now commited as a new image | 12:53 |
prologic | committed* | 12:53 |
prologic | docker images | grep e77aa24be505f99a4372d565bef7583d116be2dfe71ce2ff0a2dad95a942e655 | 12:53 |
prologic | you should now be able to | 12:53 |
prologic | docker run -v ... e77aa24be505f99a4372d565bef7583d116be2dfe71ce2ff0a2dad95a942e655 <command> | 12:53 |
prologic | oh | 12:53 |
prologic | you wokred it out :) | 12:53 |
prologic | cool :) | 12:53 |
prologic | Romster, also NB: You *are* kind of using Docker for use-cases it was not really intended for -- i.e: you are breaking some of it's boundaries :) (but I don't see this as a bad thing if some of your use-cases help to shape Docker) | 12:54 |
prologic | so get involved :) | 12:54 |
Romster | yeah i confused myself then i was like git commit without even looking or thinking | 12:55 |
Romster | because i type that a million times. | 12:55 |
prologic | it's like a saw; if you try to hammer nails with it it won't work too well :) | 12:55 |
Romster | i'm always pushing the limits of everything i use. so it seems. i have unusual use cases. | 12:55 |
prologic | heh | 12:55 |
Romster | the mount is my personal use. | 12:56 |
Romster | i share ccache over all my chroots and now docker for my build system. | 12:56 |
Romster | anything portable i wont do outside binds. | 12:57 |
prologic | ahh i see | 12:58 |
Romster | this is semi automated CI atm | 13:00 |
Romster | i'm using features i'm not aware of well. | 13:00 |
Romster | i have read about portability and the design goals of docker. | 13:00 |
prologic | yeah I expect your CI image won't run here :) | 13:03 |
Romster | it would except that mount | 13:03 |
Romster | http://rom.ster.zone/desktop_compile.png | 13:08 |
Romster | -j20 distcc ccache docker | 13:09 |
teK_ | with what factor does -j20 speed up the build? | 13:10 |
Romster | it's hard to quote. | 13:10 |
Romster | each build is different from each project. and having ccache on does skew that. | 13:11 |
Romster | it's considerably faster though | 13:11 |
Romster | i'd have to get real numbers though without ccache running. | 13:11 |
teK_ | just estimate :) | 13:11 |
Romster | i could do that if you like. | 13:11 |
Romster | probably at least half the time. | 13:11 |
Romster | llvm i can build in 20 minutes iirc | 13:12 |
Romster | i'll get some real tests and pastebin | 13:12 |
teK_ | I think I can do that on my i7, too. Wait a minute ;) | 13:12 |
teK_ | just out of curioisity don't waste time for me :) | 13:13 |
Romster | you wanna know between my desktop -j4 or -j1 ? | 13:13 |
teK_ | -j20 | 13:13 |
teK_ | < =======> Building '/usr/ports/packages/llvm#3.4.1-1.pkg.tar.gz'. | 13:14 |
Romster | maybe i'll do 3 tests -j1 and -j4 on my desktop only then -j20 over distcc | 13:14 |
Romster | ok -j20 then... | 13:14 |
teK_ | if you like | 13:14 |
Romster | =======> Building '/var/ports/packages/llvm#3.4.2-1.pkg.tar.xz' | 13:15 |
Romster | xz takes a bit longer to compress so we'll see | 13:15 |
Romster | oh fuck i need to restart i got ccache going... | 13:16 |
prologic | I would say -j20 has no more effect than -j9 | 13:16 |
prologic | if you have an 8 core CPU | 13:16 |
prologic | :) | 13:16 |
Romster | ah i'll let this finish first. | 13:16 |
Romster | i got 4 core 2 duos and 3 quad cores prologic | 13:16 |
prologic | oh are you doing distcc? | 13:16 |
Romster | yes | 13:16 |
prologic | distributed compiling? | 13:17 |
prologic | ahh | 13:17 |
Romster | yep | 13:17 |
prologic | makes perfect sense now :) | 13:17 |
Romster | see my screenshot | 13:17 |
prologic | many CPU(s) in total across the cluster? | 13:17 |
Romster | 20 | 13:17 |
prologic | ahh | 13:17 |
prologic | totally makes sense :) | 13:17 |
Romster | http://rom.ster.zone/desktop_compile.png | 13:17 |
Romster | =======> Building '/var/ports/packages/llvm#3.4.2-1.pkg.tar.xz' succeeded. | 13:17 |
Romster | but that's with ccache also on... | 13:17 |
prologic | *nods* | 13:18 |
Romster | #export CCACHE_PREFIX="distcc" | 13:18 |
prologic | man if you can find me some cheapish 1RU (rack mountable) servers I'll throw a few in my rack :) | 13:18 |
Romster | ======> Building '/var/ports/packages/llvm#3.4.2 | 13:18 |
Romster | real test over distcc | 13:18 |
Romster | not everything speeds up but it helps on the big stuff. | 13:19 |
Romster | if oyu had a few i7s n distcc it would probably beat my entire setup | 13:19 |
prologic | ahh | 13:20 |
prologic | you've learned an important theoretical thing about parallelism | 13:20 |
prologic | I forget the law | 13:20 |
prologic | but basically if you distributed N tasks over M Nodes; it will take as long as the longest task to complete | 13:21 |
Romster | newtons law? | 13:21 |
prologic | so no; for small jobs it doesn't help :) | 13:21 |
prologic | no | 13:21 |
Romster | =======> Building '/var/ports/packages/llvm#3.4.2-1.pkg.tar.xz' succeeded. | 13:21 |
Romster | done teK_ how is yours? | 13:21 |
prologic | parallel computing theory | 13:21 |
Romster | it only works as good as how many parrreral make jobs it can run and not wait on a job to finish on a slow node. | 13:22 |
Romster | that could hold up the entire compilation | 13:22 |
prologic | exactly | 13:23 |
Romster | [11:15:01] <Romster> =======> Building llvm ccache | 13:23 |
Romster | [11:17:47] <Romster> =======> Building '/var/ports/packages/llvm#3.4.2-1.pkg.tar.xz' succeeded. | 13:23 |
Romster | so supper fast on ccache | 13:23 |
Romster | [11:18:34] <Romster> ======> Building '/var/ports/packages/llvm#3.4.2 distcc -j20 | 13:23 |
Romster | [11:21:20] <Romster> =======> Building '/var/ports/packages/llvm#3.4.2-1.pkg.tar.xz' succeeded. | 13:23 |
Romster | 3 minutes something llvm | 13:24 |
Romster | ok way faster than i was thinking. | 13:24 |
Romster | have to say though llvm can do saturate my entire distcc cluster. not all ports do. | 13:24 |
teK_ | still building | 13:24 |
teK_ | 3.4.1, though ;) | 13:25 |
Romster | why so old << | 13:25 |
teK_ | because this machine still is on 3.0 | 13:26 |
teK_ | I was not lying when I said I'd be super busy 'til May | 13:26 |
Romster | ah | 13:26 |
Romster | prologic, i was aware of the usefulness of how much faster it can go. | 13:27 |
Romster | there are so many factors involved. | 13:27 |
Romster | overall it's still pretty damn fast. | 13:27 |
teK_ | the -j20 build did not use ccache, right? | 13:27 |
Romster | first one did second one not. | 13:27 |
teK_ | ok | 13:28 |
Romster | see up ^ | 13:28 |
teK_ | sudo prt-get update -fr llvm 5304.31s user 114.48s system 610% cpu 14:47.11 total | 13:28 |
teK_ | but this is whith the installation/extraction | 13:28 |
teK_ | 14.5 minutes I'd say :) | 13:28 |
Romster | 2 mins 46 seconds ccache and 3 minues 14 seconds with straight -j20 distcc | 13:29 |
teK_ | i7-2600K, tmpfs and -j7 | 13:29 |
Romster | this is also with slower xz compression. | 13:29 |
Romster | so i spent several more seconds compressing than gz | 13:29 |
teK_ | I read that Romster. but the second entry did not exclude ccache and you could have overlooked.. | 13:29 |
Romster | oh and i didn't use tmpfs in this docker image. | 13:29 |
Romster | oh hang on.... | 13:30 |
Romster | i might be wrong... | 13:30 |
Romster | i commented out export CCACHE_DIR="/var/cache/ccache" but that wont work when i got CC and CXX set to ccache... sorry. | 13:31 |
Romster | lets do that again. | 13:31 |
Romster | sdtar -p -o -C /usr/ports/work/llvm/src -xf /var/ports/distfiles/llvm-3.4.2.src. | 13:32 |
Romster | ah hard to get the first line | 13:32 |
Romster | http://rom.ster.zone/desktop_compile_llvm.png proof it uses all my cores | 13:35 |
Romster | =======> Building '/var/ports/packages/llvm#3.4.2-1.pkg.tar.xz' succeeded. | 13:38 |
Romster | and done | 13:38 |
Romster | ok so 6 minutes and 21 seconds approx | 13:38 |
Romster | still going tek | 13:44 |
teK_ | 6 minutes for -j20, no ccache, tmpfs | 13:44 |
teK_ | ? | 13:44 |
Romster | i need todo a -j4 on my pc only without distcc too. | 13:44 |
Romster | 6 minutes for -j20, no ccache no tmpfs | 13:44 |
Romster | storage in docker is on a real LV on mdadm raid1 | 13:45 |
Romster | i should setup tmpfs next | 13:45 |
Romster | i oly just got ccache bind mount to my system working earlier. | 13:45 |
Romster | in docker that is. | 13:46 |
Romster | prologic, this is why i want CI | 13:47 |
Romster | i have a kick ass build system | 13:47 |
prologic | sure | 13:49 |
prologic | find me heaps of spare time | 13:49 |
prologic | or better yet win me the lotto :0 | 13:49 |
prologic | :) | 13:49 |
Romster | haha | 13:50 |
Romster | don't we all | 13:50 |
*** fengshaun has quit IRC | 13:50 | |
Romster | teK_, surely it's done already by now | 14:02 |
*** EmoSpice_Home has joined #crux | 14:04 | |
teK_ | i did not re-run if you meant that. Still 14.5m | 14:05 |
EmoSpice_Home | Good morning all. | 14:06 |
Romster | teK_, oh right i missed that | 14:07 |
teK_ | hello EmoSpice_Home | 14:08 |
Romster | hi EmoSpice | 14:09 |
*** doomicide has joined #crux | 14:09 | |
*** plow has joined #crux | 14:18 | |
*** EmoSpice_Home has quit IRC | 14:24 | |
*** SiFuh has quit IRC | 14:32 | |
*** BitPuffin has joined #crux | 14:35 | |
*** doomicide has quit IRC | 14:38 | |
*** EmoSpice_Home has joined #crux | 14:40 | |
*** diverse has joined #crux | 14:55 | |
*** hhhhhhhh has joined #crux | 15:00 | |
diverse | hhhhhhhh: alright, you are finally here | 15:01 |
Romster | i haven't tried rust yet i been fixing docker and other stuff. | 15:01 |
hhhhhhhh | diverse: hi | 15:02 |
diverse | hhhhhhhh: could you fix up my makefile so that it compiles my libmeta, rust has been complaining about my Cmd<T> trait not detecting the type parameter when it should. Also I would like the makefile to also look for a prebuilt tar crate on the system instead of having source in my project (because shit changes a lot) | 15:04 |
diverse | Romster: btw, I plan to build a 2nd i7 machine to run in parallel, so I will go up from -j8 to -j16 ;) | 15:16 |
Romster | cool | 15:18 |
*** jdolan has quit IRC | 15:20 | |
hhhhhhhh | dive | 15:23 |
hhhhhhhh | oops | 15:23 |
*** hhhhhhhh has quit IRC | 15:47 | |
*** Feksclaus has joined #crux | 15:49 | |
*** jdolan has joined #crux | 15:53 | |
*** leo-unglaub has joined #crux | 16:36 | |
*** EmoSpice_Home has quit IRC | 16:36 | |
*** EmoSpice_Home has joined #crux | 17:38 | |
*** EmoSpice_Home has quit IRC | 17:47 | |
nwe | can I someone see from which ports a package are installed from.. | 17:57 |
*** asie has joined #crux | 18:01 | |
*** SiFuh has joined #crux | 18:19 | |
nwe | somehow* | 18:34 |
*** asie has quit IRC | 18:35 | |
jaeger | nwe: the system doesn't track that currently... but it's usually from the repo "prt-get info package" returns | 18:43 |
nwe | jaeger: okey :) thanks for the answer | 18:46 |
jaeger | np | 18:46 |
nwe | Im playing around with nginx and image_filter :) | 18:46 |
nwe | and I was unsure if I used my port or the nginx port from contrib :) | 18:46 |
jaeger | no builtin way to know for sure, unfortunately. If they're significantly different you can figure it out, I'm sure, but not from the ports/pkg system | 18:51 |
jaeger | that is something on my rewrite list :) | 18:51 |
nwe | oh :) I figure out that I using my ports on that system :) | 18:53 |
nwe | I have some other modules builtin so :) | 18:53 |
nwe | hmm I wondering if I should try docker on my laptop.. | 18:53 |
*** nullmark has quit IRC | 19:11 | |
*** nullmark has joined #crux | 19:12 | |
*** Feksclaus has quit IRC | 19:31 | |
*** sonus has joined #crux | 19:43 | |
*** leo-unglaub has quit IRC | 20:30 | |
cruxbot | [core.git/3.1]: [notify] file: updated to 5.22 | 20:36 |
frinnst | https://www.youtube.com/watch?v=ZXDQzexYEJA&feature=youtu.be&html5=1 | 20:45 |
frinnst | thats pretty fucking funny | 20:45 |
frinnst | nsfw | 20:45 |
frinnst | im such a child | 20:48 |
frinnst | did I break the channel? | 21:20 |
frinnst | http://imgur.com/ochHuow | 21:20 |
*** EmoSpice_Home has joined #crux | 21:21 | |
dxlr8r | gcc compile fails with this message: | 21:21 |
dxlr8r | /usr/ports/core/gcc/work/src/build/x86_64-unknown-linux-gnu/32/libstdc++-v3/include/ext/enc_filebu | 21:23 |
frinnst | use a pastebin | 21:23 |
dxlr8r | f.h:63:1: fatal error: can't write PCH file: No space left on device | 21:23 |
frinnst | paste.nu, paste.lisp.org etc | 21:23 |
dxlr8r | it was only two lines, but yeah | 21:23 |
dxlr8r | I could | 21:23 |
frinnst | heh | 21:23 |
frinnst | I was expecting more :) | 21:23 |
frinnst | No space left on device <- pretty telling | 21:23 |
dxlr8r | it says "no space left" | 21:23 |
dxlr8r | yeah, but I have "plenty" | 21:23 |
dxlr8r | over 3GB | 21:23 |
dxlr8r | does it use that much? | 21:23 |
frinnst | yeah, thats not "plenty" | 21:23 |
frinnst | probably, yes | 21:24 |
dxlr8r | hehe :P I think so :P only for simple terminal usage | 21:24 |
dxlr8r | well, guess I have to shut down and dedicate more space | 21:24 |
dxlr8r | Filesystem Size Used Avail Use% Mounted on | 21:24 |
dxlr8r | /dev/root 7.1G 3.3G 3.4G 50% / | 21:24 |
frinnst | yeah but the source is pretty big and all the built objects does use a lot of space | 21:24 |
frinnst | firefox requires more than 8gb to build for reference | 21:25 |
dxlr8r | I have tons of space on an external though (writethrough in virtualbox) | 21:25 |
dxlr8r | have no X stuff. disabled that entirely | 21:26 |
dxlr8r | I could just temporary create a symlink to an external disk so i compiles | 21:26 |
dxlr8r | then I don't need to shut it down | 21:26 |
frinnst | yeah. Its the downside of running a source-based distribution. building stuff does take resources | 21:26 |
frinnst | you could specify the build-dir in pkgmk.conf | 21:27 |
dxlr8r | yeah, so I noticed. I always thought it would be enough since I would only use it for simple server usage | 21:27 |
dxlr8r | nice | 21:27 |
frinnst | I have a 32gb usb stick attached to my raspberry pi for building stuff :) | 21:29 |
frinnst | Its a bit overkill, but yeah | 21:29 |
dxlr8r | hehe | 21:29 |
dxlr8r | I have several TB attached to it, but not the root volume. the root volume is an image stored in ntfs on my windowns machine | 21:30 |
dxlr8r | but the disks are passthrough | 21:30 |
dxlr8r | so very little overhead if any :) they are fast! | 21:31 |
frinnst | can you resize or hotadd disks in virtualbox? | 21:31 |
dxlr8r | yeah, but then I would have to shut it down etc. | 21:31 |
dxlr8r | # PKGMK_SOURCE_DIR="$PWD" | 21:32 |
dxlr8r | it says "pwd" but the config file is in /etc/ | 21:32 |
frinnst | depending on the guest filesystem you could then resize it on the fly | 21:32 |
dxlr8r | so why does it store it in /etc/ports ? | 21:32 |
*** EmoSpice_Home has quit IRC | 21:32 | |
frinnst | the commented line is just an example. use /path/to/my/big/disk | 21:32 |
dxlr8r | nah, easier to use on of the connected disks | 21:32 |
frinnst | oh, source dir | 21:32 |
frinnst | PKGMK_WORK_DIR="/usr/ports/work/$name-$version-$release-work" | 21:33 |
dxlr8r | ahhh, I thought it was the default | 21:33 |
dxlr8r | and not an example | 21:33 |
frinnst | thats what I use for example | 21:33 |
cruxbot | [contrib.git/3.1]: libreoffice: 4.2.7 -> 4.2.8 | 21:33 |
dxlr8r | I use ext4 on everything, so shouldn't be a problem | 21:34 |
frinnst | http://blog.pluralsight.com/resize-vmware-linux-drives | 21:35 |
dxlr8r | will think about it. expand root vs use my connected raid | 21:36 |
dxlr8r | raid is probably just as fast. even though the root disk image is on my ssd | 21:36 |
dxlr8r | but some overhead since it's an image and not passthrough | 21:37 |
*** JSchilli1 has joined #crux | 22:10 | |
JSchilli1 | is it dangerous to be prt-getting two packages at once? | 22:21 |
jaeger | It's usually fine to be building two and once but if the pkgadd calls run concurrently one will fail | 22:39 |
jaeger | you can just run it again | 22:39 |
rmull | As an embedded developer I'm always amazed when a single piece of software requires multiple gigabytes of space | 22:42 |
dxlr8r | yeah, to me several GB is still a lot | 23:09 |
dxlr8r | born in the 80'ties :P | 23:10 |
dxlr8r | I decided to move the work folder to another volume instead of increasing the space. didn't want to fiddle with expanding with ext4 and risking my data | 23:11 |
dxlr8r | to much hassle to restore :) | 23:12 |
dxlr8r | don't want to dedicate to much space to root, I need that for my games :P | 23:12 |
dxlr8r | (root is in SSD) | 23:12 |
*** lnds has quit IRC | 23:15 | |
*** doomicide has joined #crux | 23:53 |
Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!