IRC Logs for #crux-devel Saturday, 2015-01-17

prologicmorn'nall01:05
prologicmorn'n* :)01:05
prologicgoing to try and rebuild our media backend today01:05
prologicwee :)01:06
prologicgot us a NEO MiniX X5 mini (Android TV Media Hub)01:06
diversecool01:07
prologicthink I'll try to Dockerize thebackend too if I can01:13
prologicand I think jaeger is right of course01:13
prologicit's far easier sometimes to just re-install from scratch :)01:13
prologicI think it has CRUX 2.7 on it01:14
diversemight as well01:15
RomsterteK_,01:18
jaegerprobably quicker to reinstall than try to upgrade that one01:19
prologicour is doesn't do a dhcp request to get an ip address does it?01:19
prologicour iso*01:19
jaegernot until you tell it to01:19
prologicrats :)01:20
prologicwe should -- is it too hard to implement?01:20
jaegerwouldn't be hard but it doesn't load network modules automatically either01:20
prologicahh01:21
prologicyou'd have to compilea lot of common network modules as modules ?01:21
prologicand would the bootup need to make an assumption about the eth device?01:21
jaegermany of them are already there01:21
prologicenp0s0?01:21
prologicI'd better to an initial re-install at the tv then01:22
prologicthat's where it used ot be :)01:22
prologicget it to ssh readyiness :)01:22
jaegeryou can even do the entire install via ssh if you boot the ISO, load the network module, configure an address, start sshd, set the root password01:23
prologicsure :)01:24
prologicbut then I may as well do the install at the same time :)01:24
prologichehe01:24
Romstersure exit mesa3d and llvm and llvm-32 but leave mesa3d-32 to me -_-01:25
Romsters/exit/edit01:25
prologicnow where did I put that CRUX 3.1 CD I burnt a whiel ago :)01:25
jaegerIf you don't want to update mesa3d-32, don't :P01:26
Romsterit must be kept in sync <_<01:34
Romsterrealy only is jaeger myself and tek when he feels like prodding around in compat-3201:35
Romsterjaeger, i'm wondering if it's time for a crux 3.1.1 with all the changes that's gone on.01:38
Romsterthat can also inject the new dependency for xorg too.01:39
diversewell, jaeger just does a new updated ISO every so often, but what I would suggest is linking the updated ISO.01:39
Romsterbut not many even use the updated iso... thnk it needs tobe at least added onto the download page.01:39
diversethat's what I would like01:40
Romsterwell we could put a unoffical updated iso.... but really unofficial.. when it's the same guy that does the offical iso's01:41
Romsterwould be nice if it was also a torrent for the updated iso and synced on the mirrors.01:41
diverseah, that might be a pain depending how frequently jaeger updates the ISO01:41
Romsterprobably a month or two depending on changes.01:42
jaegera torrent would be fine for an official 3.1.1 ISO but a pain in the ass for an updated ISO that changes01:42
Romstereh it would just became obsolete and a new torrent be released.01:43
Romsterprobably wouldn't hurt to put a date on the crux-3.1-20150115.iso etc01:44
Romsterjust ideas01:45
jaegerI did that in the past but you can see the date in the download listing so I dropped that01:45
Romsterfor a single download is fine but if you were to torrent it.01:46
diverseRomster: I think your Crux 3.1.1 idea is better01:46
Romsteri'd be mroe than happy to scrape and keep track01:46
Romsterwell i suggested crux 2.8.1 ages ago when we had updated gcc glibc and a ton of other stuff and we delayed 3.0 because of other issues upstream and testing01:47
Romsterand everyone agreed then01:47
diverseah01:47
Romsterbut that was realy really major diff on sysup01:47
Romsteron large ports including toolchain01:48
Romsterfrinnst,01:48
RomsterNEW       -rwxr-xr-x      root/root       usr/lib/dri/gallium_drv_video.la01:48
RomsterNEW       -rwxr-xr-x      root/root       usr/lib/dri/gallium_drv_video.so01:48
Romster=======> Building '/var/ports/packages/mesa3d#10.4.2-1.pkg.tar.gz' succeeded01:48
diversejaeger: Romster is willing to keep in sync with the updated ISOs, so why not?01:48
diverseah I got that NEW diff too01:49
Romsterhmm need to test again.01:49
Romsterfires up another docker container01:49
Romsteri still need to finish off CI building for me.01:51
Romsteri have a brilliant idea to solve binary packages too. need to test that idea after.01:52
Romsteronce i build every port in its own container save the deptree and prt-get depinst port name. to a file along with a hash of each built package. and save all that to hashed files, ifi did this for every port then... again with other ports installed and find the total number of differences.01:53
Romsteronce i have a list i can figure out how to offer it up for pkg-get depending on what ports there system has already.01:54
Romsterso missing libraries will be a thing of the past01:55
Romsterand in the process i can find what is broken.01:55
Romsterand what links to what with finddeps01:56
Romsteri am brilliant.01:56
Romsterprologic, beat that idea <<01:56
Romsteru-fail ?01:59
diverseI had to google what that was02:00
Romstersome EFI bot loader?02:01
Romsterboot*02:01
jaegerembedded/SoC boot loader, not EFI02:01
Romsterah02:02
diversenope, imagine putting a boot loader on your cell phone (not just smartphones)02:02
jaegerI built a crux rootfs and packages for the odroid-c1 for crux-arm but the boot loader is flaky as hell02:02
jaegersometimes it boots in a reasonable time... sometimes it takes 40 minutes or more02:02
Romsterthat's worse than just failing reliably02:02
diversethe heck?02:02
jaegerchanging nearly ANYTHING in the boot.ini causes it to fail02:02
jaegerbeen fighting with it for a while, I just gave up02:03
Romsteris there any other boot loaders?02:03
jaegernot for this device02:03
Romsterthat sucks02:03
Romsterat that point u-boot devs with your setup and let them debug02:04
Romsterproviding they have that hardware to test02:04
Romsterwhy does mesa3d link with binary nvidia O_O02:07
Romsteris that so other apps can talk though mesa3d API than to nvidia directly?02:08
Romsterthrough*02:08
Romsterhmm what i'll also do is setup my other kernels to handle docker and share a common filesystem between them. then i can fire up a container on any of the machines02:10
Romsterand not fuss over distcc/ccache for testing.02:11
Romsterspread my load out over many machines that way02:11
Romsterload average: 25.72, 13.87, 10.5102:12
Romsteralways need more power.02:13
Romsteronly a 4 core and it's on 25...02:13
Romsterif only i had a really big super computer <<02:13
Romsteryou know what anoies me teK_ having to use -O from portsdb03:15
Romsterwget 'https://crux.nu/portdb/?a=getup&q=romster' -O romster.httpup03:16
Romsterelse i get a=getup&q=romster as the filename. is there any ways to fix that modrewrite perhaps?03:16
Romsterso we can just see https://crux.nu/portdb/foo.httpup03:17
jaegerteK_: http://crux.ninja/distfiles/03:37
jaegerstill uploading, I'll work on a cleaning script tomorrow, most likely.03:37
*** mavrick61 has joined #crux-devel03:43
Romsterwhat will happen on github clashing files...03:51
jaegerat present, md5sum mismatches03:53
Romsteryeah same as what will happen with my http://crux.ster.zone/distfiles/03:54
Romsteri did however do this http://romster.me/distfiles/hash/03:55
Romsterwith a script.03:55
Romsterbut pkgmk isn't smart to look in distfiles/hash/$name-$version/md5/$md5sum/04:00
Romsterbut it wouldn't take much effort but then everyone will probbly prefer to make pkgmk name files... like example.com/v1.2.3.tar.gz#name=foo-v$version.tar.gz04:01
Romsterhttp://crux.ster.zone/ i added log files for easy access and url lists.04:16
Romsterhttp://www.versionsort.com/list/crux-ster-zone-distfiles.txt so i can see what versions if more than 1.04:26
jaegerteK_: I lied and worked on it tonight, somewhat04:57
jaegerit's not particularly efficient but it seems to do the job04:58
jaegerhttp://pastebin.com/aDn21qs704:59
jaegerit lists files in the distfiles dir that aren't referenced by a Pkgfile05:00
jaegerin other words, files that need to be removed when the distfiles mirror is updated05:01
jaegerhttp://pastebin.com/WpaEDcJK <-- there's a run of it on my VPS05:29
Romsteris it crux based or can it just look at a checkout of Pkgfiles?05:42
jaegerit uses a ports tree to get a list of required files from Pkgfile source=() lines05:56
jaegerI'll share it tomorrow after I've had a chance to test it a bit more. For now I'm going to bed, tired and probably making mistakes somewhere05:57
diversehave a good night jaeger05:57
jaegeryou too, thanks06:00
prologicjaeger, iso successfully updated?08:21
prologichaven't started rebuilding the media box yet :)08:21
prologichad to find a video cable in a tangled mess of em in a box08:21
prologicand then we had a visitor :)08:21
prologichttp://crux.ninja/updated-iso/08:25
prologic15-Jan08:25
prologicI guess that's a yes :)08:25
prologicsow08:34
prologicwow08:34
prologicour internet at home has improved since switching to InterNode's Business grade DSL08:34
prologicgot the ISO in ~13m08:34
prologicaveraging 9Mbps08:34
Romsteri be home users get less priority08:35
prologicabsolutely they do08:35
prologicI worked for a small DC/VOIP company a while back08:35
prologicnon-business users get bumpcus all bandwidth08:35
prologicnormally (I'm 99% sure) most ISPs give non-business grade users a pooled level of bandwidth08:36
prologicwith really shit contention ratios08:36
Romsterand heaps of latency spikes08:37
prologiccan't believe I'm burning a CD08:47
prologicI don't seem to have an ISB sticks08:47
prologichaha08:47
prologicchrist almighty08:52
prologicthe burn fialed08:52
RomsterO_O08:54
prologichttps://gist.github.com/therealprologic/41183fa5af7d8b37d7ad09:19
prologichmm09:19
prologicnot sure what I'm doing wrong09:19
Romsterprologic, check the md5sum of your burned cd09:22
Romsterand you do have CONFIG_ISO9660_FS=y set?09:24
prologicyes09:24
prologicand yes09:24
Romsterout of ideas...09:24
prologicthis is just plain bizare09:24
Romsterwant my kernel config to look over it works for me.09:24
prologicThis is a hybrid ISO though09:25
prologicdoes it need any special wodim options?09:25
Romsternope09:25
prologichmmmmm09:25
Romsterwodim -v dev=/dev/sr0 /home/romster/downloads/crux-3.1.iso09:26
Romsterall i did09:26
RomsterCONFIG_UDF_FS=y09:28
Romsteri also happen to have CONFIG_UDF_NLS=y set too09:29
Romsteri don't know really i haven't hit that issue.09:29
prologicthis is nuts09:30
prologicI can't keep wasting CDs09:30
Romstergo use a cdrw09:31
Romsterbut you probably don't have any of those...09:32
Romsteryou sure the burn is bad?09:32
Romsterfire it up on another pc.09:32
Romsteror even VM09:32
prologicumm09:57
prologicyou can't put a cd into a vm09:57
prologic:)09:57
prologichmm09:58
prologicthat's interresting09:58
prologicif you remove the cd and put it back in09:58
prologicit's fine09:58
prologicI remember cdrecord would actually pop the disk automatically09:59
prologicbut wodim doesn't seem to do that09:59
*** leo-unglaub has joined #crux-devel10:07
diverseprologic: wasn't there a command that allows you to open the drive slot? What was it, 'inject'?10:17
diverselike for example: wodim && inject10:17
diversedone10:17
diverseoh yeah, it was "eject"10:18
diverseyou can use that to open or close it10:18
Romsteri've used a cd in a VM before.... in virtualbox use sr011:17
Romsterheeh yeah i used to ssh and eject11:18
Romsterhilarity ensued11:18
Romsteror beep11:18
teK_jaeger: interested in your cleanup step11:45
teK_Romster: I'd have to fiddle with the Header (wrt portdb, let me check)11:46
frinnstRomster: ack. strange though11:54
frinnstRomster: do you have a mesa3d buildlog?12:53
Romsteri was earlier and it doens't do it in a clean container. but i did notice unrecognized options: --enable-gallium-egl, --enable-gallium-gbm12:55
Romsterthere now --enable-egl --enable-gbm12:56
Romsterbut that still doens't solve them two new files showing up.12:56
frinnstok, silly me missing those. I've run the 10.4 branch for a while now13:03
frinnstneed some breakfast and i'll dig into it13:03
frinnstthis years winter has completely ruined me. slept 13hrs today and probably averaged ~10hrs/day this week13:06
frinnstneed sunlight!13:06
Romsteromg13:09
Romsterguess lack of sun will do that.13:10
diversefrinnst: be sure to take some D3 Vitamins13:10
frinnstI am13:17
frinnstRomster: do you have libva installed?13:19
frinnsti'd bet you do and thats why you get a footprint NEW13:20
frinnstmoar coffeeeeee13:20
diversefrinnst: also take an aspirin, it will help open your blood vessels in your brain, thus make you more alert as well13:24
frinnstnah, I'll just go for a walk. usually does the trick13:25
frinnstbut thanks for the tips ;)13:25
diverse(It's what I do to make myself into programmer-mode)13:26
diverse(along with a caffine pill)13:26
diverse(and D3 because I rarely get sun)13:27
diverse(and of course, take it with cold water)13:30
Romsterjust made coffee13:31
Romsterfrinnst, well i have many containers of stuff with and without libva13:31
Romstertries with.13:31
*** leo-unglaub has quit IRC13:43
teK_frinnst: take some vitamine D13:46
Romsterhttps://crux.nu/gitweb/?p=ports/xorg.git;a=commitdiff;h=e13a3b912f892e42952930d52b3e605a1f7a6f7913:46
Romstertell me whats wrong withthat..13:46
Romstergallium_drv_video.so and gallium_drv_video.la is libva.13:48
Romsterbut it doesn't link dynamically with libva13:49
teK_Romster13:49
teK_% wget -S https://crux.nu/portdb/\?a\=getup\&q\=romster13:49
teK_(...)13:49
teK_  Content-Disposition: attachment; filename="romster.httpup"13:49
teK_(...)13:49
teK_Saving to: 'index.html?a=getup&q=romster'13:50
teK_seems as if  wget just sucks :p13:50
Romsteryeah why13:50
teK_works with chromium13:50
Romsteroh13:50
teK_the official trick IS to include the c-c header13:50
teK_c-d header13:50
Romsteryeah i've done it in firefox correctly as well.13:50
teK_wget has a --content-disposition  option13:50
Romsterargh i can't add libva as libva depends on mesa3d wtf hmm13:52
Romsteroh that works but not on be default because buggy.13:54
Romsterwonder if curl is any better13:54
frinnstRomster: you mean -               --enable-gallium-egl \ ?13:58
frinnstits enabled by default13:58
Romster--enable-gbm \ is in there twice.14:01
Romsterbut that's minor.14:01
Romsteryeah the other is a default.14:01
Romsterand your right libva gives them extra files but we can't add it to mesa3d as cyclic dependency14:02
Romsterlibva depends on mesa3d14:02
frinnstgawd damnit14:16
frinnstim to fucked up to fix this at the moment, feel free14:17
teK_jaeger: I added your mirror14:20
Romsterdrip of caffeine14:32
Romsterfrinnst, i fixed it for you14:37
*** leo-unglaub has joined #crux-devel14:46
*** leo-unglaub has quit IRC14:50
*** leo-unglaub has joined #crux-devel14:54
jaegerlooks like the cleanup uis working. From my log last night:15:19
jaegerremoved '/home/jaeger/distfiles/smplayer-14.3.0.tar.bz2'15:19
teK_:)15:19
jaegerI'll share it soon, have to go AFK for a bit now15:20
teK_take care15:22
*** leo-unglaub has quit IRC15:32
*** leo-unglaub has joined #crux-devel15:33
prologichmm I must be getting rusty15:45
prologicEXT3-fs: Cannot mount file system because of unsupported features (24)15:46
jaegerteK_: http://jaeger.morpheus.net/linux/crux/files/tmp/15:47
jaegerteK_: update-distfiles is running hourly on my VPS, it calls distfiles15:48
teK_will check it out in a minute15:48
teK_gotta play taxi ;)15:48
jaegernp15:48
prologicwtf why is lilo trying to mount / as Ext3 then Ext2?15:48
jaegerI doubt lilo is trying to mount anything but with that said, you probably built ext4 support with the ext2 and ext3 option enabled15:49
prologicI did15:49
prologicbut /dev/sda1 is ext415:50
prologicsorry I meant kernel :)15:50
jaegerturn off that option and those messages will go away, I'm betting. I could be wrong, though, haven't tested it15:50
prologicit's not just the messages15:50
prologicthe system won't boot15:50
teK_I think I even can getrid of thesecond redirect btw..15:52
teK_BRB..15:52
prologic/sbin/init exists but couldn't execute it (-8)15:54
prologicis the error I'm getting15:54
prologicmake the ext3/ext2 mesasges go away by append="rootfstype=ext4"15:54
prologichmm16:00
prologicI think I somehow had initramfs enabled in the kernel config16:00
prologicmust be a default :)16:00
prologicbecause I did "make oldconfig"16:00
Romsteryeah that is16:00
jaegerthat's fine16:01
jaegerhaving it enabled doesn't require one16:01
prologichmm ok16:02
prologicwell I'm stumped if it doesn't boot now16:03
prologicnot sure why it can't execute /sbin/init16:03
Romsterits like having kernel modules for FS you don't use16:03
Romsterharmless16:03
prologicso trying to boot into the system via the ISO (Rescue): CRUX root=/dev/sda1 -- seemed to boot okay but the system hard locks up at the tty login16:09
jaegerwhat kind of hardware is it?16:10
prologicstarting to think the SSD has conked it16:10
prologicCore i3 system16:10
prologicI think it's got an OZC Agility SSD as it's /16:10
RomsterOCZ says it all, known to be unreliable.16:17
Romsteri've heard lots of reports on OCZ, but I haven't seen one first hand.16:17
jaegerSome people have OCZ drives that last for years. Those people are lucky. :)16:18
Romsterhad a few samsung SSDs fail though16:18
Romstervery very lucky16:18
jaegertoshiba is going to have to work damned hard to bring that name back16:18
prologichttp://comments.gmane.org/gmane.linux.debian.user/47996616:19
prologicaccording to what I've read16:19
prologicthis is likely the cause of hw problems16:19
prologicgeez16:19
prologictalk about a time to die when I try ot rebuild the OS16:20
prologicmay as well run memtest oni t over night16:20
prologicrule our bad ram16:20
prologicthen replace the stupid SSD16:20
RomsterI managed to upgrade to systemd, which brought a whole bunch of problems16:20
Romstercoughs16:20
prologicmight just install on USB stick?16:20
prologicyou did what?16:20
jaegerdoes specifying init=/bin/bash work?16:21
Romsteri quoted the post16:21
prologicjeager: I'll try but I doubt it16:21
prologicsince specifying root= locked up the system16:21
prologicbrb16:21
Romsterif you can backup that disk and zero it out?16:21
jaegerwell, root=<root partition> would still try to exec init, I believe16:21
Romsternot sure if smart runs on SSD16:22
Romsterthey work diffently.16:22
jaegermost of them support SMART16:23
RomsterIntel 330/335 Series SSDs16:23
Romsteroh it does support short and long tests and the conveyance test16:24
RomsterMedia_Wearout_Indicator -16:25
Romstermake sure of that prologic16:25
Romster016:25
jaegerprologic: to clarify, specify root=/dev/sda1 AND init=/bin/bash16:27
jaegeralso, is sda1 correct? I usually make the first partition my swap16:27
prologicdidn't work16:29
prologicjaeger, yeah that's right16:30
prologicI normally don't configure any swap on my systems16:31
prologic/dev/sda1 is right16:31
prologicalso when I did: CRUX root=/dev/sda1 init=/bin/bash16:31
prologicit seemd to want to bring up the login getty(s)16:31
Romsterwho needs swap with 192GB ram right prologic16:31
prologicI expected to drop into bash16:31
prologicbut anyway it locked up16:31
prologicrunning memtest86+ on it now16:32
Romsterif it hangs up power it off remove ram and clean contacts with a bit of a4 type paper.16:33
Romsterbeing careful with the capacitors/resistors near the contacts.16:33
Romsteri've had a ton of machines with bad connections.16:33
Romsterjust make sure your not all staticy when you do that... so you don't fry the mobo/ram.16:34
Romstersame potential as the mobo ram earthed case.16:35
Romsterhavne't seen one die from static yet but there is always a first.16:35
*** c0x has joined #crux-devel17:25
*** Feksclaus has joined #crux-devel18:16
*** crash_ has joined #crux-devel18:54
*** Roosevelt has joined #crux-devel19:34
*** leo-unglaub has quit IRC19:39
teK_Hi Thomas,19:58
teK_Thanks for your feedback! We're always working to improve GitHub and we consider every suggestion we receive. I've added your idea about adding repository names to the release filename19:58
teK_Hi Thomas,19:58
teK_Thanks for your feedback! We're always working to improve GitHub and we consider every suggestion we receive. I've added your idea about adding repository names to the release filename19:58
teK_asdf19:58
teK_+to our internal Feature Request List.19:58
teK_We can't say if/when we may add a feature, however your feedback has definitely been recorded.19:58
teK_jfyi19:58
*** leo-unglaub has joined #crux-devel20:07
*** crash_ has quit IRC20:51
*** crash_ has joined #crux-devel22:48
prologicso I've been running memtest for ~6hrs so far (56% done)23:24
prologicno errors so far23:24
prologicshould I keep running it till 100%?23:24
prologicor would it be unlikely at this stage that the ram is bad?23:24
prologicso 13 passes, no errors23:39
prologicfrom what I've read it'sunlikely to be the ram in this case23:39
prologicI think that's enough :)23:39
prologictime to get a new OS drive23:39
crash_13 passes is good i just run it for some hours and then call it a day :)23:39
prologicyeap fair enough23:40
prologicI assume there are no issues installing an OS on a USB Stick?23:41
prologicI've tbh never really done that before :)23:41
prologicsince the SSD is clearly dead I don't really have any other choice satm :)23:41
crash_it really depends on what brand you have on the usb, some will live longer and some shorter :P23:41
crash_maybe you can use tmpfs as much as possible to reduce the writes to the usb.23:42
prologicWell most (if not all?) writes are to /var right?23:46
prologicSo what solutions exist to prevent writes to /var ?23:47
prologicAny data stored on this media box is actually stored on a spinning drive23:47
prologicmysql db, mythtv recordings, etc23:47
crash_you can remap /var to tmpfs in fstab?23:49
prologicI can23:53
prologicjust looking at various ways to do this23:53
prologicsome folk have written scripts to do this23:53
prologicto sync /var on shutdown23:53
prologicor just /var/log23:53
prologictbh I don't really care all that much about what goes in /var on this machine23:53
prologic:)23:54

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!