IRC logs for #baserock for Friday, 2016-01-15

*** edcragg has quit IRC00:45
*** gtristan has quit IRC05:08
*** gtristan has joined #baserock05:44
*** CTtpollard has joined #baserock08:20
*** franred has joined #baserock08:24
*** bruce_ has joined #baserock08:26
*** toscalix has joined #baserock09:21
*** jonathanmaw has joined #baserock09:28
*** edcragg has joined #baserock10:05
*** ratmice______ has quit IRC10:44
*** ratmice______ has joined #baserock10:47
radiofreeqtwebengine failing because it needs /dev/urandom :\12:40
pedroalvarezat build time...12:42
radiofreeyes12:42
radiofreemount-bind /dev ?12:44
pedroalvarezdoes it really need it at build time? or it only wants to check that it's there for runtime?12:45
gtristansemi-related note... I think sandboxlib could do with a real /dev/null12:47
radiofreeit's failing at build-time with http://fpaste.org/311188/52862026/12:47
radiofreealso was some kind of "petrify" script reintroduced somewhere?12:49
* radiofree doesn't fancy changing the sha's for a dozen "v5.5.1" refs12:49
pedroalvarezradiofree: http://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/morph.git/log/?h=sam/petrify-hack12:50
*** Guoww has joined #baserock12:51
*** Guoww has left #baserock12:51
pedroalvarezradiofree: there is a /dev/urandom created by fhs-dirs chunk12:53
*** Guoww has joined #baserock12:53
pedroalvarezbtw... what is "devices" in a chunk definition?12:54
radiofreei was about to ask the same thing :D12:54
* radiofree tries adding that12:54
pedroalvarezradiofree: but.. it should be there, in your build I mean.. It depends in fhs dirs12:54
pedroalvarezs/in/on12:54
radiofreeoh12:55
radiofreeit's not12:55
radiofreei'm using ybd if that makes a different (probably does?)12:55
pedroalvarezahh12:55
pedroalvarezmaybe12:55
pedroalvarezmaybe ybd doesn't know about this "devices" thingy12:55
radiofreei manually ran the linux-user-chroot command (which was a bit of a bugger to extract....) to gain a shell and it wasn't in /dev12:55
radiofreein fact nothing was in /dev (apart from /dev/shm)12:55
*** Guoww has quit IRC12:56
*** gww has joined #baserock12:56
radiofreepaulsherwood: ?12:56
pedroalvarezhttp://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/fhs-dirs.git/commit/fhs-dirs.morph?id=268b374f02d93a5df6f8f21e62a46c82e700cd9212:57
radiofreeso that's devices for the chunk right?12:57
radiofreefor build time12:57
*** Guoww has joined #baserock12:58
radiofreeor maybe not... it was previously an install-command script12:58
*** Guoww has left #baserock12:58
*** gww has left #baserock12:59
*** Guoww has joined #baserock12:59
*** gww has joined #baserock13:01
*** Guoww has quit IRC13:01
*** gww has quit IRC13:01
*** gww has joined #baserock13:02
*** gww has left #baserock13:02
*** gww has joined #baserock13:02
*** gww has quit IRC13:04
radiofreeadding it certainly doesn't fix things13:04
radiofreei'll try with morph13:11
*** gtristan has quit IRC13:22
radiofreepedroalvarez: is there arm cache for morph?13:23
radiofreeERROR: Conflicting fields "morph, and build-system" for qtwebengine in stratum strata/qt5-tools-qtwebengine.morph. ._O13:26
radiofreei'm not even building a system that goes anywhere near qt5-tools-qtwebengine?13:26
radiofreeno cache by the looks of it13:28
paulsherwoodradiofree: sorry, i've been out of the loop... what's the issue?13:29
radiofreeqtwebengine fails because there's no /dev/urandom, which may or may not be provided by fhs-dirs13:30
radiofreeor by "devices:"13:30
radiofree(in the chunk)13:30
radiofreei'm trying to create a simple test13:30
paulsherwoodsounds like a definitions issue, not ybd? :)13:30
radiofreewell if it works under morph (using devices:) then it's a ybd issue?13:30
radiofreepedroalvarez: is there any arm cache anywhere?13:31
radiofreei need to add the cache server to conf i think13:31
paulsherwoodybd handles devices same as morph, i though13:31
paulsherwoodt13:31
*** paulw has joined #baserock13:32
paulsherwoodhttps://github.com/devcurmudgeon/ybd/blob/master/ybd/sandbox.py#L30713:32
radiofreehmm.. i didn't see "Creating device node" in the log13:33
SotKIs it running the build commands before creating the devices here? (https://github.com/devcurmudgeon/ybd/blob/master/ybd/assembly.py#L160)13:36
radiofreeoooh... looks like it?13:38
* radiofree rejigs and tries again13:38
paulsherwoodah... would that logic be better reversed?13:39
radiofreedevices are for install time i guess?13:40
radiofree16-01-15 00:01:46 [1/17/308] [qtwebkit] Creating device node /src/ybd/tmp/tmpyUJvwD/qtwebkit.inst/.//dev/urandom13:40
paulsherwoodw00t :)13:40
radiofreeno it didn't work13:40
radiofreehttp://fpaste.org/311200/65244145/13:41
paulsherwood:(13:41
*** gtristan has joined #baserock13:45
radiofreeso "devices" are supposed to be installed with the chunk? in which case i should have a /dev/urandom right? since it's created by fhs-dirs13:48
radiofreeso if i enter the shell with the same chroot command (http://fpaste.org/311207/52866267/ - it might be nice to dump this when it fails) i have /dev/null13:58
radiofreebut nothing else13:58
radiofreecat /baserock/fhs-dirs.meta suggests /dev/urandom is installed via that13:59
paulsherwoodradiofree: sorry, this is too deep for me, i'm afraid14:01
paulsherwoodneeds a real linux person :/14:01
pedroalvarezradiofree: does the fhs-dirs artifact really have it?14:09
radiofreepedroalvarez: the artifact does14:09
radiofreei think the mkdnod is failing when you extract it14:10
pedroalvarezmorph's artifact has:14:10
pedroalvarezcrw-r--r-- root/root       1,9 1991-08-24 23:00 dev/urandom14:10
radiofreeyes14:10
radiofreeif i mknod dev/urandom in the build folder, *then* chroot into it, it works14:10
rjekYou shouldn't use random data that old.14:10
radiofreewell.. works... onto another new error now14:10
radiofreeok, well it's certainly buildable14:12
radiofreehowever i can't see how to fix this in ybd14:12
radiofreei did cd /src/ybd/tmp/myfailedfolder && mknod .... dev/urandom && enter-chroot /src/ybd/tmp/myfailedfolder14:13
pedroalvarezwould be good to know if that builds in morph14:13
radiofreei can "make" now, and it 'works' (now missing some dependencies, but it is at least past that error)14:13
radiofreeso there's something wrong with the extraction of fhs-dirs by the looks ofr it14:14
radiofreehmm.. tar xvf "fhs-dirs.key" works14:15
paulsherwoodradiofree: pls could you raise an issue on ybd with as much info as you have?14:19
*** bruce_ has quit IRC14:29
* radiofree does a quick morph v ybd test14:30
radiofreeyep14:31
radiofreethere's something about how ybd is extracting the fhs dirs artifact14:31
radiofreehttp://fpaste.org/311221/45286837/14:32
* radiofree will try using the artifact from morph in ybd14:33
*** bruce_ has joined #baserock14:38
radiofree...14:42
radiofreeok...14:44
radiofreeso i scp'ed my artifact cache over from a mustang to a jetson... which included the "fhs-dirs.CACHE.unpacked/" folder....14:44
radiofreescp didn't scp the nodes...14:45
radiofree"Troubleshooting is not rocket science. For all issues you may blame yourself. You are most likely the reason of failure."14:45
radiofreeok now there's something wrong with how /dev/shm is mounted in the chroot14:49
radiofree--mount-bind /src/ybd/tmp/sandboxlib.1mZSpI.tmpfs/dev/shm /dev/shm  doesn't work for me, but /dev/shm does14:49
radiofreenot sure what sandboxlib is doing there14:50
radiofreethe error in question http://fpaste.org/311227/86944514/14:50
*** Lachlan1975 has joined #baserock14:59
radiofreehow does morph handle mounting /dev/shm as tmpfs?15:16
pedroalvarezmorph mount's it as a tmpfs for every staging area15:18
radiofreeit'll work in morph then15:18
radiofreeyeah15:18
radiofreenone on /dev/shm type tmpfs (rw,relatime)15:19
pedroalvarezmaybe not for every staging area, but for every command it runs in the staging area15:19
radiofreeno mound-bind for /dev/shm used there15:19
radiofreeybd passes --mount-bind /src/ybd/tmp/sandboxlib.1mZSpI.tmpfs/dev/shm /dev/shm   to linux-user-chroot15:19
pedroalvarezsandboxlib passes? or is it ybd?15:20
radiofreewell there's "mounts.extend([(None, '/dev/shm', 'tmpfs'), (None, '/proc', 'proc'), ])"15:21
radiofreethat's in ybd15:21
radiofree/dev/sda1 on /dev/shm type ext4 (rw,relatime,data=ordered)15:22
radiofreeso it's absolutly not being mounted as tmpfs....15:22
radiofreeis this because i've changed the tmp directory in ybd.conf?15:23
radiofreeif  /src/ybd/tmp/sandboxlib.1mZSpI.tmpfs/dev/shm were created in /tmp/.... then it would actually work15:23
radiofreeaccording to sandboxlib it's supposed to create these tmpfs folders within an existing tmpfs (which would work), but it's clearly not her15:26
radiofrees/her/here15:26
rdalei create a symlink from ~/ybd to /src/ybd15:26
rdaleoh sorry, so do you15:26
radiofreethat's not going to work /src is my ssd15:27
* radiofree adds a load of "print" to sandboxlib15:27
radiofreedef tmpfs_for_user() looks like the badger15:27
radiofreeah.. set XDG_RUNTIME_DIR15:31
*** rdale has quit IRC15:33
radiofreethis is hardly ever set in baserock systems though15:33
radiofreeit it isn't set then it fallsback to just using tempfile without a directory15:34
*** rdale has joined #baserock15:35
radiofreewhich ends up being /src/ybd/tmp as per my ybd config15:35
paulsherwoodradiofree: you can force base: to whatever, and same for other dirs15:37
radiofreepaulsherwood: i know i can, but i think forcing the fallback to use /tmp would be better here http://git.baserock.org/cgi-bin/cgit.cgi/delta/sandboxlib.git/tree/sandboxlib/linux_user_chroot.py15:38
radiofreeconsidering doing this mount-bind for /dev/shm when TMPDIR != a tmpfs fs is never going to work properly15:39
* radiofree just does export XDG_RUNTIME_DIR=/tmp for now15:43
*** bruce_ has quit IRC16:04
radiofreehmm looks like qtwebengine is one of those things that assumes you'll always cross compile it16:09
pedroalvarezbecause nobody is as crazy as we are... :)16:10
*** toscalix has quit IRC16:41
*** CTtpollard has quit IRC16:54
*** jonathanmaw has quit IRC17:56
* richard_maw jut pushed https://gerrit.baserock.org/#/c/1744/ which is about adding shallow push support17:58
richard_mawit's the lorry-controller part17:58
richard_mawthe lorry part is demanding that I squash my commits17:58
richard_mawwe didn't used to need to do that, at least for morph changes17:59
richard_mawpedroalvarez: any ideas?17:59
richard_mawoh, nevermind18:00
richard_mawfirst response for googling is a messed-up change-id18:00
* richard_maw knows how that snuck in18:01
richard_mawthere we go18:03
richard_mawhttps://gerrit.baserock.org/#/q/status:open+project:baserock/baserock/lorry+branch:master+topic:shallow-support18:03
*** edcragg has quit IRC18:30
*** Lachlan1975 has quit IRC19:33
*** toscalix has joined #baserock23:28
*** toscalix has quit IRC23:30

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!