IRC logs for #baserock for Monday, 2015-10-12

*** gtristan has joined #baserock01:37
*** gtristan has quit IRC03:43
*** gtristan has joined #baserock05:01
*** zoli_ has joined #baserock05:56
*** zoli_ has quit IRC06:33
*** zoli_ has joined #baserock06:34
*** zoli_ has quit IRC06:43
*** zoli_ has joined #baserock06:43
*** paulwaters_ has joined #baserock06:43
*** gtristan has quit IRC07:01
*** gtristan has joined #baserock07:13
*** zoli_ has joined #baserock07:16
*** ssam2 has joined #baserock07:18
*** ChanServ sets mode: +v ssam207:18
*** zoli_ has joined #baserock07:24
*** zoli__ has joined #baserock07:29
*** zoli_ has quit IRC07:29
*** zoli_ has joined #baserock07:31
*** zoli__ has quit IRC07:31
gtristanhmmm, that's odd07:40
gtristansystem-integration hooks were working yesterday, and not today07:40
*** fay_ has joined #baserock07:42
*** edcragg has joined #baserock07:46
gtristanaha, I see07:47
gtristanjjardon, we need --enable-compat-libs at least for now in systemd07:48
gtristanhttp://paste.baserock.org/atehomibut07:48
gtristanaccountsservice configure... checking for SYSTEMD... no.07:49
* gtristan reapplies the patch07:49
*** toscalix__ has joined #baserock07:59
*** bashrc has joined #baserock08:04
*** CTtpollard has joined #baserock08:11
*** mariaderidder has joined #baserock08:15
*** ctbruce has joined #baserock08:19
paulsherwoodany idea why git operations could take forever on my local drive when i share it via nfs?08:19
paulsherwoods/forever/10 times as long or worse/08:19
*** edcragg has quit IRC08:20
KinnisonSlower on the server or on the client?08:20
paulsherwoodclient i think08:20
KinnisonWhich version of NFS ?08:22
paulsherwoodhow would i establish that?08:22
Kinnisonthe mount flags should tell you08:23
KinnisonSomething like vers=408:23
paulsherwood172.28.128.1:/Users/paulsherwood/Projects/vagrant_ubuntu/src on /src type nfs (rw,vers=3,udp,addr=172.28.128.1)08:23
KinnisonIs this entirely local?08:24
KinnisonIf so, boosting rsize and wsize, and using tcp rather than udp might improve performace08:24
KinnisonNFS is, sadly, a bit of a dark-art for optimising08:25
paulsherwoodyes it's entirely local. on 'server' (ie outside vagrant) things remain fast08:25
KinnisonWhat OS is 'server' ?08:26
paulsherwoodmac os08:26
KinnisonMight be worth checking to see if there's any documentation on macos <-> linux NFS performance tuning08:26
* Kinnison only knows linux <-> linux stuff08:26
paulsherwoodis it really possible performance could be > 10x worse due to bad tuning?08:28
KinnisonCertainly I used to see that back when I was running ARM boards off NFS-root.  I could get a 5-20x speed boost by configuring it properly08:28
KinnisonAlso it's worth checking if the lock services are working08:28
Kinnison(it'd be lockd under linux)08:28
KinnisonWithout them, the client has to fall back to slower less reliable locking (or hope that it doesn't matter)08:29
paulsherwoodthis is a much deeper hole than i was expecting :/08:29
paulsherwoodi guess software is usually like that08:30
ssam2i've been through a "vagrant looks really cool, I should try it" --> "vagrant is totally useless because if this limitation" cycle several times now08:30
ssam2*of08:30
paulsherwoodlol08:30
* paulsherwood wanders away from the internet08:31
* Kinnison wonders if paulsherwood tried the virtualbox FS sharing thing and what its performance was like08:33
*** zoli_ has quit IRC08:35
*** zoli_ has joined #baserock08:39
paulsherwoodKinnison: i beleive it was the same, but i can re-check08:42
*** jonathanmaw has joined #baserock08:42
*** gary_perkins has joined #baserock08:44
jjardongtristan: hey!! Congrats! :)08:54
gtristanjjardon, whew... at least the basics are there now :)08:55
gtristansomething to build on08:55
*** tiagogomes_ has joined #baserock09:00
tiagogomes_morning09:11
* tiagogomes_ will attempt to review today https://gerrit.baserock.org/#/c/1018/09:11
*** wschaller has joined #baserock09:18
*** flatmush has joined #baserock09:23
ssam2thanks tiago :)09:23
ssam2i've reviewed the gnome patches, all looks fine to me, I haven't submitted any of them in case someone else wants to give a more thorough review09:24
*** edcragg has joined #baserock09:28
ssam2but i will submit them all before the end of the day if nobody else does09:30
*** franred has joined #baserock09:30
tiagogomes_I'll try to have a look09:31
franredgtristan, great work!!09:36
gtristanfranred, mostly detective work, thanks :)09:39
franredgtristan, heh, that always happen in baserock (franred wants to think that this also happens in any software integration work)09:42
pedroalvarezSo, to have the network working on a rpi2 I have to restart 'systemd-netwokrd'09:55
pedroalvareznetworkctl says "failed"09:55
richard_mawpedroalvarez: what does journalctl -b -u systemd-networkd.service say?09:56
richard_mawPreviously the lorry-controller database would grow to 750M in 10 days. In the 3 days that it's been running with job cleanup after 1 day, it's at 87M.09:57
pedroalvarezfor eth0:09:57
pedroalvarez- Gained carrier09:57
pedroalvarez- Could not start IPv6 router discovery...09:57
pedroalvarez- Failed09:57
pedroalvarezrichard_maw: good!09:57
richard_mawpedroalvarez: it at least indicates a reduced growth09:58
richard_mawpedroalvarez: though I didn't think to check on it yesterday to see how large it was, so I don't know if it's reached its maximum09:58
richard_mawpedroalvarez: hm, no indication as to why IPv6 router discovery failed?09:59
richard_mawalso, which version of systemd?10:00
Kinnisonrichard_maw: I'd suggest we monitor it for a few more days and see if it grows more than 87M10:01
pedroalvarez"richard_maw: Address family not supported by protocol"10:02
pedroalvarezbad quoting10:02
pedroalvarezrichard_maw: "Address family not supported by protocol"10:02
pedroalvarezI didn't worry about that error, because it happens as well when I restart networkd, and network works10:03
* Kinnison wonders if pedro's kernel doesn't have ipv6 support10:04
richard_mawwouldn't surprise me10:04
richard_mawit's not on by default in ARM dev-boards10:05
KinnisonWhich considering cheap ARM devboards are an obvious IoT development platform is daft10:06
pedroalvarezit's enabled as a module10:06
pedroalvarez(this is a massive defconfig file from the raspberry pi linux repo)10:07
*** gtristan has quit IRC10:07
richard_mawpedroalvarez: the error is probably from socket(AF_INET6, SOCK_RAW | SOCK_CLOEXEC | SOCK_NONBLOCK, IPPROTO_ICMPV6);10:08
Kinnisonpedroalvarez: did you install the modules?10:08
Kinnisonpedroalvarez: and did you ensure depmod was run?10:08
* pedroalvarez runs `depmod -a`10:09
pedroalvarezthe modules are being installed10:10
pedroalvarezthat was it?10:10
pedroalvarez:)10:10
paulsherwoodKinnison: vboxfs time to git clone linux: 1m40.389s10:11
edcraggpedroalvarez: sorry i missed your pings on friday, was busy and not paying attention to irc10:13
edcraggdid you have any luck with partitioning?10:13
pedroalvarezedcragg: a lot of luck :)10:14
edcraggreally?10:14
edcraggsounds promising10:14
pedroalvarezyes, I want to get them merged :)10:14
edcraggok, excellent :)10:15
pedroalvarezI can deploy now a disk image, that you can `dd` to a sdcard and then put it on a rpi2 :)10:15
edcraggindeed, that was the idea ;)10:15
edcraggglad it worked for you10:17
edcraggi did also modify the jetson flash script to work with partitioned images, i just didn't submit it to james as it turned out to be fairly inefficient time-wise compared with the way it is now10:17
Kinnisonpedroalvarez: cool10:17
richard_mawpedroalvarez: if you've got device pass-through working, you ought to be able to install directly to the SD card, which is faster and allows the card to last longer10:18
paulsherwoodpedroalvarez: what was your trick for livestreaming?10:18
pedroalvarezrichard_maw: the jetsons have an sdcard slot :)10:18
* paulsherwood is considering creating a new network of build machines + artifact server on scaleway10:18
pedroalvarezpaulsherwood: I used hangouts10:19
pedroalvarezyou can stream directly to youtube10:19
paulsherwoodaha, ok.10:19
paulsherwood'hangouts on air has been disabled by your domain administrator'10:28
pedroalvarezhm.. I guess I used my own gmail account?10:29
paulsherwoodprobably10:30
* Kinnison notes that OBS is pretty easy to set up and use10:31
Kinnisonand does very nicely for streaming, to YT and to twitch and beam10:31
KinnisonSadly I think it can only do one target at a time10:31
*** CTtpollard has quit IRC10:31
Kinnisonso you'd need some kind of output demultiplexer10:32
jjardonpedroalvarez: I've enabled the videocore support in mesa to have 3D acceleration, so GNOME should work in the rpi2 when we merge Tristan patches; maybe worth a try :)10:32
pedroalvarezjjardon: I was about to start asking graphics questions for the Pi10:33
pedroalvarezI'd like to test if that works, today10:33
*** wschaller has quit IRC10:34
pedroalvarezjjardon: will I have to specify a backend when running weston?10:35
pedroalvarezmeh, I will try :)10:35
franredpedroalvarez, other test would be check the XFCE system, if it is still alive and/or usable10:36
jjardonxfce has been fixed recently; and it doesn't need 3D support so it should work10:38
pedroalvarezmy goal is to have Genivi Demo Platform on it10:40
pedroalvarezso, weston/wayland10:40
pedroalvarezjjardon: is the patch to enable videocore in gerrit? or has already been merged?10:43
jjardonpedroalvarez: its already there (vc4 in the mesa chunk)10:47
pedroalvarezaha! I was grep'ing videocore10:48
*** wschaller has joined #baserock10:48
pedroalvarezthis is good!10:48
*** CTtpollard has joined #baserock11:08
pedroalvarezjjardon: so... if I run weston, it tries to use the drm backend11:15
pedroalvarezjjardon: if I use the fbdev backend, it works, but maybe a bit slow11:15
pedroalvarezbut it works!!!! \o/11:17
CTtpollard:D11:19
*** franred has quit IRC11:20
paulsherwoodw00t!11:21
*** rdale has joined #baserock11:23
*** franred has joined #baserock11:24
pedroalvarezhm.. "rpi-backend.so"..?11:29
*** tom__ has joined #baserock11:32
*** CTtpollard has quit IRC11:36
*** wschaller has quit IRC11:44
*** zoli_ has quit IRC11:45
*** wschaller has joined #baserock11:56
*** zoli_ has joined #baserock12:04
pedroalvarezI don't understand why it's not being created12:16
*** tom__ is now known as CTtpollard12:54
*** wschaller has quit IRC13:04
*** wschaller has joined #baserock13:17
jonathanmawwell, it seems ybd didn't like following a symlink to another disk for its cachedir13:20
jonathanmawthe good news is it's building now, the bad news is it's building from scratch13:20
paulsherwoodjonathanmaw: check the top of its output... has it actually picked up your config?13:23
jonathanmawpaulsherwood: it's not caching all the gits again, and it's logging the build commands to a place on /src, so that's a win13:25
paulsherwoodjonathanmaw: but not seeing your artifacts???13:25
jonathanmawpaulsherwood: I don't think I managed to build any artifacts before my build got stuck, though it did manage to cache all of them, I think13:26
paulsherwoodif they're there, it shouldn't need to rebuild them?13:26
jonathanmaws/them/the git repos/13:26
paulsherwoodah. i'm tyalking artifacts13:27
jonathanmawWould there be a kbas server I can point my build to?13:27
paulsherwoodsoon13:28
paulsherwoodjonathanmaw: http://212.47.227.68/artifacts/*13:31
paulsherwoodit has some things building stuff, but it will take a while to populate13:31
pedroalvarezOk, i believe that rpi-backend.so is not being installed because weston can't find the Raspberry Pi BCM headers at build time13:47
pedroalvarezhm...13:47
pedroalvarezI don't want to have weston depending on the BSP13:47
*** gtristan has joined #baserock13:47
richard_mawif the headers are stable, then you could just copy them into a branch of weston13:48
pedroalvarezthat is actually a good idea13:48
pedroalvarezhm.. not sure13:57
pedroalvarezmaybe I can just remove the check: http://paste.baserock.org/ekunenirug13:57
pedroalvarezCopy all of these files to $DESTDIR: https://github.com/raspberrypi/firmware/tree/master/hardfp/opt/vc14:00
pedroalvarezand then generate a .pc file manually14:00
pedroalvarezi don't know how many of these blobs are actually needed :/14:01
*** wschaller has quit IRC14:06
*** bashrc has quit IRC14:08
*** bashrc has joined #baserock14:10
pedroalvarezOh, I think I can build it from source :) https://github.com/raspberrypi/userland14:12
paulsherwoodjonathanmaw: would you like to upload to that, as well as download?14:15
jonathanmawpaulsherwood: is that done by setting serve-artifacts: True in ybd.conf?14:16
pedroalvarezbtw: ERROR: Conflicting fields "morph, and build-system" for glib in stratum strata/glib-common.morph.14:17
pedroalvarezthis is Morph complaining, I guess 'ybd' doesn't fail with that14:17
paulsherwoodjonathanmaw: no, it's done by setting kbas-password in ybd.conf14:18
jonathanmawpaulsherwood: I take it your kbas server isn't using the default kbas-password, then14:18
paulsherwoodjonathanmaw: no. if it did, nothing would be able to upload. the code expressly refuses to work if that's the password14:18
ssam2pedroalvarez: in master? oops14:20
ssam2the error makes sense -- if you specify a predefined build system and a chunk .morph file with custom build commands, one or other will be ignored14:20
ssam2which is a potential source of "why the hell are you ignoring what i told you" errors14:21
pedroalvarezyes, makes sense to me too14:21
*** wschaller has joined #baserock14:21
richard_mawdoes it? won't that prevent overriding a build-system with custom build commands?14:21
richard_mawor am I misunderstanding this14:21
paulsherwoodssam2: iirc there were several of those... has someone changed morph in this regard?14:21
ssam2only if the predefined build system is specified in the stratum14:21
ssam2paulsherwood: not sure what you mean, sorry14:21
ssam2if you set build-system in the chunk .morph file there's no error14:22
paulsherwoodaha, that's what i meant14:22
ssam2ok :)14:22
pedroalvarezthis is for "build-system" in the stratum morphology14:22
pedroalvarezyou can either set build-system, or morph, to point to a chunk morphology, in where you can define build-system too14:23
tiagogomes_pedroalvarez are you fixing that, or should I fix it?14:29
pedroalvareztiagogomes_: please, go ahead. I have my brain in this rpi fight..14:30
paulsherwoodis this a recently introduced problem?14:32
paulsherwoodmason seemed to work14:32
pedroalvarezintroduced here: https://gerrit.baserock.org/#/c/1167/1/strata/glib-common.morph14:32
pedroalvarezpaulsherwood: I guess mason is building a previous commit, but will eventually fail14:33
paulsherwoodaha14:34
tiagogomes_https://gerrit.baserock.org/#/c/1189/14:38
jjardonpedroalvarez: for the drm part, you will need this: https://github.com/anholt/linux/tree/vc4-kms-v3d14:39
jjardonpedroalvarez: does the xwalyand backend works in weston?14:40
pedroalvarezerr, that fork is "26275 commits ahead"14:42
paulsherwoodouch14:42
pedroalvarezbut it might be the good one to get this working, idk14:42
pedroalvarezjjardon: havent tested xwayland, nope14:43
pedroalvarezjjardon: weston is built with "--disable-xwayland"14:43
jjardonpedroalvarez: I think you are using an old definitions14:44
pedroalvarezjjardon: so, you mean that with that kernel, I can use the drm-backend?14:44
pedroalvarezjjardon: nope, this is weston-genivi14:45
jjardonah, ok14:45
jjardonis weston-genivi a fork of weston?14:45
pedroalvarezit looks like that, with some patches to get the ivi-shell working14:46
pedroalvarez(iirc)14:47
jjardonpedroalvarez: about the drm question: yes, I think so. More info here: http://dri.freedesktop.org/wiki/VC4/14:48
CTtpollardVC4 is used for the meta-raspberry pi fork from Tizen14:49
pedroalvarezI think this is the right branch: https://github.com/anholt/linux/tree/vc4-kms-v3d-rpi214:49
pedroalvarezCTtpollard: good to know, can you point me to that repo?14:50
*** tiagogomes_ has quit IRC14:50
CTtpollardpedroalvarez: git://git.s-osg.org/meta-raspberrypi.git -b 3d_accel_vc414:50
pedroalvarezThanks! It's good to have all of these info14:54
pedroalvarezI don't quite understand why, with all the people using RPi's, these patches are not in mainline14:58
paulsherwoodmaybe bjdooks could explain?14:59
* gtristan gets bombarded with 131 gerrit emails15:03
*** tiagogomes_ has joined #baserock15:05
ssam2gtristan: it does that :(15:06
paulsherwoodjonathanmaw: am i right your machine is building gcc?15:07
jonathanmawpaulsherwood: yep15:07
paulsherwoodit doesn't seem to be uplaoding anything... is that expected?15:07
jonathanmawIt seems I failed to upload m4-tarball when I was building that15:08
jjardonpedroalvarez: broadcom used to provide a binary blob; then broadcom hired anholt to develop a open source driver; so probably is not fnished yet15:08
gtristanouch15:08
jonathanmawI suspect the asterisk at the end of the kbas-url shouldn't have gone in the config15:08
paulsherwoodkbas-url should be http://212.47.227.68/15:09
jonathanmawpaulsherwood: no /artifacts?15:09
paulsherwoodcorrect15:09
paulsherwoodi suggest let it finish gcc15:10
jjardongtristan: hey! did you file a bug against accountmanager or should I?15:10
paulsherwoodit seems to be  *way* faster than the scaleway machines, though15:10
gtristanthis: https://gerrit.baserock.org/#/c/1174/2//COMMIT_MSG is certainly not the commit message, but some comment I made on gerrit instead :-/15:10
gtristanjjardon, it's past midnight, no I did not, I guess you mean accountsservice which probably needs the same systemd configure.ac patch15:11
pedroalvarezgtristan: looking at the patch, Sam updated the commit msg15:12
gtristanpedroalvarez, in any case looking at my original, it was not very verbose ... as tiagogomes_ points out I could have mentioned there that it was a dependency of udisks15:13
jjardongtristan: ok, dont worry15:14
paulsherwoodjonathanmaw: working now i see15:14
paulsherwoodi wonder if the 9 will ever catch up15:14
*** zoli_ has quit IRC15:15
jonathanmawpaulsherwood: I had cancelled the gcc build, so when I started it up again, random chance picked something else. Now I'm back to gcc15:15
ssam2gtristan: I edited that commit message, sorry15:15
jonathanmawso perhaps the 9 will catch up after all.15:15
ssam2gtristan: I just didn't want the info to only be present in a Gerrit comment where nobody will find it15:16
gtristanssam2, if nobody is going to see the conversation in a bug report or a gerrit review which justifies an added/modified line of code, this is a bit of a sad state of affairs :-/15:17
gtristanssam2, anyway no worry, I was a bit alarmed because I'm not used to gerrit, and I was worried that it auto-did-something-intrusive15:18
gtristanssam2, question, looking at: https://gerrit.baserock.org/#/q/topic:startup-gnome-shell... I see merge conflicts...is that really the case ?15:19
gtristanI'm absolutely sure that if the branch were to be applied in the submitted order, there would be none15:19
pedroalvarezgtristan: there was a patch merged recently that maybe have introduced some conflicts15:23
pedroalvarezoh, no, it hasn't been merged yet15:23
pedroalvarezhmm,.. sometimes gerrit su**s15:24
pedroalvarezthis is because we have modified the commit msg of one commit15:24
pedroalvarezcausing a new version of this patch15:24
pedroalvarezand breaking the dependencies of the patches that depended on the previous version of the patch15:25
pedroalvarezbut for this, the gerrit UI has a Rebase buttom15:25
pedroalvarezs/m$/n/15:26
* gtristan clones a fresh definitions and sees half of the patches applied15:26
gtristanehh...15:27
* gtristan guesses he'll have some overhead to burn hours on tomorrow15:27
ssam2gerrit says 'merge conflict' when it means 'this depends on another change that isn't merged yet'15:28
ssam2but it also says 'merge conflict' when there's an actual merge conflict15:29
gtristangerrit could wait to merge the branch properly only when all of the patches are approved15:29
ssam2Gerrit was basically designed with no support at all for change dependencies, and it shows15:29
ssam2it has at least some sponsored development I think via Android, maybe they could implement that15:30
ssam2it does seem like a pretty obvious improvement15:30
gtristansorry for showing frustration - seems like a huge timesink, coupled with a "add discussion to comments so they dont get lost in gerrit" comment, one has to wonder what's the point15:30
*** paulwaters_ has quit IRC15:30
ssam2that sort of patch series is sort of the worst case for Gerrit. I find it nicer than mailing list review in some ways15:34
tiagogomes_and gerrit doesn't allow to test the whole branch at once, unless I cherry-picked 15 commits15:34
gtristanmailing list reviews dont match up to proper bug tracking that's true15:34
pedroalvarezERROR: In stratum baserock:baserock/definitions|68fa5ed3acd71a70a5f6849a92be15076deef495|strata/x-common.morph|x-common-devel, chunk xorg-lib-libXi references its dependency xorg-lib-libXfixes before it is defined15:35
ssam2tiagogomes_: `git-review -d top-change-of-branch`15:35
* tiagogomes_ needs to install that git-review thingy15:35
pedroalvarezI'm worried that contributors using YBD might break the definitions for Morph :/15:36
pedroalvareztiagogomes_: It's in baserock15:37
pedroalvarezyou might have it already if you are using a baerock devel VM15:37
tiagogomes_pedroalvarez yep, I've it, there are no excuses now15:37
gtristannod, that would be me - I guess the hot question is: Is it against the spec of the .morph to reference before definition ? and then the bug is either with YBD or it is with Morph15:38
pedroalvarezreally hot15:39
gtristan:)15:39
pedroalvarezI'd like to separate the morphologies validation15:39
pedroalvarezto a library maybe that both can use?15:39
* tiagogomes_ would like they battled until only one stand15:43
*** franred has quit IRC15:45
pedroalvareztiagogomes_: :)15:45
pedroalvarezI don't like fights tbh15:46
pedroalvarezwe could be hacking on the same thing together, having extra time for focus on other things15:51
*** wschaller has quit IRC15:54
paulsherwoodhmmm chroot: failed to run command ‘/bin/bash’: No such file or directory15:54
paulsherwoodeven though `which bash` returns /bin/bash15:54
*** nowster has quit IRC15:55
pedroalvarezpaulsherwood: what exactly are you trying to do?15:55
ssam2pedroalvarez: we need pre-commit automated testing, still15:55
tiagogomes_/bin/bash needs to be inside of the chroot dir. Eg./home/me/mychootdir/bin/bash15:55
ssam2pedroalvarez: at least we have post commit testing!15:56
*** mariaderidder has quit IRC15:56
ssam2moving the validation stuff into a separate lib is a great idea though15:56
ssam2except that i think paulsherwood would not use it :-)15:56
paulsherwoodpedroalvarez: i'm trying to follow james' chroot instructions http://listmaster.pepperfish.net/pipermail/baserock-dev-baserock.org/2015-August/013252.html15:56
paulsherwoodssam2: you mean validation of definitions? would it require a boatload of morph code?15:57
pedroalvarezpaulsherwood: check what tiagogomes_ said15:58
paulsherwoodinteresting... i wonder how it worked for james, then?15:59
pedroalvarezssam2: validation and parsing, etc :)16:00
paulsherwoodin my chroot there's ash, no bash16:00
pedroalvarezpaulsherwood: ash? sh?16:00
ssam2paulsherwood: it can be done with any code, as long as it validates .morph files according to the agreed spec16:00
paulsherwoodssam2: you mean like the jsonschema approach, or your work with owl?16:01
pedroalvarezpaulsherwood: if 'sh' you can do `chroot . /bin/sh` I believe16:01
paulsherwoodpedroalvarez: i'm trying '/usr/bin/linux32 -B /src/armv7chroot/launch.sh'16:03
pedroalvarezpaulsherwood: do you have "/src/armv7chroot/bin/bash" ?16:04
pedroalvarezor "/src/armv7chroot/bin/sh" ?16:04
paulsherwood/src/armv7chroot/bin/sh16:04
pedroalvarezthen change /src/armv7chroot/launch.sh to be like: http://paste.baserock.org/usabomogey16:05
paulsherwoodpedroalvarez: ah, sorry. i'm being an idiot. i didn't see the bash reference16:06
paulsherwoodtvm16:06
pedroalvareznp, does that work?16:07
paulsherwoods/bash/chroot/16:07
paulsherwoodbetter, certainly :)16:07
pedroalvarez /bin/bash should be in the system :/16:08
ssam2paulsherwood: jsonschema would be able to do a lot of what Morph currently implements in Python16:09
ssam2although jsonschema would probably give a lot more unhelpful error messages than Morph does16:09
*** nowster has joined #baserock16:09
paulsherwoodi'd be surprised if that's true :)16:10
*** ctbruce has quit IRC16:17
*** ssam2 has quit IRC16:22
richard_mawthe thought occurs, that mkfs.btrfs includes --rootdir for pre-populating a disk image's rootfs, so if it was smart enough to detect that the files it's initializing the disk image with and the disk image itself are on the same btrfs file system, you could do a zero-copy disk image creation16:27
richard_mawat which point you could use virtual-machines for your builds, providing much greater isolation16:27
richard_mawthough if you wanted to have full zero-copy, you'd need to make a corresponding tool for extracting a tree from a disk image as well16:29
richard_mawwhich is close to what btrfs' rescue command is for16:30
pedroalvarezjjardon:  drm backend still not working with that kernel version16:42
* paulsherwood now has 9 scaleway machines, and 5 moonshot cartridges, all building definitions armv716:44
*** jonathanmaw has quit IRC16:45
paulsherwood(and a cache server they're all trying to populate)16:46
jjardonpedroalvarez: :( any helpful message in the logs?16:59
pedroalvarezjjardon: haven't enabled logging :/16:59
pedroalvarezI was trying to figure out how16:59
jjardonFor weston I think is --log=file.log17:02
*** bashrc has quit IRC17:02
pedroalvarez[23:08:32.155] no drm device found17:03
pedroalvarez[23:08:32.156] fatal: failed to create compositor17:03
paulsherwood:/17:03
pedroalvarezalright, I'm missing a couple of things17:06
paulsherwoodhmmmm.... ybd is not happy in the chroot... http://paste.baserock.org/vozidinoxu17:07
paulsherwoodthis mknod thing has to go17:08
pedroalvarezI'd guess it's needed :)17:08
paulsherwoodi guess so too17:09
* pedroalvarez sends https://gerrit.baserock.org/#/c/1190/17:22
pedroalvarezMason will crash without it :)17:25
paulsherwoodbecause of ordering?17:26
*** edcragg has quit IRC17:35
pedroalvarezYup17:37
pedroalvarezI proposed the idea of having the same library for ybd an morph to parse definitions because of that error17:40
pedroalvarezAs I said, I'm worried that people using only ybd send patches that are not morph compatible17:41
*** toscalix__ has quit IRC18:23
*** paulwaters_ has joined #baserock18:41
*** paulwaters_ has quit IRC18:45
*** zoli_ has joined #baserock19:58
*** zoli_ has quit IRC20:02
*** zoli_ has joined #baserock20:27
*** zoli_ has quit IRC20:28
jjardonHi, any idea what version of morph is being used to generate the cache?21:17
*** lachlanmackenzie has quit IRC21:21
pedroalvarezThe one used on the last release. :)22:54
pedroalvarezWhenever we do a release, we upgrade the version of Mason's22:54
*** lachlanmackenzie has joined #baserock23:12

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