=== marcoceppi_ is now known as marcoceppi
=== chihchun_afk is now known as chihchun
=== alecu_ is now known as alecu
=== chihchun_afk is now known as chihchun
oobartezwhat's the status of snappy becoming the default package manager in Ubuntu Desktop? anywhere on the net I could find some up-to-date info?06:30
oobartezI'm pretty excited about the idea but it's difficult to find a central point with current status, or info on how to get involved in this particular project06:42
ollioobartez, it's currently not plan of record for Ubuntu Desktop07:16
ollionce 16.04 is out we might start working on a modern rendition of Ubuntu, which will incorporate technologies from the phone and be snappy based, but there is no specific & detailed plan yet07:17
oobartezok, I understand it's a long haul, in the mean time I also found this: http://ubuntuforums.org/showthread.php?t=2312045&p=13432475#post1343247507:24
oobartezstill, I personally find it super interesting. are there any conceptual discussions going that I could dig through?07:24
anpok_hm where is an up to date documentation on writing an oem snap?07:33
anpok_the doc in developer.ubuntu.com/en/snappy/guides/oem differs from examples in http://bazaar.launchpad.net/~snappy-dev/snappy-hub/snappy-systems/files07:35
anpok_is it gadget or oem currently?07:38
anpok_whats the immutable-config good for?07:38
=== dvladar_ is now known as dvladar
dholbachgood morning08:07
JamesTaitGood morning all!  Happy Thursday, and happy Create a Vacuum Day! 😃09:33
anpok_oem_snap/bananapi_1.0_all.snap failed to install: Signature verification failed: No signatures, or no origin signature. (exit code 10)09:41
anpok_trying to create a new image using a just recently created oem snap..09:41
=== kickinz1 is now known as kickinz1-sprint
=== kickinz1 is now known as kickinz1-sprint
noizerHi, I want to make some services in snapcraft 2.1 but when I do that it tells met services was unexpected10:54
noizerthats a part of my yaml10:56
noizerany clues10:56
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
dholbachhey beuno, do you know when the new reviewers tools will land in myapps? I think all new snaps are being rejected right now? at least the one I uploaded with 'snapcraft upload' got rejected.11:51
beunodholbach, hi. Rejected with what message?11:58
dholbach - meta/readme.md does not exist lint_snappy_readme.md11:59
dholbach - (NEEDS REVIEW) squashfs pkg lint_is_squashfs11:59
dholbach - unknown entries in package.yaml: 'apps,description,summary' lint_snappy_unknown11:59
dholbachwith the most recent reviewers tool I think most of the above are fixed11:59
dholbachbeuno: ^12:00
beunodholbach, the readme is fixed, but I don't think squashfs verification has landed yet, has it?12:02
sergiusensdholbach, hmm, there shouldn't be a package.yaml12:03
dholbachsergiusens: lp:~dholbach/+junk/moon-buggy is what I tried to upload with 'snapcraft upload'12:04
beunodholbach, I think we'll need a week or so more until 16.04 snaps auto-approve12:06
dholbachauto-uploaded auto-approving snaps would be something really nice to demo :)12:09
beunodholbach, yes indeed  :)12:15
sergiusensdholbach, maybe clean and rebuild, I don't see package.yaml in the latest snapcraft builds12:15
kyrofaGood morning12:15
sergiusenskyrofa, morning12:16
noizershort question i want to make services in snapcraft 2.1 but it seems i don't work with the normal services tag12:16
kyrofanoizer, no, the syntax has changed in version 212:17
noizerkyrofa ok where can I find the syntax then?12:17
kyrofanoizer, https://github.com/ubuntu-core/snapcraft/blob/master/docs/snapcraft-syntax.md12:18
dholbachsergiusens, kyrofa: I'm still seeing "sh -c /usr/lib/update-notifier/update-motd-updates-available 2>/d..." as a hanging process12:18
sergiusensdholbach, I have no idea what you are talking about12:19
noizerkyrofa how does it work with the ports then?12:19
dholbachsergiusens: http://paste.ubuntu.com/14876880/12:19
sergiusensdholbach, in what point of snapcraft does it hang?12:20
kyrofanoizer, what do you mean?12:21
sergiusensdholbach, or at what point is it called12:21
dholbachsergiusens: that's in the 'pull' stage12:21
noizerI need something with sockets but then i tought i need services. So services with the previous version there you can open some ports. But is it possible now with the deamons?12:21
sergiusenspindonga, question about the store; every time we upload with the store API through snapcraft  we have to re-set the supported releseases, is there a way to make that stick?12:22
dholbachsergiusens: first apt-get update is called, then something calls /usr/lib/update-notifier/update-motd-updates-available12:22
dholbachbasically after "apt update" pull hangs12:23
kyrofanoizer, sure it is. You just need to set the right permissions. If you're binding to a port you'll need network-service12:23
noizerok i will look in to it :p12:24
ogra_mvo, somehow /etc/writable stopped working with the switch to squashfs12:25
sergiusensdholbach, so if it is when installing build packages, we don't call update https://github.com/ubuntu-core/snapcraft/blob/master/snapcraft/repo.py#L6012:25
sergiusensdholbach, but we do call it here: https://github.com/ubuntu-core/snapcraft/blob/master/snapcraft/repo.py#L23212:25
sergiusensdholbach, nothing fancy though, just using python apt in a different rootdir (this will go away soon for everything but ros)12:26
dholbachsergiusens: when the apt lists are obtained in any case12:26
sergiusensdholbach, yeah, you hit this when adding stage packages12:27
dholbachI wonder how we can turn off bits like update notifier12:27
dholbachit must be possible12:27
sergiusensdholbach, maybe mvo knows12:27
dholbachmvo: do you know an apt config option which would turn off update-notifier scripts and stuff (when using a local sources.list)12:29
noizerkyrofa is there already documentation of network-services?12:29
sergiusensdidrocks, hello there! Was there any issue with the upload? If you are too busy I can bother dholbach :-)12:31
dholbachwhat what?12:32
sergiusensdholbach, I just distro patched someting until a release is avail http://paste.ubuntu.com/14871051/12:32
sergiusensfor a broken package12:32
kyrofanoizer, check out this example: https://github.com/ubuntu-core/snapcraft/blob/master/examples/shout/snapcraft.yaml12:33
ogra_mvo, oh, ignore me ... /etc/writable is fine, it is the atomic write issue once again12:35
noizerkyrofa thx12:35
dholbachsergiusens: on it12:36
didrockssergiusens: please bother him, yeah! here, it's quite crazy :p12:37
kyrofanoizer, does that make sense? You're just saying "my service needs to listen on the network"12:38
noizerI think it make sense12:38
noizeri will try it12:38
noizerbecause when when I'm testing it with normal command the permission was denied kyrofa12:39
noizer(tornado sockets)12:39
kyrofanoizer, do you know about snappy-debug?12:40
noizeri did not used it before12:40
kyrofanoizer, this may be helpful, from our brilliant dholbach: https://github.com/ubuntu-core/snapcraft/blob/master/docs/debug.md12:40
noizerok pray all dholbach :D12:41
sergiusensdidrocks, I imagined as much12:42
sergiusensdholbach, thanks !!!12:42
* sergiusens hugs dholbach 12:42
* dholbach hugs sergiusens back12:49
dholbachkyrofa, noizer: to be fair... the content was written mostly by Jamie IIRC :)12:50
=== bpierre_ is now known as bpierre
=== JamesTai1 is now known as JamesTait
mvodholbach: you can clean the dpkg::post-invoke, is snapcraft using "import apt"? if so I can provide an example in a bit (meeting)12:59
kyrofamvo, yessir it is13:00
mvokyrofa: please try apt.apt_pkg.config.clear("dpkg::post-invoke")13:03
anpokis there a way to use an oem snap with ubuntu-device-flash without signing it?13:03
noizerkyrofa what does the type migration-skill?13:04
kyrofaThanks mvo! dholbach it sounds like you have a good failing test case, think you can throw that in and give it a try?13:06
kyrofanoizer, so for 16.04 the ability to share things between snaps is calls "skills"13:06
noizeroooh ok :D13:07
kyrofanoizer, there's still in development, so right now there's only a single skill called the "migration-skill" that essentially offers backward compatibility with the old stuff13:07
noizerkyrofa ok thx :D13:07
kyrofanoizer, thought not really backward as it changes the yaml anyway... but you get the idea13:07
kyrofanoizer, there have been a few threads on snappy-devel regarding this. If you're using 16.04 snappy you might consider getting on there so you know what's happening :)13:11
noizerkyrofa ok i will check it out today :D. first, I will take care of my application so it is working on snappy13:12
dholbachkyrofa: I'm not quite sure where to throw it in to make it work - I tried in a couple of places, but it doesn't seem to have an effect13:19
dholbachI tried L60 and L232 in snapcraft/repo.py, tried after the apt import and elsewhere13:19
mvodholbach: just anywhere in the code, its a global option13:19
mvodholbach: having said this, I have no idea if what I said actually made sense, maybe I should read backlog first ;)13:20
dholbachmvo: /usr/lib/update-notifier/update-motd-updates-available is still being called13:20
dholbachwhich in our context hangs (as non-root user)13:20
dholbach(mv /tmp/tmp.HSo85YHW5T /var/lib/update-notifier/updates-a...)13:20
mvodholbach: hm, the apt.apt_pkg.config.clear() should prevent this. are there more ways that snapcraft drives apt in addition to the import apt?13:22
dholbachmvo: these are all notable occurences: http://paste.ubuntu.com/14877293/13:23
mvodholbach: the subprocess call there looks suspicious :)13:29
mvodholbach: is that were it crashs and burns?13:29
dholbachI think it's before: http://paste.ubuntu.com/14877327/13:30
dholbachthat's where it tries to install build-packages which is not my local case13:30
mvodholbach:  apt_cache = apt.Cache(rootdir=rootdir, memonly=True) will reset the config, could you please stick the clear() after thi sline?13:34
dholbachmvo: no dice13:36
dholbachprinting out apt.apt_pkg.config.get("dpkg::post-invoke") before and after the clear() gives nothing13:36
dholbachmvo: ah ok - I used .value_list now13:40
dholbachit's "['adequate --help >/dev/null 2>&1 || exit 0; exec adequate --debconf --user nobody --pending', 'if [ -d /var/lib/update-notifier ]; then touch /var/lib/update-notifier/dpkg-run-stamp; fi']13:40
dholbach" before the .clear()13:40
dholbachand [] after13:40
dholbachcould it be that the motd bits are somewhere else?13:41
mvodholbach: hm, one sec13:44
dholbach /etc/apt.conf.d/99update-notifier:APT::Update::Post-Invoke-Success {"/usr/lib/update-notifier/update-motd-updates-available 2>/dev/null || true";};13:45
mvodholbach: try APT::Update::Post-Invoke-Success:13:45
mvodholbach: try APT::Update::Post-Invoke-Success13:45
mvodholbach: yeah, if you clean that, it should be ok13:45
* dholbach crosses fingers13:45
dholbachcome on apt, make me happy :)13:45
ogra_what a day ... so much ap5t talk in the snappy channel13:46
kyrofaogra_, hahaha13:46
dholbachsergiusens_, kyrofa: http://paste.ubuntu.com/14877432/13:46
dholbachI need to run for a bit, so feel free to push this one13:46
kyrofadholbach, thanks for that!13:47
sergiusens_anpok, with --developer-mode iirc13:47
sergiusens_dholbach, mvo thanks13:48
dholbachsergiusens_: I did a fresh build and upload of moon buggy - still the same errors/warnings from the reviewers tools13:48
dholbach1) meta/readme.md does not exist lint_snappy_readme.md - 2) (NEEDS REVIEW) squashfs pkg lint_is_squashfs - 3) unknown entries in package.yaml: 'apps,description,summary' lint_snappy_unknown13:49
dholbachreally need to run now - see you later13:49
sergiusens_dholbach, oh, now I recall mvo saying he mixup the error message; it should say snap.yaml there;13:52
sergiusens_there is still support required for those tools though13:52
mvodholbach: yeah, the store is a bit confused, the click-reviewers-tools are not fully updated yet13:52
mvodholbach: eh, s/store/c-r-t/13:52
didrockssergiusens_: I have some snapcraft.yaml which just download debs and a shell script. I wonder if I can say "please create a armhf .snap" (with snapcraft1) easily?13:53
pindongasergiusens_, is this on subsequent uploads for the same pkg? ie, just new versions?13:53
didrocksas I have no compilation at all13:53
pindongasergiusens_, this == having to re-set the releases13:54
kyrofadidrocks, the problem will be pulling the .debs from the arm repos13:54
didrockskyrofa: exactly!13:54
kyrofadidrocks, pretending you didn't have that to worry about, you can actually put the .snap together yourself and just run snapcraft snap <directory>13:55
kyrofadidrocks, but I'm not sure there's a good way to get the .debs without running on target or hacking snapcraft13:55
didrockskyrofa: sounds like an option. I'm looking as well in the code if I can spoof your arch detection :p13:55
didrocks(if you are using a chroot for download from the archive)13:55
didrocksor are you using the system sources.list?13:55
kyrofadidrocks, heh, going the hacking snapcraft route eh?13:56
kyrofadidrocks, hold on, let me find you a link13:56
didrockskyrofa: seems to be in snapcraft/common.py13:56
didrocksheh 'dpkg-architecture', '-qDEB_BUILD_ARCH'13:56
didrocksand the arch triplet13:56
kyrofadidrocks, check out snapcraft/repo.py13:57
didrocksah nice, it seems you not relying on system sources.list13:57
didrocksbut rather building one13:58
kyrofadidrocks, right. stage-packages and build-packages are different13:58
kyrofadidrocks, you want the stage-packages stuff, which is repo.py13:58
didrockskyrofa: thanks for the pointer! going to mess from there! :-)13:59
kyrofadidrocks, good luck! Interested to hear how it goes14:00
didrocksyeah, well… snapcraft1 and it's changing with the classic dimension then14:00
kyrofadidrocks, remember you can always download the .debs yourself and unpack them14:00
didrocksso, more of a stop gap :)14:00
didrocksin download/ from the part14:01
anpoksergiusens_: ah that worked .. now it cannot find the package.yaml in the gadget snap14:01
didrocksbut 134 of them14:01
didrocksso getting lazy! :)14:01
kyrofadidrocks, ugh!14:01
sergiusens_anpok, I'll refer to mvo for that14:01
sergiusens_pindonga, yes, new versions of the same package14:02
pindongasergiusens_, I guess we could default to any release set for the latest published version14:03
pindongasergiusens_, could you file a bug on lp:software-center-agent please?14:03
beunoor maybe store it locally?14:03
beunoin a local config14:03
sergiusens_didrocks, your code is old, we removed the dep on dpkg-dev hence we don't have dpkg-architecture anymore either14:03
beunolast release targeted?14:03
sergiusens_pindonga, I agree on that14:04
sergiusens_beuno, well I was asking about the store api at first ;-) Since I need to add 'release' to snapcraft.yaml to support cleanbuild anyways14:04
beunosergiusens_, I worry a bit about the server auto-filling data14:05
beunobut only a bit14:05
anpoki creates a gadget/oem? snap that has a meta/package.yaml .. u-boot binary .. now it ubuntu-device-flash fails with 'Determining oem configuration\nno oem package found'14:05
pindongabeuno, this being for new version uploads only, I thought relying on existing releases if none is specified was ok14:06
pindongaof course if the pkg contains the releases in the metadata, we should grab it from there14:06
beunoit's probably ok, yeah14:07
beunoit just feels like the wrong place to define intention14:07
=== alexabreu is now known as alex-abreu
sergiusens_beuno, it's not autofilling14:12
sergiusens_beuno, or not less than it used to14:12
sergiusens_beuno, if I upload the snap from the ui it is filled in with the previous stuff; at least it was a week a go14:12
sergiusens_beuno, I don't mind if it is an api; maybe this can also play with the stability level (grade) in snap.yaml14:13
didrockssergiusens_: it's snapcraft v114:14
sergiusens_beuno, I'd much like to define which pockets to upload to, something like 'snapcraft upload --releases ubuntu-core,ubuntu-personal,...'14:14
didrockssergiusens_: from git14:14
sergiusens_didrocks, oh, right14:14
didrockssergiusens_: we have to use it for mwc :p14:14
=== sergiusens_ is now known as sergiusens
awe__mvo, I tried installing hello-world.mvo on my new amd64 rolling image and it fails.  Is this a known issue?14:16
kyrofadidrocks, oh yeah, that reminds me-- snapcraft snap <directory> is in version 2. If you're using 15.04 that functionality still exists in snappy itself, which you can use via snappy build14:16
awe__sergiusens, I managed to get everything re-worked last night14:17
kyrofadidrocks, works the same way though14:17
awe__one question... since I can't get hello-world working, what's the name of the user-specific writable env var?14:18
kyrofaawe__, $SNAP_USER_DATA14:18
kyrofa(if 16.04)14:18
awe__great, thanks14:18
awe__did SNAP_APP_DATA_PATH also get shortened?14:18
awe__( for 16.04 )?14:19
kyrofaawe__, $SNAP_APP_USER_DATA_PATH if 15.0414:19
kyrofaawe__, indeed, $SNAP_DATA14:19
ogra_SNAP_DATA ?14:19
* ogra_ still cant memorize them 14:19
awe__is the old env var still supported though in the latest images?14:19
kyrofaogra_, it's okay, me too14:19
kyrofaawe__, yes, just deprecated14:19
awe__ok, great; will change now14:19
sergiusensawe__, I guess email subjects are a thing of the past; https://lists.ubuntu.com/archives/snappy-devel/2016-January/001392.html :-)14:22
ogra_ppisati, mvo, i'm nearly readey witrh my rpi firmware update for gadget and oem snap .... i was wondering if we should perhaps bump the default kernel version to the latest xenial one though (which means i'd need to update the dtbs in the snap)14:22
sergiusensawe__, what fails with hello-world? If it is installation, try snappy install hello-world/edge14:22
ogra_(that is ... 4.2 to 4.3)14:23
awe__sergiusens, ok, one sec14:23
ppisatiogra_: +114:23
ogra_ppisati, any possible regressions (i.e. config changes) i shold be aware of ?14:23
kyrofamvo, did you ever learn anything about that weird systemd %h thing?14:24
awe__sergiusens, that works14:24
awe__mvo, never mind my last statement about hello-world14:25
kyrofamatiasb, I uploaded another version of that owncloud snap, but I'm not seeing it in snappy again. When you get a sec would you mind patching that framework thing on it?14:26
matiasbkyrofa, ack, sure, in a bit14:26
ppisatiogra_: nothing on purpose at least :)14:27
ppisatiogra_: let me check one thing14:27
ogra_ppisati, heh, ok14:27
kyrofamatiasb, thank you!14:27
matiasbkyrofa, fyi, we expect to get a fix for this deployed today, hopefully14:27
kyrofamatiasb, excellent! Thanks for letting me know :)14:27
anpokthe problem seems to be that the oem/gadget snap looks different.. when in stalled in temp the package.yaml ends up being here:14:31
pindongasergiusens, +1 to the idea of passing releases via api14:31
anpoks/in stalled/installed14:32
matiasbkyrofa, updated, you should see latest revision now14:33
kyrofamatiasb, perfect, works!14:37
diwicI rebooted - which apparently installed some new rootfs? - and now pulseaudio doesn't start:14:37
kyrofadidrocks, try the new version of owncloud.kyrofa when you get a sec, app store should be working14:38
ppisatiogra_: http://paste.ubuntu.com/14877727/14:38
ppisatiogra_: that's the config diff between latest 4.2 and x/master 4.3 raspi2 kernel14:39
ppisatiogra_: other than that, it works fine on my board and in my test14:39
ogra_ppisati, do you plan to bump to 4.4 before final ?14:39
ppisatiogra_: we make use of =m as mush as possible, so it should be fine14:39
ogra_(giving us the free graphics driver)14:39
ppisatiogra_: we already have a 4.414:39
ppisatiogra_: but it's the first cycle14:39
ppisatiogra_: so it's marked as unstable14:40
ogra_not on rpi yet14:40
ogra_ah, k14:40
ppisatiogra_: yep, on rpi2 too14:40
ogra_well, but you plan to bump it ?14:40
ppisatiogra_: ys14:40
ogra_ppisati, hmm, then i'll probably push the update til 4.4 is consumable14:40
diwicthe wrapper script claims it cannot find the executable14:40
ogra_and stay with 4.2 til then14:41
ppisatiogra_: probably yes14:41
ppisatiogra_: i really like X because we syncronized with an upstream LTS kernel14:41
ppisatiogra_: so align nicely14:41
ppisatiwhen everyone moves there14:41
didrockskyrofa: oh nice! will definitively do:)14:43
elopioping pitti. I need help trying to reproduce the snapcraft autopkgtest. Canonistack has timed out all week before starting to run the tests.14:51
elopiodo you have any tips to reproduce locally?14:51
pittielopio: it works in qemu?14:56
pittielopio: looks like it's trying to contact https://repo.maven.apache.org/maven214:56
pittielopio: possibly it's not respecting $https_proxy for that?14:56
pittitests need to use the proxy in the environment, they can't talk directly to the network14:57
elopiopitti: I haven't even gotten there with qemu: https://paste.ubuntu.com/14863428/14:58
elopioI have everything timing out in different ways :(14:59
pittielopio: wow, how big is that tree?15:02
pittielopio: you can try with --timeout-copy=3000 (default is 300, i. e. 5 mins)15:03
pittibut I guess something else is wrong, unless your tree really is Gigabytes big15:03
* pitti runs "adt-run snapcraft --- qemu /srv/vm/adt-xenial-amd64-cloud.img"15:04
elopiopitti: ummm, we pollute the repository when we build snaps. I'll try cleaning everything up first.15:06
pittielopio: ah, in your ./ ? yes, do that, or run "snapcraft" instead of ".//" to use apt-get source15:07
pittielopio: I started it, and I now see ".......EEE..."15:07
pittielopio: i. e. some tests fail, but it's runing15:07
pittielopio: wrt. trying to download from maven: don't we have tha particular module packaged? we have hundreds of maven packages15:07
pittielopio: i. e. if you add this as a test dependency, then it would avoid having to call out to the internet15:08
elopiopitti: ok! seems promision. /me tries.15:08
elopiopitti: we decided for some reason to use upstream. Not sure about what was that reason now.15:08
elopiosergiusens: ^ ?15:08
pittielopio: ok, then I figure you either don't respect $https_proxy, or something cleans the environment15:09
sergiusenselopio, pitti well one of the purposes for those examples is to use upstream; I I can look into seeing if http_proxy gets wiped or not15:13
elopiothanks pitti! it's running now locally. I can start debugging now.15:13
jdstrandmvo: hey, fyi, both systems are continuing to install hello-world.canonical 1.0.1815:23
jdstrandI didn't check to see if they stopped during the night15:23
jdstrandI can if you need me to15:23
mvojdstrand: meh, thanks15:24
kyrofamvo, sergiusens much simpler reproducible case for the broken $SNAP_USER_DATA here: https://bugs.launchpad.net/snappy/+bug/154189815:40
ubottuLaunchpad bug 1541898 in Snappy "Broken $SNAP_USER_DATA for services" [Undecided,New]15:40
* dholbach hugs sergiusens15:53
kyrofasergiusens, jsonschema question (context: https://github.com/ubuntu-core/snapcraft/blob/master/snapcraft/yaml.py#L343)15:55
kyrofasergiusens, if I load the YAML via yaml.load (which gives me a dict) and hand the dict to jsonschema, how can jsonschema _possibly_ back that out to a line number in the yaml?15:55
noizerkyrofa what is the correct syntax for the copy plugin?15:57
kyrofanoizer, copy has one parameter, "files", which is a map of source->destinations15:58
kyrofanoizer, so something like: http://pastebin.ubuntu.com/14878198/15:58
noizerkyrofa ok but in Private chat you said i did it wrong for my nginx.conf file15:59
kyrofanoizer, right, you have a copy plugin with no files-- just a stage-package15:59
noizeryea ok that was wrong but the part under that it will copy the nginx.conf over the other one but it don't work16:01
kyrofanoizer, you may need to exclude that conf file from the stage-packages16:02
kyrofanoizer, if two parts try to write the same file but they differ, snapcraft will throw a fit16:02
kyrofanoizer, since it doesn't know which one you actually want16:02
kyrofanoizer, look in that syntax file for the `stage` and `snap` keywords for the parts16:04
kyrofanoizer, you can use those to exclude the config file16:05
sergiusenskyrofa, it is impossible16:06
sergiusenskyrofa, but you can get the key that cause the issue16:06
kyrofasergiusens, yeah okay, that's what I thought16:06
kyrofasergiusens, gotcha16:07
renatHi guys!16:32
renatWe have application. Screenly Viewer. Slide Show application which uses opengl to show it's slides.16:33
beunorenat, hi!16:33
renatWhen I start the binary - everything works well, and it works. But when I'm trying to start it as a systemd service. I loads. Shows the first asset and stops. Just using 100% cpu.16:34
renatbeuno, Hi!16:34
beunorenat, lets start off with some basic information16:34
beunowhat release of snappy are you using?16:35
renatIt doesn't show any apparmor and seccomp warnings16:35
beunorenat, ok, lets ping some folks who might be better suited to help debug16:36
renatbeuno, thanks!16:36
beunojdstrand, sergiusens, kyrofa, any pointers? ^16:36
beunokgunn, maybe you have some thoughts?16:36
apwmvo, yo, you know things ... ubuntu-core 15.04 has extended support relative to 15.04 which is just EOL'ing16:36
apwmvo, as i understand things you have an overlay PPA which will take over to supply updates for things after EOL of 15.04 ...16:37
kyrofarenat, are you using mir or something?16:37
apwmvo, so we are trying to confirm that you are needing vivid kernel updates for that, and how those would make it to you16:37
renatkyrofa, hi! No. I'm now 100% sure how it works. But it uses Qtt OpenGL16:38
kgunnrenat: and what exactly are you running on ? amd64? vm? arm?16:38
renatkyrofa, it's Raspberry Pi2 ARM16:38
renatkyrofa, it's Raspberry Pi2 armhf16:38
beunoapw, ricmm might be better suited to answer that16:38
kyrofarenat, are you using the qml plugin in snapcraft, then?16:38
apwricmm, ^^16:38
renatkyrofa, no. We use our patched version of Qt as I know. Build everything into squashfs using buildroot. I create a snap by extracting everything from the squashfs and setting appropriate paths in LD_LIBRARY_PATH and all other libraries.16:40
kyrofarenat, oh16:40
beunoI think 15.04 doesn't support squashfs?16:40
beunodoes it?16:40
kyrofabeuno, right16:41
kgunnbut he says it loads/works....16:41
kgunnonly until he tries to start it as a systemd service does it not work (stalls on first frame)16:41
kgunnrenat: if it shows 100%cpu...what's the process using cpu?16:42
renatkgunn, executable itself16:43
kgunnrenat: mmm, but seems you have a ton of stuff inside that snap...16:44
renatkgunn. I'm placing logging functions to different places. But don't get any output.16:44
renatWhat is difference between running as executable and as a systemd service?16:45
kgunnrenat: that i really don't know...others would be better to speak16:45
kyrofarenat, I'm confused. You say it's Qt and OpenGL. You must be using a display server of some kind?16:47
renatkyrofa, it uses framebuffer16:47
kyrofarenat, ah, okay16:49
renatIt should be something systemd or cgroups related.16:49
kyrofarenat, have you tried running this as an app with sudo?16:50
renatkyrofa, Yes. It works16:51
kyrofarenat, have you tried via su as well?16:51
kyrofarenat, real=effective uid16:51
renatkyrofa, worked again16:54
kyrofarenat, when running as a service, I'm assuming it doesn't fork right?16:55
renatNo. It doesn't I use exec in all wrappers16:55
renatNo. It doesn't.  I use exec in all wrappers16:56
=== kickinz1-sprint is now known as kickinz1|afk
kyrofarenat, by running under su you've eliminated the differences as far as I can tell. I'm really not sure why it's not working when running as a service. Everything else should be the same16:57
kyrofarenat, what you need is gdb :(16:59
renatkyrofa, no. 1 more difference. Different slice. system.slice vs user.slice.16:59
kyrofarenat, hmm, yeah you can either start adding prints to the application to figure out where it's looping, or wait for sergiusens or mvo to give you some more pointers17:01
sergiusensrenat, snappy service logs [snap] gives no output?17:04
jdstrandrenat (otoh, I can't think of something. however, it is possible you aren't seeing security denial due to kernel rate limiting. I suggest: sudo snappy install snappy-debug ; sudo reboot ; login ; sudo snappy-debug.security scanlog ; exercise your snap17:06
renatThanks. trying.17:09
jdstrandbeuno: awe__ asked me to upload bluez to rolling using the shared acount (fine). there is an amd64 and an armhf snap with the same version number. anything special I need to know when doing this?17:10
kyrofajdstrand, they can't have the same version right now17:10
kyrofaUnless that's changed since yesterday17:10
jdstrandawe__: fyi, ^17:10
kyrofajdstrand, so I just append -amd64 and -armhf to the versions17:10
awe__ok; so  build one as -117:11
awe__and the second as -217:11
kyrofajdstrand, other than that, seems to work great!17:11
awe__or that17:11
kyrofaawe__, yeah, whatever doesn't imply that the versions are actually different, you know?17:11
jdstrandkyrofa: thanks!17:11
jdstrandbeuno: nm17:11
sergiusensawe__, kyrofa wait, for 16.04 we have multi arch pockets in the store since monday17:12
sergiusensis that right matiasb ^ ?17:13
sergiusensmaybe for 15.04 two; the store doesn't care about releases17:13
kyrofaawe__, yeah multi arch uploads work, but the versions need to be different (same package)17:13
kyrofasergiusens rather, sorry17:13
sergiusensright version~$arch seems saner than diffrent package names to me17:13
kyrofasergiusens, yeah17:13
kyrofasergiusens, that's what I was trying to say-- sorry I wasn't clear17:14
sergiusens"this is the armhf version of the snap"17:14
ogra_we should really have a meta field for that17:14
sergiusensit even matches how you would mention it17:14
sergiusensogra_, we do ;-)17:14
ogra_why do i then need $version-arch ?17:15
matiasbsergiusens, awe, yes, you can upload multiple versions for different arch/release combination in the same package17:15
sergiusensogra_, because the store doesn't upload every arch atomically17:15
sergiusensogra_, it's like setting an arch in a deb and uploading N times swithing the arch field17:15
ogra_sergiusens, yeah, nothing you would ever do with a deb :)17:15
renatsergiusens, snappy service logs doesn't give more information unfortunately.17:16
ogra_but i guess thats a price we have to pay for binary uploads17:16
renatjdstrand, already tried scanlog. Nothing related to apparmor or seccomp17:16
* awe__ ducks out for a bit; bbl17:16
jdstrandrenat: ok, so if you did the reboot (that is important so the kernel forgets about what it rate limited) and scanlog gave no output, it is something else17:17
jdstrandrenat: the first place I would look is 'snappy service logs <foo>'17:18
jdstrandthat will give all the systemd log output17:18
renatjdstrand, already done. Unfortunately somewhere infinite loop appeared. And it doesn't give additional logging17:18
renatIs it possible to start process inside custom cgroup hierarchy?17:23
kyrofasergiusens, elopio https://github.com/ubuntu-core/snapcraft/pull/296 when you're able17:26
kyrofaHuh. I can't reach github all of a sudden17:53
kyrofaAh, and as soon as I say something, it's back17:54
renatkyrofa, What if I try to move to 16.04 snappy version? Can I just chroot to our squashfs there, and run program from inside the chroot?17:55
kyrofarenat, no not really-- you'd need to run it with ubuntu-core-launcher, with the right environment, etc.17:57
kyrofarenat, though that brings up my `snappy shell` question again. Chipaca, you around?17:57
kyrofaelopio, coveralls doesn't seem to be reporting back to https://github.com/ubuntu-core/snapcraft/pull/29618:15
kyrofaelopio, travis says it submitted it successfully, but coveralls keeps saying "no data"18:16
kyrofaelopio, any ideas?18:16
elopiokyrofa: no, seems just to be dumb. I re-ran it.18:28
kyrofaelopio, ah, okay18:29
kyrofasergiusens, the fix to bug #1541353 is to just parse the snapcraft.yaml, yes?18:29
ubottubug 1541353 in Snapcraft "'snapcraft upload' says 'Invalid package filename.'" [High,Triaged] https://launchpad.net/bugs/154135318:29
kyrofasergiusens, or would you go so far as to unsquashfs and extract metadata from the snap itself?18:30
LefterisJPbeuno: Hey mate, I am uploading my snap to the MWC store and no matter what I do it seems that the icon.png from the snapcraft.yaml is not included.18:42
elopiokyrofa: one more time. Now it's travis with a mismatch hash18:43
LefterisJPbeuno: if I upload the snap to a local snappy device the icon shows up fine in the WebDM interface, so it's probably an error with the store.18:45
beunoLefterisJP, hi18:46
sergiusenskyrofa, the fix is to remove the regex18:47
beunoLefterisJP, is this a 15.04 snap or a 16.04 snap?18:47
kyrofasergiusens, but the upload code needs a name18:47
kyrofasergiusens, it's not just a validation18:47
sergiusenskyrofa, oh, doesn't the upload code contruct a name in the code itself?18:47
kyrofasergiusens, via the regex18:47
sergiusenskyrofa, oooh18:48
sergiusenskyrofa, one second, I'll get you pointers18:48
kyrofasergiusens, so I figure, get the name from the yaml18:48
sergiusenskyrofa, extract these two functions https://github.com/ubuntu-core/snapcraft/blob/master/snapcraft/commands/snap.py#L4718:49
sergiusenskyrofa, and also figure out why it refers to 'package.yaml' and how it works :/18:49
sergiusensput them in common18:49
kyrofasergiusens, uh oh :P18:50
kyrofasergiusens, but yeah, okay, same page18:50
sergiusenskyrofa, expose format snap_name with docstrings and all tat magic18:50
sergiusenskyrofa, I don't understand how it works though; given all tests still pass18:50
kyrofasergiusens, indeed, I'll look into it18:51
sergiusenskyrofa, oh, that first function is not used18:51
kyrofasergiusens, perhaps the tests setup that directory with a package.yaml18:51
sergiusenskyrofa, just export format_snap_name in common18:51
kyrofasergiusens, it's used if called with DIRECTORY18:51
kyrofasergiusens, but yeah18:52
sergiusenskyrofa, ah, we have no test for that!18:52
kyrofasergiusens, oops18:52
LefterisJPbeuno: It's a 15.04 snap that's going in the MWC store18:52
kyrofasergiusens, making bug now18:52
sergiusenskyrofa, units yes, not tests tests18:52
sergiusenskyrofa, even if we did, our prelayed out dir wouldn't of catched this18:52
sergiusenskyrofa, the easy way to have a test for this is to run `snapcraft snap` and then `snapcraft snap snap`18:53
beunoLefterisJP, while I look into that, can you add the icon separately to the app in the store?18:53
kyrofasergiusens, oo, I like it18:53
sergiusenskyrofa, it may have merit for a separate bug18:53
kyrofasergiusens, yeah I think so18:54
LefterisJPbeuno: I think I can but then I won't be able to confirm it works fine when it does :)18:58
beunoLefterisJP, I can do that separately  :)18:58
kyrofasergiusens, yeah, just a bad test setup: https://github.com/ubuntu-core/snapcraft/pull/29719:04
kyrofasergiusens, old, rather-- not bad :)19:04
elopiokyrofa: sergiusens: I'm going to lie in bed for the rest of the day.19:20
kyrofaelopio, feel better!19:20
awe__so earlier this morning, I installed hello-world using: 'sudo snappy install hello-world/edge19:23
awe__when I try now19:23
awe__I get snappy package not found19:23
ogra_blame beuno ?19:23
* ogra_ guesses the store was fixed during the day19:24
awe__kyrofa, is there anyway around a full rebuild of a autotools based snap when you only change a single file?19:27
awe__right now I find myself having to do a clean first19:27
awe__and this takes forever on a ripi219:28
kyrofaawe__, yeah I feel your pain. Unfortunately not, but it's something I'm hoping to solve soon19:28
kyrofaawe__, my ownCloud snap takes literall all day (or night) long to build on the rpi19:28
sergiusenskyrofa, right, which goes down to the fact that only units can't save you ;-)19:33
sergiusenselopio, get better19:33
kyrofasergiusens, indeed. I added an integration test as well19:34
kyrofasergiusens, so hopefully that doesn't bite us again19:34
=== dustino_ is now known as dustino
=== cprov_ is now known as cprov
kgunnjdstrand: ping21:22
kgunnjdstrand: hey, was just updating to skills...and wanted a sane eye21:22
kgunnthat format look right?21:22
kgunnlike the "migration-one/two" can just be random/meaningless to me?21:23
jdstrandkgunn: you want to uses caps and not security-template21:23
jdstrandkgunn: plus you can combine those into one skill21:24
jdstrand  type: migration-skill21:24
jdstrand  caps: [display-server, unix-listener\21:24
kgunni get it21:24
kgunnaltho...interesting, i thot the whole reason was to get rid of the "caps" nomenclature21:25
* kgunn didn't really get the passion of the move to term "skill"21:25
kgunnbut thanks!21:25
jdstrandkgunn: it is, this is a 'migration-skill'21:26
pedroniswell migration-skill is just an interim thing, so it crosses the old and new world (also terminology wise)21:26
kgunni figured21:26
jdstrandwhen the security parts of skills a worked out, that can go away21:26
kgunnjdstrand: just looking ahead, is the goal to be nailed down at 16.04?21:27
jdstrandzyga and I will be talking soon to sync up once I get the review tools changes done for 16.0421:27
kgunnthanks again21:27
jdstrandkgunn: everyone wants that. from my side, I'm good with doing it for 16.04, but there might be things on the skills side I'm not aware of that might prevent it21:28
jdstrandI think it is ok to assume 'yes'21:28
robert_ancellelopio, around?21:46
=== chihchun is now known as chihchun_afk
Chipacakyrofa, i'm around now, what was the question?21:50
sergiusenskyrofa, you should join the fun and get onto the slack channel :-P22:03
sergiusensChipaca, I think I know22:04
sergiusensChipaca, systemd service units are not replacing %h properly22:04
sergiusensChipaca, convert hello-world.env into a daemon and you will see22:05
sergiusensChipaca, %h should be /root, instead it's plain `/`22:05
Chipacasergiusens, can you pastebin the output of hello-world.env-as-a-daemon?22:05
sergiusensChipaca, ok, first I need to unxz the latest image; give me 2'22:06
Chipacasorry i'm being lazy22:07
Chipacarough day22:07
Chipacathinking of taking tomorrow off or somehting22:07
zygajdstrand: with pleasure22:10
sergiusensChipaca, look at SNAP_USER_DATA http://paste.ubuntu.com/14883057/22:14
sergiusensChipaca, notice the double `/` there22:15
* sergiusens heads out for aikido22:15
Chipacasergiusens, yes yes. But is that *all* the output from env? (I want to see the environ :-) )22:21
Chipacasergiusens, my suspicion is that for some reason HOME isn't set22:21
sergiusensChipaca, it is22:23
sergiusensChipaca, don't units set HOME accordingly or is that some sort of system setting?22:23
sergiusensthe manpage says %h resolves to /root by default22:23
sergiusensI need to go22:23
sergiusenswill bbl22:23
sergiusensif you want to drop lines here I'll read later :-)22:24
Chipacasergiusens, go go22:24
plarselopio: I've cherry-picked your older patches into current trunk, and I'm trying to also cross-build the test binary for arm but I'm getting an error. Have you seen anything like this?22:26
plars /tmp/gotestbuild/src/github.com/ubuntu-core/snappy/progress/progress.go:200: undefined: isatty22:26
grim__does anyone know if there are snappy images archived anywhere?  The core updates cause some of our snaps to fail to install.  We'd like to try to install on core versions 1-current to figure out what is going on.23:20
neutrinoAny way to create self hosted snappy store for private snapps?23:51

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!