=== laza_ is now known as laza === xnox_ is now known as xnox [06:45] good morning [08:10] good morning [09:49] Good morning all! Happy Thursday, and happy Get Out Your Guitar Day! 😃 [10:45] hi, I'm having trouble with my mirror, can I redirect snapcraft 2.2 (when pulling) to use the main archive instead? [10:46] whats the new SNAP_APP_DATA_PATH env? [10:46] hm. guess i would need to upload hello world again [10:47] didrocks: you made a central place for examples in github right? [10:52] okay, the mirror is up again now [11:01] Hi guys is nginx supported on snappy? [11:01] Chipaca, tickle [11:02] noizer, if you manage to package it, it is supported :) [11:02] nicee :D just i can't get it started xD [11:03] ogra_ where can we find the image for the dragonboard for snappy? [11:03] are you not on the mailing list ? [11:04] oooh ok sorry i didn't have time to look at it the last couple of days [11:04] http://people.canonical.com/~ogra/snappy/all-snaps/dragonboard/ [11:04] i mailed yesterday ;) [11:04] (snappy-devel that is) [11:05] as for ngnix ... i think the spreedbox guys use it in their spreedbox snap in the store [11:05] no idea where their code is (and this is for 15.04 only iirc) [11:05] you could install their snap on a 15.04 armhf install and inspect the installed snap though [11:06] ok thx ogra_ === pindonga` is now known as pindonga [11:40] ogra_, tockle [11:46] asac, he has, but it seems bare bones now https://github.com/ubuntu-core/demos [11:50] Chipaca, damn, i forgot what i wanted to ask you now :( [11:51] ogra_, \o/ [11:51] Chipaca, oooh ... ubuntu-snappy ftbfs on arm64 since a while again, i need to re-build images after fixing the resizing for dragonboard setups [11:52] would be good if someone could take a look (and since mvo is away ...) [11:52] ogra_, ouch [11:53] i'll take a look in a bit [11:53] thanks ... take your time though ... i'm still working on the resize fixes [11:57] sergiusens: ok thats demos then... guess moving snappy-examples to 16.04 in its current place is worthwhile then still... [12:07] asac: yeah, it needs sorting, and it quite empty, but it should be a central place [12:07] the difference I see is that demos will be more complete real pieces [12:07] and snapcraft should keep examples separated [12:07] (as more "simple way of doing X") [12:11] noizer: try https://github.com/zyga/devtools/blob/master/ubuntu-image and tell me if it works for you [12:12] zyga what is that? [12:13] is that for the dragon board or is that for my nginx [12:13] ogra_: hi, i'm trying to boot the new all-snap dragonboard image, but i'm not seeing any output on hdmi. do you have any ideas on why not? [12:14] ogra_: same setup works ok with the install image from 96board website [12:14] joc__, because the boot console still points to serial ... once it booted you should get a login prompt on hdmi though [12:14] (i definitely do here) [12:14] joc__, also regard the big fat warning in the readme [12:14] (to not use a card bigger than 4GB) [12:16] ogra_: will it resize and reboot before the hdmi login prompt? [12:17] it resizes before it mounts anything [12:17] (so thats a yes) [12:18] i'm hoping to have that bit fixed today :) [12:18] ok thank you, plars was saying that he had one successful boot before the resize would stop his board working so I was hoping for that - he may have been using serial though [12:18] kyrofa, when you get on, mind looking at https://launchpad.net/snapcraft/+milestone/2.2.1 ? [12:20] joc__, i havent tried booting without the serial board attached, perhaps that blocks somewhere else though ... [12:20] * ogra_ gives that a try [12:24] noizer: that's for building images easily [12:24] noizer: including dragonborad [12:26] zyga can I make then custom images for myself? [12:27] zyga ps I will first make my nginx up and running [12:28] Who will also join the clinic tommorow? [12:28] noizer: depends on what you mean by custom [12:28] noizer: anyway, have a look, the tool is _tiny_ [12:29] zyga ok [12:29] joc__, hmm, yeah, looks like it hangs somewhere at the bootloader if the serial board isnt attached .... [12:30] Good morning [12:30] joc__, so for now, please attach the serial board, i'll research that [12:31] Good morning kyrofa [12:32] sergiusens, what should I be looking at here? [12:39] good morning kyrofa! [12:40] kyrofa, the PRs :-) [12:40] Hey noizer, didrocks :) [12:40] * sergiusens needs to go to the bank, will bbl [12:40] sergiusens, ah, that's what I'm on now [12:40] sergiusens, alright, seeya! [12:41] ogra_ What is the name of the application of spreedbox? [12:41] noizer, spreedbox :) [12:41] (iirc ) [12:41] kyrofa Hi do you know where I can get more information about netwerk listeners etc [12:42] ppisati, did you ever try to boot without serial board attached ? seems it hangs at the bootloader [12:42] noizer, you mean what permissions are encompassed by it? [12:42] yes [12:42] ppisati, dragonboard that is [12:42] ogra_: uhm no, i never trie [12:42] d [12:42] let me tru [12:42] let me try [12:42] kyrofa its for the rolling version. Or will it be explained tomorrow on the clinic? [12:42] seems to not be kernel related ... i just did set console=tty0 and it still hangs [12:44] ogra_ I installed the 15.04 version on my Raspberry Pi 2 and tried to install spreedbox with the following command: sudo snappy install spreedbox but he dont find it [12:45] ppisati, wow .... trying to set "stdin, stdout or stderr" at the uboot prompt with setenv actually complains [12:46] i cant set it to serial,lcd [12:46] noizer, I'm not aware of any documentation on each one. jdstrand might know more [12:48] ogra_: ## Error inserting "stdin" variable, errno=22 [12:48] :O [12:48] yeah [12:49] i know i can set that var on the rpi [12:49] i did that multiple times [12:49] ogra_: i was using https://github.com/hallor/u-boot.git / dragonboard-for-mainline-v2 [12:50] ogra_: i just did a git fetch, and there's a 'dragonboard-for-mainline-v3' [12:50] let's see [12:51] ah [12:52] though i fear there is some build time hardcoding going on here [12:56] ogra_: same story, doesn't boot and you can't set stdin&c [12:57] yeah, i suspected that [12:57] i just pinged in #96boards ... lets see, perhaps a known issue [12:57] ogra_, i can't find the build failure email, could you fwd it me? [12:58] i have to dig it up [12:58] Cghbtw, vivid failed too (on arm64, powerpc and ppc64el) [12:59] https://launchpad.net/~snappy-dev/+archive/ubuntu/image/+packages [12:59] arm64 xenial is here https://launchpad.net/~snappy-dev/+archive/ubuntu/image/+build/8962323 [13:01] => setenv stdout serial,lcd [13:01] ## Error inserting "stdout" variable, errno=22 [13:01] => setenv stdout serial [13:01] => [13:01] ppisati, ^^^ [13:01] so its the parameter [13:01] just setting it to serial works [13:01] oh ok [13:01] kyrofa did you already installed an apache in a snap? [13:02] i guess we miss a build time option somewhere in uboot to enable lcd at all [13:02] noizer, indeed, yes [13:02] noizer, that owncloud snap uses apache [13:02] though i still dont get why it hangs hard ... it shoudlnt [13:03] general question: if the app has some files in /usr/share/ - but with snappy, they're instead in $SNAP/share or $SNAP/usr/share, what's the smartest way to have the app (or one of its library dependencies) look there instead of in /usr/share ? [13:03] kyrofa owkay i will look for the yaml because I can't run nginx :s [13:03] noizer, how come? [13:03] noizer, I can share my plugin with you if you like [13:04] yea i would like that maybe i can get some info for my nginx from there [13:05] noizer, alright. It's still a work-in-progress, no decent documentation or anything, but it works: https://github.com/kyrofa/owncloud-snap/blob/apache_plugin/parts/plugins/x_apache.py === nessita_ is now known as nessita [13:05] ppisati, ah, seems we just need to set CONFIG_LCD in dragonboard.h [13:06] kyrofa You didn't installed it with stage-packages? why? [13:06] noizer, you give it a source (like any other plugin) and it'll unpack it into htdocs. You also specify the modules you want built/enabled, and you can specify some sections of the config and a startup script if you want [13:07] noizer, because debian packaging uses a different apache layout than the standard, and it's all setup in a post-inst [13:07] postinst, rather [13:07] kyrofa ok. [13:07] noizer, since Snapcraft only unpacks debs, that stuff never gets run [13:08] noizer, so it led to a lot of hackery. Not to mention this way I could tune it exactly the way I wanted and get it down to a few megs [13:08] so nginx won't run with the deb package? [13:08] noizer, it might, it might not. Depends on how it's packaged [13:08] ogra_: let m,e try that [13:09] noizer, that was the first lesson I learned making that massive snap: debian packages only get you so far in snappy [13:09] ok so maybe the best way is from source code. [13:09] noizer, indeed [13:11] kyrofa, ok thats a lot harder then using a deb package :p [13:11] kyrofa lets give it a try [13:11] noizer, yeah it's a bit complex. Apache was particularly painful because it writes its prefix _everywhere_, config files, scripts, etc [13:12] hopefully is nginx not so painful :p [13:12] ogra_: nope, it doesn't compile with only CONFIG_LCD [13:12] noizer, heh, good luck [13:12] kyrofa thx. if I succeed maybe then i will post an easy example of it for other users of Snappy [13:13] ppisati, how does it fail ? [13:13] noizer, the way snapcraft works requires software that can be configured and built in one location, and run in a complete different location, with different paths. Some software works fine that way. Some though, like apache, do not [13:13] ogra_: http://pastebin.ubuntu.com/15015948/ [13:14] ogra_: CONFIG_LCD is just the machine independent code, every hw needs some specific bits to make it work [13:14] crap [13:14] ogra_: we should talk to guy who ported uboot to this board [13:15] ppisati, thats hallor in #96boards i think [13:18] kyrofa is that not a good idea to use docker in the snap. Because docker have already a build with nginx? [13:19] noizer, hmm... I'm not sure I understand [13:19] Ok you know docker? So when I'm installing docker and on the docker side you can install nginx easily (normally) [13:20] noizer, alright, I'm with you so far [13:21] So when im installing nginx on docker. into my snap then can I maybe refer to my data in my nginx.conf? [13:22] noizer, that's where you lose me. How does docker relate to your snap? [13:22] hmm i don't know is it possible to use docker and nginx into my snap? [13:22] kyrofa i didn't use docker before so I'm really noob at docker [13:24] noizer, okay so you're thinking "It's easy to put nginx on docker. It's hard to put nginx into the snap. So can I get docker into my snap instead and put nginx on there?" [13:24] noizer, sound about right? [13:24] kyrofa yes :p [13:24] noizer, I know a docker framework exists on 15.04, though I'm not sure what the plans are for 16.04 [13:25] oooh ok xD i will do it the hard way and I think the best way then :D [13:25] noizer, you'd have to depend on the docker framework and make your .snap that way. I've not done that, though there are people here that have [13:26] noizer, yeah, there's non-negligible overhead associated with things like docker. You don't want it if you don't need it [13:27] kyrofa ok. I will you update on my status of nginx. If you want that :D [13:27] noizer, sure :) . It may not be as bad as you think! [13:28] kyrofa ok hopefully xD [13:30] noizer, I strongly suggest you experiment with it outside of snapcraft until you understand the build process [13:32] kyrofa [13:32] ok i will do that first [13:33] ppisati, try bui9lding with CONFIG_USB_KEYBOARD (and perhaps with CONFIG_SYS_USB_EVENT_POLL_VIA_CONTROL_EP) [13:36] ogra_: compiling... [13:40] dragonboard410c => setenv stdin serial,usbkbd [13:40] ## Error inserting "stdin" variable, errno=22 [13:40] ogra_: ^ [13:40] bah [13:40] with these: [13:40] ok, i gues si have to disable bootdelay then [13:40] +#define CONFIG_USB_KEYBOARD [13:40] +#define CONFIG_SYS_STDIO_DEREGISTER [13:40] ogra_: you mentioned something else before [13:40] ogra_: setenv stdin ... [13:41] i only know serial.usbkbd [13:41] err ... comma indeed [13:41] ogra_: oh ok [13:42] * ogra_ tries the evil way [13:42] ubuntu@localhost:~$ sudo fw_setenv stdin serial,usbkbd [13:42] ubuntu@localhost:~$ [13:42] muhahaha ! [13:42] oh [13:42] did it work? [13:42] ubuntu@localhost:~$ fw_printenv |grep stdin [13:42] stdin=serial,usbkbd [13:42] ubuntu@localhost:~$ [13:42] it sets it [13:43] weather it will boot is another question :) [13:43] reading uboot.env [13:43] In: serial [13:43] Out: serial [13:43] nope [13:43] ignores it [13:43] doh :( [13:44] Environment size: 4424/131067 bytes [13:44] => printenv stdin [13:44] stdin=serial [13:44] wow [13:44] it even unsets it during boot [13:44] (it accepted my bootdelay change) [13:45] well, people wanting to debug the bootloader can always set bootdelay from the running system, so i guess i'll default to bootdelay=0 [13:46] ogra_: k, i sent an email to the uboot guy, just in case (you are in cc:) [13:46] thanks [13:47] he already said there might be a bug in hserial [13:48] joc__, thanks a lot for bringing that up !!! [13:48] (no dev would ever have catched it since we all boot with serial board attached ) [13:55] ogra_: np, thanks for the info [14:04] noizer: re documentation, install snappy-debug then do 'snappy-debug list -i' [14:04] noizer: the 16.04 names will be changing based on skills work though [14:05] hopefully those will be settled within the next couple weeks [14:13] noizer, questions over here please :) [14:15] jdstrand ok [14:16] ogra_, https://launchpad.net/~snappy-dev/+archive/ubuntu/image/+build/8962323 [14:19] ogra_, the good news is, I did nothing more than rebuild [14:19] ogra_, but that is also the bad news [14:19] ogra_, there are tests in there i'm not proud of :-( [14:19] that depend on the test starting and ending the same minute [14:46] Chipaca, well, at least we have a binary now to make the image build, thanks a lot ! [14:46] Hi I have a question about Ubuntu as an OS [14:58] sergiusens, FYI, as soon as netbase is in an all-snaps image I'll make MOOS an example. Until then, if you're curious: https://rainveiltech.com/posts/that-s-one-snappy-moos [15:00] kyrofa, nice === vrruiz_ is now known as rvr [15:00] kyrofa, oh, thanks for the reminder :P [15:00] * ogra_ hacks the seeds [15:00] ogra_, ha! Thanks again on that :) [15:01] ogra_, can you upload new OS images? or is it only mvo? or does slangasek have the priv as well? [15:01] * sergiusens heads back home [15:01] sergiusens, walk safely. Look both ways [15:02] sergiusens, i dont know what account mvo used for them ... i'd rather have hiom do it [15:03] fgimenez: why do we have offline slaves? Did I break something? [15:03] kyrofa, seeded and pushed, next image should have it [15:04] * kyrofa hugs ogra_ [15:04] elopio, i have no idea, was waiting to ask you :) [15:04] I don't know. Didn't notice yesterday. [15:05] i'll remove them [15:08] Hey didrocks I wanted to ping you regarding the owncloud snap. Is that in a good state for demo? We still have some time to fix things if necessary [15:08] kyrofa, an arm64 buiold would be so awesome :) [15:08] kyrofa where can i see the deamons running in a snap? [15:08] ogra_, ha! I'm working on it, the dragon keeps dying [15:09] noizer, snappy service status [15:09] kyrofa, even with the new image ? [15:09] ogra_, ah, I haven't tried yet. I'll get back to you [15:09] great [15:11] kyrofa: I guess it's good enough for the demo, I played with it and didn't spot anything bad :) [15:12] thanks for pushing this btw! [15:12] didrocks, you GUESS? Psshh [15:12] didrocks, and no problem, it's nice to get it some exercise [15:13] kyrofa: I still think (but maybe once 16.04 is out) we need to centralize all those "demos" in the git repo [15:13] but I tend to wait for 16.04 for now (at least, that it starts to stabilize with snapcraft 2) [15:13] so I hope for end of March/early April :) [15:13] didrocks, yeah it's too much of a moving target right now [15:14] seen that :p [15:14] ogra_, nooo, I don't have a 4GB SD card [15:14] didrocks, heh, yeah I bet you have :P [15:14] kyrofa, you can resize the writable partition before first boot [15:14] just use gparted [15:15] ogra_ can I see why it is inactive my service? [15:15] or gnome-disks [15:15] noizer, snappy service logs ? [15:15] ogra_, your README leads me to believe there will be nothing to resize? [15:15] or something along these lines [15:15] noizer, the only reason it would be down is if it crashed. Check the syslog for reasons (or your application's log, if any) [15:16] noizer, or yeah, snappy service logs [15:16] kyrofa, dd to a big SD ... then re-plug and open gnome-disks ... pick the writable partition and resize it to occupy all free space ... then the resite on boot wont kick in [15:16] I keep forgetting about that command [15:16] *resize [15:16] ogra_, ahhh, okay [15:17] it checks for more than 10% free space on the SD ... if thats not there it wont run [15:17] ogra_, kinda neat that it's there at all, honestly [15:17] yeah, i just wish i had picked the right partition tool from the start [15:18] (who would have thuoght we ever have u-boot devices with GPT partitions !) [15:18] ogra_, it gets ingrained and it's difficult to switch, eh? [15:18] ogra_ i got no error in my syslog or in my log of the service [15:18] well, it needs a bunch of code changes ... it is just annoying to test since you need to dd all images for all devices to SD multiple times [15:19] enervingly time consuming [15:19] ogra_, ah, sure [15:30] * zyga reflashed his BBBs [15:35] fgimenez: hangout. [15:36] elopio, yep omw [15:56] ogra_, do I need to do anything fancy with these switches on the dragon to boot? [15:57] you need to turn the SD one on [15:57] and leave all others off [16:03] could it be taht running udf as root (not sudo)... is not working at all? [16:04] i get weird error that it cannot create /root/.cache etc. [16:07] noizer, by the way, if you're curious about the actual apparmor snippets relating to each capability, check out /usr/share/apparmor/easyprof/policygroups/ubuntu-core/16.04/ [16:09] okey i see [16:09] ps i tried to run the standard nginx ( installed with stage-packages) and it do not work at all :s [16:09] noizer, /usr/share/seccomp/policygroups/ubuntu-core/16.04 for seccomp filters [16:10] asac, i never ran it as root, might interfere on some levels [16:10] noizer, yeah, not surprising [16:10] kyrofa i can see it start but not that it crashes in the logs :s [16:11] ogra_, gnome-disk is just showing be a bunch of unknown partitions. I'm in trusty... too old? [16:11] s/showing be/showing me/ [16:11] kyrofa so i need to compile it myself [16:11] kyrofa, you should see at least two usable ones (the last two) [16:11] but so I can see it says make serve [16:11] to build it [16:11] ogra_, nope... everything is unknown [16:12] kyrofa, the rest will most likely still be unknown today, they are part of the bootloader and use very specific IDs [16:12] thats weird [16:12] noizer, maybe. You might be able to get away with some sed magic [16:12] the last two are just vfat and ext4 [16:12] with no special signature [16:13] gparted sees that, but then gets super confused when I try to resize the ext4 saying the partition doesn't exist [16:13] hmpf [16:13] gnome-disk doesn't actually show the last two at all, now that I look closer [16:13] ogra_, just 1-7 [16:13] kyrofa sed magic? [16:13] well, you can always use gdisk on the cmdline [16:14] ogra_, ah, I didn't actually know about gdisk. Tried using fdisk to obvious dispair [16:14] yeah, the old tools all kind of lack if it comes to GPT [16:15] noizer, if for example it's dying because it's trying to save logs in a non-writable place, you can try copying a new config file over the top that uses the right paths [16:15] hmmm. running bluetoothctl hangs my pi2 ... or at least my ssh session goes down [16:16] woot, all boards operational :) [16:17] ok i will do that first [16:20] asac: maybe get beefier psu [16:20] asac: I'm using a big 5-port 10A supply [16:20] asac: works great for a small farm [16:21] hmm. you say this could be power? [16:21] it happens even without a dongle attached [16:21] hmmmmm [16:22] asac, anything in syslog or dmesg ? [16:23] ogra_: not really [16:23] let me see after this reboot [16:23] the last two times nothing was written [16:23] even though i had entries from yesterday in there [16:23] so something got persistet, but not these boots it seems [16:23] also try to use serial to make sure its only ssh/network that goes down [16:24] yeah it seems to really not persist the syslog [16:24] culd be your USB hub dying ... (since everything on the rpi is USB) [16:24] if it dowenst shut down cleanly [16:24] guess thats a bit of a bad feature :) [16:24] ogra_: could be, but i dont attach anything and just run bluetoothctl [16:24] and it dies [16:24] my pi2 is in an orange box [16:24] cant connect serial [16:25] if you have a pi2 with 16.04 just install bluez5 [16:25] asac, you have a USB NIC ... [16:25] anywsay, i connectd it now to strong USB power supply [16:25] let me try again [16:25] nope same... with 4A [16:25] yeh [16:25] just dead the second i run bluetoothctl [16:25] without attaching any dongle and nothing [16:25] k... llet me try to find my other pi2 where i can make a serial avail [16:26] i think i had another one [16:26] ogra_: which pins do i connect the ttl to? [16:27] ugh ... i dont have my rpi nearby [16:28] asac, http://workshop.raspberrypiaustralia.com/assets/console-cable-connections.jpg [16:28] (you dont need VCC usually) [16:29] yay, i like raw boards with serial console :OP [16:29] works [16:29] lets see what happens there [16:29] they look so nice on the desk [16:29] yeah, my desk is plastered with them :) [16:30] right, but its your life style to have raw boards on your desk ... i am trying to get to a clean desk (not succeeded yet) [16:30] hehe [16:30] my desk is clean ! [16:30] (everything around it is a mess admittedly) [16:31] ok first thing i notice is: http://paste.ubuntu.com/15017149/ [16:31] lol [16:31] so thats probably not the best indicuation of successful bluez5 usage [16:31] ouch [16:31] yeah [16:31] yeah [16:31] dead [16:31] morphis, ^^ [16:31] console dead when running bluetoothctl [16:31] no oops [16:31] nothing [16:31] end of life [16:31] bang [16:31] guess this is bogus stuff [16:31] thts with a USB dongle ? [16:31] no [16:32] nothign attached but serial, power through USB and LAN cable [16:32] is tere any BT onboard on the rpi ? [16:32] i dont think so [16:32] right, so why would it start then :) [16:32] kyrofa, elopio here's a weird one https://github.com/ubuntu-core/snapcraft/pull/316 [16:32] ogra_: what would start? bluez5? [16:32] yeah [16:32] i think its a deamon that should just there [16:32] until something gets plugged in [16:32] if there isnt a BT device there is nothing to attch to [16:32] also how can bluetoothctl binary just crash everything [16:33] sure, but its a service [16:33] no? [16:33] ask the kernel :) [16:33] the kernel doesnt even spit out anything before it dies [16:33] ppisati: ^^ [16:33] 16.04 pi2 ... install bluez5 and run bluetoothctl [16:33] it will be dead without oops [16:33] do you get the obex messages too if you have a BT dongle pluggged in ? [16:33] ok let me reboot and plug it in [16:34] well, powercycle ... rather than reboot :) [16:34] yeah [16:34] asac: but is the board dead? led blinking? can you still ping it? ssh-in? [16:35] ssh dies [16:35] noise][: I think I hit the same issue as you were discussing yesterday, but I didn't understand what it was until now. I specified the release for my product as "rolling" instead of "rolling-core". Is there an easy way for me to edit the product? [16:35] ppisati: ssh is dead ... thats how i noticed and connected serial [16:35] serial also dead [16:35] i will check LEDs next time [16:35] give me 1 sec [16:35] yeah, we have heartbeat on by default [16:35] leds should tell you something [16:35] obey is not starting [16:35] even with dongle [16:36] bluez still fails even with dongle plugged while booting [16:36] sounds broken :) [16:36] do you see the right driver being loaded on plug/unplug ? [16:36] (in dmesg) [16:36] ok so right now the led is red [16:36] i can still use console [16:36] i am typuing bluetoothctl again [16:36] (nothing is blinking though even though it works) [16:37] the LED is still red [16:37] hmm, i thought the rpi kernel had heartbeat on by default [16:37] no blinks [16:37] me too [16:37] thast 16.04 ? [16:37] yeah [16:37] 4.4 kernel? [16:37] ok let me reboot from scratch and try the pinging of iface [16:37] can you ckeck the kernel version with uname ? [16:37] can also do that [16:37] [ 0.000000] Linux version 4.3.0-1006-raspi2 (buildd@kishi13) (gcc version 5.3.1 20151206 (Ubuntu/Linaro 5.3.1-2ubuntu2) ) #6-Ubuntu SMP Tue D) [16:38] ppisati: ogra_:^ [16:38] seems disk io makes the LED blink [16:38] its blinking right now [16:38] while booting [16:38] asac: so i just start bluetoothctl and the board dies, correct? [16:38] yeah [16:38] just do that [16:38] dead [16:38] no dongle needed [16:38] ok, let me try [16:38] end of day :P [16:38] lol [16:38] end of life :) [16:39] noise][: otherwise I can just delete it, just checking first if there's a better way [16:39] yeah, wanted to write that, but then i can still powercycle, so i think its more like a day... :) [16:39] but guess every boot is a life [16:39] then its fine [16:40] guess i could also have fun and bring another board to my desk ... my good old bbb [16:40] really depends on the game though [16:43] asac: so, i just tested on a 4.2 raspi system [16:43] asac: if i start blueoothctl, it takes over the tty and i loose it [16:43] asac: e.g. if i'm in a ssh connection, it freeze [16:43] asac: but the board is ok [16:43] asac: serial works, and the heartbeat led keeps beating [16:44] asac: indeed if from serial i kill the bluetoothctl process, i'm back in business [16:44] oh... so it might also take the tty on our serial for snappy? [16:44] let me ping to test [16:44] asac: actually even the ssh is still up [16:44] asac: this on a 4.2 kernel [16:45] i shall try with a 4.3 [16:45] just in case [16:45] asac: i guess so [16:45] oh, it might a bug tough [16:45] i'm no ruling out that [16:45] just that the board is alive, use ssh and then try with the serial [16:45] ping still works indeed [16:45] but my serial is down too [16:45] ssh in, bluetoothctl over ssh and it freeze [16:46] than try the serial [16:46] system is ok [16:46] *then [16:46] asac, try changing the kernel commandline [16:46] and drop the console=tty0 [16:46] (with fw_printenv and fw_setenv) [16:46] * ppisati tries with a 4.3 kernel, you never know [16:47] snappy has two console= options on the cmdline [16:47] ogra_: from userspace? fw_printenv is not there in uboot [16:47] asac, indeed from userspace [16:47] in uboot just use setenv and saveenv [16:48] yeah only that uboot dies when running printenv alone :( ... doesnt return to prompt after its finished [16:48] let me boot this thing to linux then [16:49] huh ? [16:49] definitely works here [16:49] ogra_: isnt that somewhere in a file in /boot? [16:49] that i can edit rather than running fw_ ? [16:49] no [16:49] it is a file in system-boot but it is binary [16:50] ogra_: so i run fw_setenv 'mmcargs=setenv bootargs "${args} console=ttyAMA0 root=${mmcroot}"' [16:50] ? [16:51] that wiped the whole mmcargs line :( [16:51] yeah, wrong quoting [16:51] * asac hopes he can fix it before reboot [16:52] fw_setenv mmcargs 'setenv bootargs "${args} console=ttyAMA0 root=${mmcroot}"' [16:52] ok [16:52] and check with fw_printenv [16:53] ok that looks better [16:53] * asac crosses fingers and reboots [16:54] ogra_: didnt help [16:54] seems to highjack all ttys [16:55] that i have [16:55] what a beast [16:55] well, was worth a try [16:56] so what now? guess coffee and break [16:57] yeh, and a cigraette [16:57] +r [16:57] ogra_, gdisk won't let me create a partition of type FFFF... know any workarounds? [16:58] kyrofa, the type deosnt matter [16:58] ogra_, oh, okay [16:58] not rfor the last two [16:58] mmm coffee [16:58] just make sure to not touch the first ones [16:58] fgimenez: is it safe to restart a slave? [16:59] ah nevermind, It's not needed :D [16:59] elopio, sure, docker restart will give you a pristine one [16:59] elopio, ok :) [17:00] we are missing update-ca-certificates -f in the slaves [17:00] uncertified slavery ! [17:01] all our papers are in order. This is legal. [17:01] hah [17:01] fgimenez: should I put it in the swarm-slave? [17:02] * ogra_ sends khaleesi and the dragons to check that [17:03] ogra_: this totally badass scary guy will be waiting for you: https://wiki.jenkins-ci.org/download/attachments/2916393/logo.png?version=1&modificationDate=1302753947000 [17:03] elopio, mm i thought this was done in an ancestor of that images, let me check [17:03] OMG ! [17:03] he will treat us with extensive friendlyness ! [17:04] elopio, ogra_ careful sometimes he gets angry https://raw.githubusercontent.com/jenkinsci/jenkins/master/war/src/main/webapp/images/rage.png [17:04] lol, that should be our default logo. [17:04] hah, i picked up a sticker of that at SCaLE :) [17:05] if you manage to get a java exception on the server you'll see him like that :) [17:06] challenge accepted! [17:08] elopio, it's done only on the master https://github.com/ubuntu-core/jenkins-ubuntu/blob/master/Dockerfile#L5 [17:09] 2016-02-04T21:26:39.484191Z ubuntu-core-launcher cp: cannot create regular file [17:09] ��‘/etc/dbus-1/system.d/bluez5_bluez_5.37-1-armhf.conf��’: Permission denied [17:09] think thats not good [17:09] yeah, your UTF-8 is broken :P === Taco is now known as Guest91403 [17:10] well, the fac that it cannot put that stuff there [17:10] is ther eason why bluez is not activated [17:10] # Allow replacing our dbus policy configuration file until [17:10] # snappy has a better way to do this. [17:10] /etc/dbus-1/system.d/bluez_* rw, [17:10] is in apprmor profile [17:10] yes,, it has a hacked up sercurity template iirc [17:10] fgimenez: I know. What I'm not sure is if I should add it to the two slaves, or to the swarm. [17:10] ogra_: right, which seems to not work well [17:11] Hey I just installed snappy on my RPi and then instakk thexkcd-webserver. How do I start xkcd? [17:11] asac, probably because snappy moved a lot since the package was uploaded [17:11] asac, especially the security/skills stuff [17:11] Guest91403, it starts automatically [17:11] jdstrand: did we drop support for the old way to do profiles now? [17:12] * ogra_ forgot on which port ... might be 8080 [17:12] Thats what I read online so how do I view it then? [17:12] i am on ubuntu-core 2016-02-03 16.04.0-7.armhf canonical [17:12] Guest91403, point a browser at your rpi [17:12] okay [17:13] http://webdm.local:8080/ [17:13] that might work [17:13] ok i manually copied the dbus conf there now [17:13] ets see [17:14] Hi all, I have a snap for both the amd64 and armhf architectures that needs a Java JRE. It looks like there's a "jdk" part but it's unclear what plugin it should use. [17:15] whizzo: java plugin or maven plugin should do the trick in snapcraft [17:15] ogra_, no that's webdm [17:15] Guest91403, xkcd is just on port 80 [17:15] kyrofa, on 8080 ? [17:15] ogra_, yeah [17:15] whizzo: err jdk plugin [17:15] since when [17:15] or maven or ant [17:16] ogra_, no [17:16] it used to be 4200 [17:16] ogra_, 4200 or something, sorry [17:16] yeah :) [17:16] ogra_, 8080 isn't anything unless you're using kvm [17:16] 4200 isa webdm [17:16] you can go to the app there and there is a link to its real port [17:16] ogra_, and forwarded 80 to 8080 [17:16] kyrofa, right, i wasnt sure if the xkcd demo runs on 80 or 8080 [17:16] asac, when I try part "jdk" with plugin "jdk" I get an error that 'source' is a required property [17:16] ogra_, yeah, just 80 [17:16] k [17:16] ogra_, last time I tried anyway [17:17] whizzo: yeah you need to do something there [17:17] asac, does it matter what? [17:17] source: . [17:17] that might work [17:18] whizzo: try to just specify . [17:18] its a bug [17:18] (it works with the python plugin ... ) [17:18] not sure if jdk plugin is suposed to be use standalone [17:18] if you use ant or mavent as build system just go for thse directly [17:19] ogra_: so after putting the dbus file in blace, bluez and obex daemons are happy [17:19] but bluetoothctl still highjacks everything [17:19] heh [17:19] asac, it seems to be building with that change. [17:19] man what nasty stuyff does this thing do [17:19] yeah [17:19] whizzo: cool. lets hope it also works :) [17:19] asac, thanks for the help... i'll see if it actually works properly [17:19] whizzo: what build system are you using? [17:20] * asac wonders if there is anything beyond ant and maven used in java world [17:20] asac: I'm using maven but my maven build is above snapcraft -- it builds for several platforms. [17:20] ic so you need multi arch/cross feature [17:21] whizzo: so you have JNI bindings? [17:21] asac: the maven build lays down the directory that I run snapcraft against [17:21] i think you might run better with doing one snap for amd64 and another for armhf for now [17:21] asac: did something change? I used to be able to build a snap that included both archs... [17:21] yeah i get it. think best if maven could just be used tip to tail, but then, give it a try. the jdk plugin surely will get you the jre bits [17:22] whizzo: yeah you can still do that, but snapcraft doesnt support doign that [17:22] in the sense of building two runs with different toolchains [17:22] i think what you try to do could work [17:22] just need to add the magic wrappers etc. in place [17:22] but you cant get multiple archs for jre [17:22] yeah, snapcrft definitely supports just copying binaries around [17:22] in one build [17:22] * ogra_ does that all the time [17:22] asac: ok, i'll give the snap that just built a try [17:22] so you might want to use the multi-arch syntax and see if that explodes [17:22] :) [17:23] right [17:23] hmm. wonder why we build bluez5 without-systemd [17:24] or rather what would happen if its build that way [17:24] ask tony or morphis [17:24] * asac kind of suspects it might be interfering with auto tty by systemd [17:24] but i guess it would need even more apparmor hackery [17:24] not even sure what systemd would bring in ... shrug :) [17:24] * asac will try to look at code [17:24] but has not high hopes [17:25] hmmmmmm [17:26] let me try to do two ssh logins [17:26] dont overload the poor board ! [17:26] elopio: zyga: either of you (or anyone else) ever see this error when building an x86 image with mvo's udf? [17:26] snap unpack failed with: exec: "unsquashfs": executable file not found in $PATH () [17:26] :) [17:26] so yeah its just that its bogus [17:26] readline [17:26] doesnt return [17:26] oh [17:26] *sigh* I realized what's up with it as soon as I pasted it [17:26] :) [17:26] heh [17:26] :) [17:27] ok got it... dbus permissions are not granted, hence bluetoothctl just sits there forever waiting for a dbus response [17:27] yay [17:28] lol [17:28] looking at code can be revealing :P [17:28] hehe [17:28] how do you know which port an app will use? [17:28] Guest91403: if you go to webdm you can find a link there afer navigating to the app itself [17:29] Guest91403: otherwise check the package.yaml in the snap and look for port [17:29] some are nice and specify which port they want to use [17:29] in 15.04 [17:29] if all that fails run sudo lsof | grep LISTEN [17:29] and you will see what it listens to [17:29] * ogra_ thinks having a "snappy list-ports" command would be nice [17:30] listing yll the ports assigned to the packages [17:30] *all [17:30] snappy list-ports doign a magic lsof would work yeah. [17:30] yeah, or netstat [17:30] netstat doesnt help you finding the ports for a procss, no? [17:30] i can only see all ports open [17:30] but not figure who is doing what [17:31] it lists the binary if you use the right options [17:31] zyga: so how can i give my bluez5 snap powers to use dbus? [17:31] it has amigration-skill etc. [17:31] but taht seems to not work [17:31] ubuntu@localhost:~$ sudo netstat -tulp [17:31] Active Internet connections (only servers) [17:31] Proto Recv-Q Send-Q Local Address Foreign Address State PID/Program name [17:31] tcp 0 0 *:22 *:* LISTEN 981/sshd [17:32] you mean go to webdm.local in my browser and then there will be a link? [17:32] Guest91403: http://webdm.local:4200 -> navivate to the app [17:32] and then there is a link at the top\ [17:32] Guest91403, go to http://webdm.local:4200/ [17:32] ya i did that [17:32] ogra_: ok didnt nkow about -tulp i guess [17:32] :) [17:32] there is a link in the details page of the app [17:32] I see it install on my RPi but when I click it I just see information about it [17:32] asac: is there a log to check to see why my app didn't start once the snap was installed? [17:33] whizzo: snappy service logs PACKAGENAME [17:33] has the logs [17:33] hast the aggregate of systemd logs for services [17:33] :) [17:33] we also still have good old /var/log/syslog if you prefer fishing ;) [17:34] whizzo: for debugging its sometimes easier to not use service: but rather binaries: [17:34] and then see how it explodes when running that command [17:34] sandbox env etc. should be same, so when that starts working you can make it a service again [17:34] but service logs should have all [17:34] asac: It says Unknown command 'service' [17:34] hmm, is that 15.04 ? [17:34] whizzo: post: snappy list [17:35] right [17:35] if you are on an ancient 15.04 build you wont have snappy service... but recent 15.04 has it [17:35] yep [17:35] asac: ubuntu-core 2015-07-29 4 ubuntu [17:35] yeah thats too ancient [17:35] wow [17:35] I click on the app and I see links for Details, which has a small description, and then settings and reviews, which have nothing. [17:35] thats oooold [17:35] whizzo: what board are you on? [17:36] if its a pi2 i would just update to latest [17:36] Guest91403, ah, if the app would have declared a port there would be some link called "Open" [17:36] asac: I downloaded the latest OVA from the snappy site. [17:36] ok OVA is utlemming [17:36] utlemming: :)? [17:36] whizzo: can you just snappy update that thing? [17:36] sudo snappy update [17:36] and see if it explodes [17:37] (back up first) :) [17:37] asac: yep, doing it now [17:37] ok lets cross fingers [17:38] asac: when I run the snappy update, it looks like it pulls down 2016-01-20, but when I reboot it still comes up as 2015-07-29 [17:39] yeah, could be thats OVA [17:39] whizzo: you run on a mac? [17:39] asac: yes [17:39] if you are on linux just go for a normal qemu [17:39] so yeah, i am not sooo sure what to do there. guess wait for utlemming to get back to you and for now test on the arm board [17:39] rather than the amd64 virt thingy [17:40] asac: ok, will try on an ARM board. [17:41] whizzo: curious, did you see anything about grub when OVA booted? [17:42] or is that thing not using grub? [17:43] asac: it boots so fast its impossible to read the startup messages. Is there something in the filesystem I can check? [17:43] asac: all the old ways are still there via the migration skill. the goal is to have them be gone though. see https://bugs.launchpad.net/snappy/+bug/1543220 [17:43] Launchpad bug 1543220 in Snappy "skills and migration-skill in particular needs more documentation" [Critical,New] [17:44] jdstrand, yeah, but a snap that was built before the skills appeared will be broken [17:44] those are broken regardless [17:44] snap.yaml [17:44] well, it installs [17:44] but doesnt seem to apply the apparmor rules [17:44] no [17:45] there is no compat in 16.04 for the old stuff [17:45] yeah [17:45] ie, if you used 'caps' you must move to the migration skill format [17:45] * ogra_ desnt know what the blues5 package uses though [17:45] but timely it fits into the big changes with its upload [17:46] it uses the new format [17:46] then it isnt complete or something [17:47] I tested package install on a vm, all the profiles were added [17:47] apparmor seems to have " /etc/dbus-1/system.d/bluez_* rw," ... but the package cant create its dbus file [17:47] (teh service one) [17:47] he forgot to add that rule to the policy [17:47] /etc/dbus-1/system.d/bluez5_bluez_5.37-1-armhf.conf: Permission denied [17:48] ah [17:48] thats what you get when pushing someone to do last minute uploads before his vacation :) [17:58] jdstrand: well, so i have the bluez5 snap installed from store [17:58] jdstrand: it uses migration-skill [17:58] buit all those seem to not work anymore [17:58] jdstrand: i am on ubuntu-core 2016-02-03 16.04.0-7.armhf canonical [17:58] see above [17:58] there is a missing rule [17:59] that will make it fail to work [17:59] I suspect [17:59] jdstrand: hmm. so can i just fix the snapy.yaml? [17:59] or you say it wont work in all cases? [18:00] jdstrand: http://paste.ubuntu.com/15017801/ [18:00] is the current one [18:01] you need to adjust meta/bluez.apparmor to have: /etc/dbus-1/system.d/bluez5_bluez_* rwk, [18:01] ok one sec [18:01] asac: are you repacking the snap? [18:01] jdstrand: it currently has: [18:01] # Allow replacing our dbus policy configuration file until [18:01] # snappy has a better way to do this. [18:01] /etc/dbus-1/system.d/bluez_* rw, [18:01] jdstrand: no i am not [18:01] right [18:01] i am trying to assess whether there is hope to use this [18:01] ah i see [18:02] let me fix it [18:02] bluez5 [18:02] ok, I guess I didn't adjust the rule properly when moving from bluez to bluez5 [18:02] jdstrand: so the other thing i get is even nastier :) [18:02] when I try to ssh into my RPi I just get permission denied. I am using the same password that I login with though [18:02] but it is going back to bluez aiui after awe returns from vacation [18:02] that is still tbd aiui [18:02] nevermind [18:02] jdstrand: the whole dbus stuff [ 297.130585] audit: type=1107 audit(1455211145.162:20): pid=724 uid=100 auid=4294967295 ses=4294967295 msg='apparmor="DENIED" operation="dbus_method_call" bus="system" path="/" interface="org.freedesktop.DBus.ObjectManager" member="GetManagedObjects" mask="send" name="org.bluez" pid=966 label="bluez5_bluetoothctl_5.37-1-armhf" peer_pid=894 peer_label="bluez5_bluez_5.37-1-armhf" [18:02] i forgot the ubuntu@ [18:02] is jdstrand is that also caused by that line? [18:03] if so i will repack it now [18:03] not by that line, but the same reason [18:03] let me look at it [18:04] maybe the dbus paths also have 5 now? [18:04] like /org/bluez5? [18:04] and interface org.bluez5 ? [18:05] no [18:05] this is in meta/framework-policy/policygroups/client [18:05] where it says 'label=bluez_bluez_*' [18:06] it should be label=bluez5_bluez_* [18:06] what a mess :( [18:06] asac: are you planning to upload? I think I should clean this up if not [18:07] jdstrand: i dont plan to upload, just unblock [18:07] asac: ok, so adjust your policy for that. I'll upload after my meeting in a few minutes [18:07] jdstrand: there is no meta/framework-policy/policygroups/client [18:07] just seccomp [18:08] nevermind [18:08] found it [18:08] asac: sorry, meta/framework-policy/apparmor/policygroups/client [18:09] garbage collection impossible: prerequisites untrue: remove /snaps/bluez5/5.37-1-armhf/command-bluetoothctl.wrapper: read-only file system [18:09] that is copied to /var/lib/snappy/apparmor/policygroups [18:10] and then on policy compile, the contents of that put in the app's profile in /var/lib/snappy/apparmor/profiles [18:11] works :) [18:11] you are rockstar for me jdstrand :) [18:11] damn ... [18:11] nice! [18:11] wonder why this got uploaded without working :) [18:11] that is a long story [18:11] so the dragonboard always changes its MAC until i put it in a txt file somewhere on disk [18:12] hehe [18:12] dont worry. i know why i think [18:12] * ogra_ hatse the world [18:12] *hates [18:12] awe had a working snap when named 'bluez' [18:12] jdstrand: let me know when you uploaded the clean version so i can validate [18:12] i am mostly interested how i can then allow a snap to use the RFCOMM devices [18:12] store stuff, I renamed, fixed a few things but not all, but couldn't test locally due to lack of hw [18:12] so i will need your help sooner or later i guess [18:13] i hope its a rfcomm tty [18:13] it will probably be in ~1 hour [18:13] so this makes me thingk we cant easily get away with only attaching the modules to a generic initrd :( [18:13] yeah no hurry. have to go for lunch now taht bluetoothctl is not hanging anuymore :) [18:13] ogra_: what? [18:13] we can [18:13] asac, well, not if you want a persistent MAC on the dragonboard [18:13] changing macs is normal in these worlds :) [18:14] not in my world ! [18:14] ogra_: do you want a persistent mac that is the same everywhere? [18:14] * ogra_ stomps foot [18:14] ! [18:14] if not then no initrd will help you :) [18:14] otherwise putting a txt file is fine [18:14] i want a MAC like MACs are defined [18:14] in kernel snap ... just dont want the bootlogic in there [18:14] unique in the world [18:14] modules + hacked stuff for the board is ok [18:14] or for the series of boards :) [18:15] and you can only read it from the kernel cmdline that the lk sets [18:15] we could also do the text file in the gadget? [18:15] so i need to parse /proc/cmdline and extract it from there [18:15] and then write to a txt file [18:15] and i cant really do that from the rootfs [18:15] yeah thats fine [18:15] thats db specific hackery [18:15] (well, i could, but that would even be more ugly) [18:15] like in init-bottom dir [18:16] or something [18:16] asac, exactly ... so we need an opportunity to inject device specific scripts into the initrd [18:16] from the non-generic side [18:16] ogra_: yeas, but all the initrd merging does [18:16] is adding stuff to the OS one [18:16] so you can add other files there too [18:16] \beyond modules [18:16] well, that wasnt really the plan [18:16] just put the file you want in the kernel snap initrd and thats cool [18:16] its fine [18:16] but yes, we will have to [18:16] the plan can be adjusted [18:16] and its also in the spirit [18:16] yeah, but it gets more complex [18:16] as long as all the platform indpendent stuff goes into OS initrd i am happy [18:17] dont think it does [18:17] just merging (or even loop mounting) /lib/modules is a lot easier [18:17] we need scripts for firmware upload hackery sometimes too there etc. [18:17] we dont want to do that [18:17] we want to concat the initrds [18:17] i do [18:17] taht was plan from day one [18:17] yeah, thats messy [18:17] yeah, better drop that idea and join us concat folks [18:17] :) [18:17] but anyway [18:18] i guess i'll be overruled ... [18:18] well, you overruled yourself :) [18:18] hehe [18:18] ok dinner [18:18] * ogra_ used a VHS as well ... even though there was betamax [18:18] ttyl [18:18] haha [18:18] yeah loop moubnting modules feels similar to betamax [18:18] it does [18:18] and drops all the duplication :) [18:25] kyrofa, did it work for you? [18:25] elopio, want to give the PR a look? [18:25] sergiusens, oh I didn't notice the new push [18:27] kyrofa, I suppose it does given the tests pass, but I'll leave it up to you ;-) [18:29] sergiusens, indeed, that works :) . I'll let elopio take a look before merging [18:30] sergiusens: oh, sorry. Got too deep debugging in the cloud. [18:30] give me a minute. [18:36] kyrofa, yeah, a release comes after this; I hope we get a green adt run [18:36] Hello! It's Arron from the Mycroft team. [18:37] I was wondering a few things, how might we integrate continuous integration when building snaps for the raspberry pi via Jenkins? [18:38] aatchison, oh hello; your mere presence reminds me I need to pick something up [18:38] hehe [18:39] aatchison, you can look at snapcraft itself; we build a ton of snaps with travis; it shouldn't differe much; elopio is working on building that stuff on jenkins fwiw [18:39] ok, cool [18:39] I'm jsut wondering about cross-compilation [18:39] but we are building for amd64 here. [18:40] I tried to attach to openvpn via an lxd container, but there is no tun device...not sure how to create one [18:40] yeah, we currently don't have cross-compilation. You would need an armhf machine to build for rpi. [18:40] my thought was, jenkins slave in lxd -> push via ssh to the host snappy-core host machine via ssh? [18:41] I saw a similar thing with docker containers,we already have a server [18:41] hey aatchison ! [18:41] Hey! [18:42] * ogra_ feels the same as sergiusens [18:42] oh the guilt ! [18:42] lol [18:42] no worries or hurries [18:42] i have the mycroft snapcraft.yaml ope in a terminal though :) [18:42] *open [18:42] awesome! [18:42] (didnt close it since SCaLE :P ) [18:42] hahaha [18:42] aatchison: https://github.com/ubuntu-core/snapcraft/blob/master/examples_tests/tests.py#L293 [18:43] ahh, thanks:D [18:43] if you run this in an armhf machine, and your testbed is the ip to the rpi, you will have it working. [18:43] aatchison: what I didn't get was your lxd comment. Where are you missing the tun device? [18:44] aatchison, is your jenkins local? [18:45] aatchison: this is what I'm struggling to fix today: https://github.com/ubuntu-core/snappy-jenkins/blob/master/containers/jenkins-master/config/jobs/github-snapcraft-integration-tests-cloud/config.xml [18:45] jenkins, docker, snappy on the cloud. Maybe you'll find it useful. [18:45] oh elopio don't get distracted :-P [18:46] sergiusens: I'm so close... I test to fix. [18:46] well, I was trying to connect to openvpn ( our jenkins server connects to slaves that way), there is no tun device in /dev/net/ in an lxd container [18:46] but I need to make your review... agh. I'm trying :) [18:46] nice. I'll take a look at that script and see if I could push via another pi on the local network [18:48] aatchison: we have a reverse proxy that solves the communication between github and the lab that's behind vpn. [18:48] ogra_, asac: no real reason [18:48] ogra_, asac: and other than startup support isn't special in bluez for systemd [18:48] morphis, yeah, was a packaging error it seems (due to the bluez5 rename) [18:48] elopio: that's a good idea [18:48] ogra_: possible [18:48] looks like asac and jdstrand solved it [18:54] aatchison: what we haven't solved yet is the provisioning of the rpis. http://linux.codehelp.co.uk/?p=235 [18:55] hmm [18:58] I'm using SPI to burn Arduinos via Raspberry Pi GPIO, we are actually wondering if a small bootloader could be written to the SD card the other direction [19:02] aatchison: ogra_ is the bootloader guy. And plars is the hardware lab guy. [19:02] they might have tips. At least about things that didn't work :) [19:07] haha, cool [19:29] elopio, add comments ;-) [19:41] ogra_, morphis_: fyi, out of my meeting now and will work on getting a fixed up bluez5 [19:41] uploaded [19:42] jdstrand: oh, is something wrong with the snap? [19:42] in the apparmor policy [19:43] two rules need to be updated [19:43] it is what ogra_ was saying as ac and I worked out [19:43] I'm just saying, I'm going to get that into the store [20:07] is 'snappy login' persistent - i.e. you just do it once on install and then never again? [20:09] robert_ancell, yes, I don't think we kill expired keys though [20:09] sergiusens, thanks [20:10] sergiusens, and it does use a standard token? So it could be integrated with Ubuntu Online Accounts? [20:10] robert_ancell, not using online accounts, something closer to click-toolbelt (just like snapcraft) [20:11] robert_ancell, this is the gist of it https://github.com/ubuntu-core/snappy/blob/master/snappy/auth.go [20:12] sergiusens, thanks [20:15] robert_ancell, if you want to runtime check on some online account features that would be sweet; we have some go code you can copy/paste or be inspired from in lp:account-polld even though we use the glib api instead of the new qt one [20:16] sergiusens, so you think it would be appropriate for snappy to check u-o-a for credentials? [20:16] and fall back to ~/snaps/snappy/auth/sso.json if u-o-a is not there? [20:16] robert_ancell, yes; a runtime thing I guess would be valid; although I am not core to snappy anymore so maybe ask Chipaca first [20:17] robert_ancell, or wait until Monday where we will be in the same room ;-) [20:17] yeah, will do that too! [20:17] unless this is pre flight home work :-P [20:17] it is [20:18] um [20:18] so either runtime or though configuration [20:18] 'snappy login' is going away, at least in its current form [20:18] you want to talk with facu about that [20:18] Chipaca, oh, right [20:18] he is not here [20:18] and unless online accounts support macaroons already you want to talk with facu and/or beuno *stat* :-) [20:21] elopio, kyrofa https://github.com/ubuntu-core/snapcraft/pull/319 [20:22] robert_ancell, Chipaca, so [20:22] I don;t understand why online accounts comes into play [20:23] elopio, kyrofa just in case I pushed a tentative deb for that here https://launchpad.net/~snappy-dev/+archive/ubuntu/snapcraft-daily/+packages?field.name_filter=snapcraft&field.status_filter=published&field.series_filter=xenial [20:23] beuno, robert_ancell is trying to tie snappy into the authentication, access and secrets storage (and more?) that exist already on the desktop [20:24] at least that's my understanding of what he's doing :-D [20:28] JamesTait: fyi in case you are looking for something to do, https://code.launchpad.net/~jdstrand/click-reviewers-tools/lint-snapv2/+merge/285666 is ready for review [20:56] asac: I installed my snap on a BBB running 2016-01-20 and it doesn't appear to install java [20:58] asac: I take that back, it installs java but I get an "Exec format error" when trying to run the java executable [21:01] kyrofa: wtf catkin? It generates a bash file and a python file from templates, and then calls the python file from the bash file, and the python file calls python -c with a script in a string. [21:01] whyyy??? [21:02] I had to put like 30 echos to understand that, and I'm still not close to understand the error. [21:10] elopio, fwiw, is my changelog good to land? just asking since the last one was waiting on you without notice ;) [21:15] Has anyone else seen an issue with building a Java-based snap for BBB (specifying armhf as the architecture in snapcraft.yaml) and the java that is installed says "Exec format error" when you try to run it? [21:18] It looks like it installed the amd64 version instead of the armhf one [21:30] asac: fyi, updated bluez5 in the store [22:18] cd [23:28] whizzo: exec format error means you included binaries fro different architecture [23:28] you need to build on arm for arm and on x86 fof x86 [23:28] as i said, snapcraft desnt support crossing [23:28] so best off make two snaps really [23:29] then you can probably just use the maven [23:29] plugin as well [23:29] asac: I got that error when building a snap specifically for armhf only [23:30] asac: oh, I see, you're saying I can't build an armhf snap on an amd64 ubuntu machine? [23:35] asac: It seems that even though the only architecture in the snapcraft.yml file is armhf, the jdk part still lays down the amd64 version of the JVM