[07:33] good morning [07:34] hey dholbach! [07:36] salut didrocks === joc|away is now known as joc_ [12:02] Good morning [12:06] Hey, anyone here to help with snappy service problem? [12:52] Marin_, you have to be a bit more specific i guess :) [13:03] ogra_, can the auto-update be disabled in ubuntu core? [13:04] kyrofa, yeah, tghrough snappy config ubuntu-core [13:05] echo -e "config:\n ubuntu-core:\n autoupdate: false\n" | sudo snappy config ubuntu-core -- - [13:05] something like that should work [13:05] ogra_, thank you! [13:39] mvo, regarding bug 1546821 .... do we need to omit -no-xattrs too for os snaps ? (see the last comments) [13:39] bug 1546821 in Snappy "please add -all-root and -no-xattrs to mksquashfs options" [Undecided,New] https://launchpad.net/bugs/1546821 [13:41] ogra_: the OS snap is not using xattrs AFAIC [13:41] AFAIC [13:41] AFAIK [13:41] *sigh* [13:41] ogra_: this is something in snappy [13:41] mvo, so it wouldnt break if that was used ? [13:41] ogra_: I think it won't break either way [13:41] ok [13:42] ogra_: but we can not use -root-only (or wahtever it was called) [13:42] that will definitely break [13:42] just making sure that fixing bug 1557513 will be enough to drop snappy build [13:42] bug 1557513 in Snapcraft "please add option to override -all-root in "snapcraft snap"" [High,New] https://launchpad.net/bugs/1557513 === jkridner is now known as beagleslackbot === beagleslackbot is now known as jkridner [14:03] Hello! [14:06] mvo, Chipaca http://paste.ubuntu.com/15391784/ :( [14:06] (this is todays dragonboard image) [14:08] Hi ryanleesipes! [14:09] ogra_: in qemu? [14:09] mvo, ? arm64 in qemu ? [14:09] ogra_: or on real HW? [14:10] thats the dragoboard image i just built from the most recent snaps on cdimage [14:10] there is another issue [14:10] ogra_: and its running on your dragonboard? hm, the traceback looks scary [14:10] http://paste.ubuntu.com/15391819/ [14:10] ogra_: its not even touching our go code yet when it panics [14:10] mvo, well, snap install seems to work, it is only the old "snappy install" [14:11] ogra_: interessting [14:11] 2016/03/15 14:08:31.229390 kernel_os.go:167: can not get boot settings: cannot determine bootloader [14:11] ogra_: can not get boot settings is also not cool [14:11] this one looks scary though [14:11] ogra_: how does ls /boot/uboot look ? do you have the uboot.env there? [14:11] perhaps due to having grub installed now ? [14:11] ogra_: yeah, I was wondering this too [14:12] ogra_: iirc the code looks for a grub.cfg in /boot so it should not affect it, but maybe it does [14:12] oh man [14:12] ubuntu@localhost:~$ ls /boot/uboot/ [14:12] ubuntu@localhost:~$ [14:12] ... [14:12] ubuntu@localhost:~$ ls /boot/efi/ [14:12] EFI install.yaml [14:12] canonical-dragon-linux.sideload_IPAIJHReTMLK.snap snappy-system.txt [14:12] dtbs uboot.env [14:13] ubuntu@localhost:~$ [14:13] lovely [14:13] so it now mounts under /boot/efi [14:14] ubuntu@localhost:~$ mount |grep boot [14:14] /dev/mmcblk1p8 on /boot/efi type vfat (rw,relatime,fmask=0022,dmask=0022,codepage=437,iocharset=iso8859-1,shortname=mixed,errors=remount-ro) [14:14] /dev/mmcblk1p8 on /boot/grub type vfat (rw,relatime,fmask=0022,dmask=0022,codepage=437,iocharset=iso8859-1,shortname=mixed,errors=remount-ro) [14:15] and .... [14:15] ubuntu@localhost:~$ sudo mount /dev/mmcblk1p8 /boot/uboot [14:15] ubuntu@localhost:~$ snappy list [14:15] Name Date Version Developer [14:15] canonical-dragon 2016-02-12 0.7.1 canonical [14:15] canonical-dragon-linux 2016-03-15 IPAIJHReTMLK sideload [14:15] ubuntu-core 2016-03-15 IPAIJIIJHeDL sideload [14:15] Reboot to use canonical-dragon-linux version IPAIJHReTMLK. [14:15] Reboot to use ubuntu-core version IPAIJIIJHeDL. [14:15] (note that i installed webdm before i noticed that issue ... there is no webdm listed now) [14:15] it sits under /snaps though [14:17] ricmm you around? [14:23] /msg nickserv set password 6V8gEoK.HYzUY6XpAsF [14:23] lol ops [14:24] mvip, might want to change that now... ;) [14:24] kyrofa lol yeah done already [14:33] Mar 15 14:32:53 localhost kernel: [ 1416.561031] audit: type=1400 audit(1458052373.895:292): apparmor="DENIED" operation="create" profile="xkcd-webserver.canonical_xkcd-webserver_16.04-4" pid=1908 comm="xkcd-webserver" family="inet" sock_type="stream" protocol=0 [14:33] hmpf [14:33] so our latest rootfs is definitely heavily broken in various ways [14:33] it needs 'network-client' [14:34] oh, maybe this is what didrocks was talking about yesterday [14:34] (bootloader broken, snappy list not showing packages) [14:34] (snappy itself throwing exceptions) [14:35] jdstrand, yeah, i guess [14:35] didrocks: btw, the 'description' warning from the review tools for xkcd was accurate-- there is a 'description' tucked in the apps section. I made that warning more clear in trunk. the other issue I fixed in trunk [14:35] (couldn't find binary) [14:35] ubuntu@localhost:~$ snappy list|grep webdm [14:35] ubuntu@localhost:~$ snappy list|grep xkcd [14:35] xkcd-webserver 2016-03-14 16.04-4 canonical [14:35] * ogra_ scratches head [14:35] didrocks: there is a new review tools rollout to the store imminent, but these changes won't be until the next one [14:35] ubuntu@localhost:~$ ps ax|grep webdm [14:35] 1244 ? Ssl 0:02 /snaps/webdm.canonical/0.16.1/bin/aarch64-linux-gnu/snappyd [14:36] funnily it runs and i can use itr === charles_ is now known as charles [14:39] jdstrand: ah, great! We will fix the description with next upload then :) [14:42] Hey didrocks. How are things? You recovered from MWC? ;) [14:42] mvo, mind if i revert the grub stuff for now ? [14:44] ogra_: ok [14:48] mvo, at least the grub migration error on boot is now gone on the dragonboard :P [14:54] ogra_: heh [14:55] mvip: I fortunately did yeah! :) Quite busy to prepare content for 16.04 (dev website, slide decks and such) [14:55] mvip: how about you? [15:08] Hey ogra_ I'm trying to automate a way to get the writable partition on an external HD. Ideally I'd be able to create two images, one for the SD card that doesn't contain the writable partition, and one for the external HD that does. Then each image can be flashed to its respective device and voila. Does that sound possible? [15:17] kyrofa, we only have two partitions (well, plus bootloader if needed) [15:17] kyrofa, technically you want to split bootloader+boot vs writable ... [15:18] thats indeed possible to do using scripts and kpartx [15:19] (i.e use sgdisk/fdisk to find the writable partition in the img file, create a fresh img for writable, format it, mount it, copy the contents, delete the writable partition on the original img) [15:21] ogra_, ah, excellent [15:21] ogra_, I'm going to take a crack at this. I'm sure I'll have questions if that's alright? [15:21] sure [15:33] fgimenez: are you coming? [15:34] elopio, yep omw === kickinz1_ is now known as kickinz1 === chihchun is now known as chihchun_afk [16:05] didrocks: good good. Lots of events as of lately, but i guess one cannot complain ;) [16:05] didrocks: any progress on the dev website? [16:12] mvip: quite some on the getting started, the rest will take longer obviously :) [16:13] mvip: I can share with you some mockups if you are interested [16:47] zyga-phone: with all the feedback on https://github.com/ubuntu-core/snappy/pull/658 and your most recent comment, thinking I'll hold off on my portion of the review? [16:54] jdstrand: looking [16:54] jdstrand: ah, I think that has stabilized now [16:54] jdstrand: a review of that would help [16:54] jdstrand: from the security POV [16:55] ok [16:56] jdstrand: note that I've dropped the checks for UID/GID since it makes no sense in snapd context [16:56] jdstrand: and it was making the code a lot more complex given the atomic renames we do === vrruiz_ is now known as rvr [17:43] mvo, just FYI, after removing grub the dragonboard image totally behaves ... no more segfaults or anything ... seems seeding grub confused the system mmore than just through mounting everything under /boot/efi [17:44] hmm, though snappy list still doesnt show webdm [17:44] oh ! [17:44] it shows webdm if i sudo ... [17:44] seems to be a permission prob [17:46] wow, weird [17:46] so installing xkcd-webserver makes it show up without me using sudo [17:46] (but webdm is hidden from the list) [17:47] * ogra_ doesnt get that [17:47] mvip: yes? [17:47] ricmm, i had to remove grub from arm64 again, it made everything explode [17:48] ricmm, what is it neede for ? [17:48] *needed [17:48] ogra_: well obviously arm64 devices that use grub ;) [17:48] what explodeD? [17:48] the whole boot process ... everything got mounted under /boot/efi regardless [17:48] that in turn completely broke snappy itself [17:49] ubuntu@localhost:~$ mount |grep boot [17:49] /dev/mmcblk1p8 on /boot/efi type vfat (rw,relatime,fmask=0022,dmask=0022,codepage=437,iocharset=iso8859-1,shortname=mixed,errors=remount-ro) [17:49] /dev/mmcblk1p8 on /boot/grub type vfat (rw,relatime,fmask=0022,dmask=0022,codepage=437,iocharset=iso8859-1,shortname=mixed,errors=remount-ro) [17:49] (thats on a uboot system) [17:50] that caused: http://paste.ubuntu.com/15391784/ and http://paste.ubuntu.com/15391819/ [17:50] why would that have happened [17:50] removing the grub package and everything is back to normal [17:50] its a core bug then [17:51] yeah, somewheer [17:51] *where [17:51] could you send mvo/me an email about this with all the info [17:51] surely needs fixing .... but i guess we are not expecting a grub based arm64 arch within the next few weeks ? [17:51] I have one on my table right now [17:51] damn [17:51] port it to uboot :P [17:51] well, they went through an effort to make it uefi :p [17:52] anyways, just send us an email [17:52] should be a simple bug [17:52] will do [17:52] no reason why things cant coexist :) [17:52] * ricmm adds to board [17:52] well, doesnt u-d-f handle them differently ? [17:52] i.e. grub is installed out of the rootfs [17:52] it does, but it should work fine [17:52] while uboot has to come with the gadget [17:52] i.e. armhf supports it [17:53] and iirc there are also checks for mountpiints [17:53] only thing we did was add arm64 bits [17:53] right it might be doing some hackish thing to determine the bootloader setup [17:53] i.e. if /boot/grub exists (which the package created) we assume grub [17:53] by probing mountpoints [17:53] right [17:53] thats not necesarily a nice thing to do :) [17:53] yeah, i guess that causes the wrong mounts [17:53] the other bits are fallout [17:54] ok send the email will track tomorrow [17:54] thanks for pointing that out [17:54] will do [17:55] * ogra_ is close to having auto-uploads of the daily builds ready ... then we should have daily snaps in the store ;) [17:56] and with these snaps auto-resize on dragonboard finally works \o/ [17:56] \o/ [17:56] (these = http://cdimage.ubuntu.com/ubuntu-core/daily-preinstalled/current/ ) [18:07] ogra_: woot, great [18:07] ogra_: is there something I should change in ubuntu-image? [18:08] hmm, the amd64 image exposes the same weird snappy list behavior [18:09] zyga-phone, nah, i plan to auto-push them to the store ... rolling edge should always have the dailies ... promoting them to rolling stable should then stay a manual process [18:09] ogra_: nice [18:09] ogra_: will existing images auto-resize? [18:09] (perhaps we can tie that to the testing infra later though) [18:09] ogra_: or is that some firstboot speciallity [18:10] the check happens on every boot ... if there is more than 10% free space it will resize to full disk size [18:10] so technically not a first boot thing ... but likely to run on first boot and never again :) [18:11] with the smally all-snaps setup we could now default to 1GB images btw [18:12] it is ~180MB per kernel+os snap (by 2 since we have two of them installed usually) ... and ~130MB for /boot