[00:00] kyrofa, that seemed to fix it. The modules and everything load up, but it is still quite odd that a clean custom image I build with ubuntu-device-flash shows no snaps installed until I install my snap, and then it downloads ubuntu-core and then when I reboot my system, it crashes. A clean install would show three snaps, ubuntu-core, my custom-kernel and pc, but now it says no snaps installed. [00:01] wililupy, sorry, I'm not sure what the situation is regarding all-snaps images. Have you tried downloading one of mvo's? [00:02] I was just going to try that when my snapcraft snap finished and I tested that. [00:02] Looks good so far, I have a dependency missing, so I have to figure that out... [00:03] kyrofa: I am using a modified version of your fboss snap from github to work with the Wedge 100. [00:03] wililupy, ah, oay [00:03] okay* [00:04] kyrofa: I'm going to push my updates to my fork and then look at mvo's all-snap tomorrow. I am including the kmods with the snap, so it is a little easier to try with the all-snap image instead of a custom kernel. [00:17] hi ev, niemeyer - looking to start using spread asap while we're redeploy some openstack ci infra. i've got a fresh xenial instance, can launch and list lxc a-ok as my user, but can't get spread to do the examples: http://pastebin.ubuntu.com/23091151/ ideas? [00:51] beisner, yeah, try running it from source instead [00:51] beisner, the snap doesn't have permission to reach out like that [00:51] beisner, do you need help with that? [00:52] (qemu backends have the same problem) [00:52] The snap is only really useful if you're using remote backends === al_ is now known as Guest55102 [02:03] $ sudo snap refresh --beta --devmode ubuntu-device-flash [02:03] - Setup snap "ubuntu-device-flash" (9) security profiles (cannot setup apparmor for snap "ubuntu-device-flash": cannot load apparmor profile "snap.ubuntu-device-flash.ubuntu-device-flash": cannot load apparmor profile: exit status 10 [02:03] what? [02:03] PR snapcraft#759 opened: Remove the useless call to os.path.join from the go plugin [02:12] well rebooting and reinstalling fixed that [02:45] hi kyrofa thanks for the info. just for kicks i removed the snap, reinstalled it in devmode, still the same. so, go get ftw? [02:51] what does this mean ? [02:51] Cannot get permission from the store to upload this package. [02:51] trying to have launchpad build a snap package for me [02:52] it says: [02:52] If you ask Launchpad to automatically upload builds of this snap to the store on your behalf, then the login service will prompt you to authorize this request. [02:52] so i expected it to ask me or something, but it just says fail [04:36] PR snapd#1765 opened: many: move network initialization to a separate service [04:48] PR snapcraft#760 opened: Removed statements not unused by the tests === chihchun_afk is now known as chihchun [07:16] PR snapd#1766 opened: image, asserts: add new asserts.Fetcher.SavedRefs() helper [07:22] good morniong [07:22] *morning [08:18] Question about snaps and folder access / interfaces. For my snap I need to mount a folder so that the normal users can write to it and the snap can read it. [08:18] how can I achieve that? [08:24] PR snapd#1766 closed: image, asserts: add new asserts.Fetcher.SavedRefs() helper [08:37] mvo, mind if i ubpublish all the canonical-* snaps ? [08:37] ogra_: not at all === AL_ is now known as Guest80394 [08:58] Hi ! I am sorry to disturb the channel. I have a short question: I try to connect a USB drive to my snap. I did not find a way to do this with the "strict confinement" mode. It only works with the "devmode". Is there a way using interfaces ? [08:59] Guest80394: hey, I don't believe there is right now but it should be simple to add [09:00] Guest80394: look at the home interface, a similar interface could mediate access to /media or /run/media or anything like that [09:00] Guest80394: it should be very easy to contribute if you want to [09:01] ok, I will look at this ! [09:01] mvo, hmm, did we seed subiquity already ? my serial console is a huge readable mess after rebooting on all my images [09:01] *unreadable [09:01] mvo, http://paste.ubuntu.com/23092374/ [09:02] the last line fill multiple pages [09:02] '*fills [09:02] Guest80394: look at zygoon.pl, I posted some articles about this there [09:02] Guest80394: and ping me here for advice :-) [09:02] ogra_: it landed last night [09:02] ogra_: apparently a hard landing [09:02] Thanks a lot ! [09:03] mvo, extremely buggy ... and it seems to run regardless if the system is cionfigurede [09:03] *configured [09:03] ogra_: lets talk to cyphermox, mwhudson or slangasek about the bugs in console-conf: http://paste.ubuntu.com/23092374/ [09:03] on the dragonboard i now have a network config dialog [09:03] ogra_: if its too buggy we can revert the upload [09:03] ogra_: its a single line in livecd-rootfs from last night [09:04] it is more than console-conf ... unless that also does network config etc ... and it is definitely not safe for serial [09:04] ogra_: it does [09:04] ogra_: and yeah, it takes over the gettys [09:04] ah, i thought it cionfigures the coneole :) [09:04] ogra_: its a bit misleading the name [09:05] right, if it does so, it should probably use settings from d-i or so to make sure you dont end up with multiple lines of control chars :) [09:05] in any case i cant do any unattended reboot anymore [09:05] ogra_: and I want my getty back [09:06] ogra_: if its too anoying, just revert, I think that is fine [09:06] the boot hangs at a prompt (which i cant even see on the pi... but see on the dragonboard (different serial consoles i guess)) [09:08] mvo, oh, fun ... and it doesnt recognize my configured wlan0 devices on both boards ... so i'm stuck in the config screen [09:08] i'll unseed it once i reported all the bugs ... oh my [09:09] ogra_: oki doki, thanks [09:09] PR snapd#1747 closed: overlord, daemon: add collect attribute hooks [09:12] PR snapd#1767 opened: overlord, daemon: add collect attribute hooks [09:13] PR snapd#1763 closed: many: clarify/tie down model assertion [09:14] PR snapd#1616 closed: tests, integration-tests: implement the cups-control manual test as a spread test [09:14] PR snapd#1759 closed: asserts: fix GPG key generation parameters [09:17] * ogra_ sighs ... so how do i get out of this .. i cant boot at all anymore [09:18] (it tries to configure the already configured wlan devices, falis and drops me back to the start screen ... there is no way to skip) [09:18] PR snapd#1759 closed: asserts: fix GPG key generation parameters [09:32] Bug #1617231 opened: subiquity kills serial console after regular update [09:35] Bug #1617232 opened: subiquity goes into endless configuration loop, no way to get out [09:36] pitti, i fear in the light of that one ^^^ Bug 1616928 should get a higher urgency [09:36] Bug #1616928: netplan should support managing wlan devices via wpasupplicant [09:37] (assuming that subiquity actually tries to use netplan in the backend) [09:45] Bug #1617236 opened: console-conf falls over trying to create already existing logdir [09:48] PR snapd#1753 closed: asserts: add an account-key-request assertion [09:52] ogra_: ah, I can't get out either, as even though it is online it doesn't accept any email address of mine for registering to the snap store [09:52] what a mess [09:52] ogra_: but yes, wlan is another issue indeed [09:52] pitti, file a bug ..., i'll unseed it now and re-spin ubuntu-core afterwards [09:52] communication breakdown -- when this was discussed a few weeks/months ago, the plan was still "use NM for wifi".. [09:53] pitti, well, the worst is that the console-conf process actually consumes 16-25MB ... [09:53] ogra_: yep, will do; I'm sorting out some other console-conf issues with mwhudson (currently blocked on getting a VM image that actually works) [09:53] ogra_: welcome to python :) [09:53] and it spawns a new process every time i ctrl-c out of it === hikiko is now known as hikiko|ln [09:53] yeah, the worst choice for arm [09:54] i guess that bumps our ram requirements from ~64MB to 128 for embedded boards ... [09:54] :((( [09:55] for those you could skip console-conf somehow perhaps (kernel arg or pre-create the firstboot stamp file etc.)? [09:55] so you mean for all of them ? [09:55] :) [09:55] (like all boards) [09:57] so depressing :/ [10:01] cyphermox, mwhudson, slangasek, i unseeded console-conf again for now, it breaks to many things atm [10:03] woah ... [10:04] ubuntu@dragonboard:~$ ps ax -orss=,args=|grep 24228 [10:04] 24228 python3 /usr/bin/console-conf [10:04] ubuntu@dragonboard:~$ sudo kill -9 24228 [10:04] ubuntu@dragonboard:~$ ps ax -orss=,args=|grep 24228 [10:04] 24228 python3 /usr/bin/console-conf [10:04] very reluctant to be killed :P [10:08] Restart=always [10:09] Bug #1617250 opened: 401 errors trying to install a snap [10:09] Bug #1617251 opened: console-conf unkillable [10:09] pitti, weeeelll ... wouldnt a restart assign a new PID ? [10:09] but systemctl stop console-conf.service also is broken as there are about three endless loops [10:09] ogra_: ^ [10:09] ogra_: err yes, kill -9 not working?? [10:09] right [10:09] it keeps the PID [10:09] no matter how often i kill -9 [10:31] PR snapd#1744 closed: snap: use "up to date" instead of "up-to-date" [10:42] so, i've just made a snap that contains a single Go binary (app) that just makes HTTP connections. to make it work, I had to configure it with network, firewall-control and network-bind plugs. Anyone know why? It seems like I should only need "network". [10:42] also, when I didn't have the required plugs, the binary just hung when I ran it. Is that the expected failure mode? [10:43] it needs to create a socket ... "network" is only for client side things ... for the socket creation you need -bind ... [10:43] not sure why you would need firewall though [10:44] ogra_: how can you be a network client without creating a socket? [10:44] (i definitely dont need it for any of my network server snaps ) [10:45] ogra_: ah, it's possible i don't need firewall-control - i added that first [10:45] ogra_: i'd have thought that network-bind was only necessary for listening on network ports, not making outgoing connections [10:46] https://github.com/snapcore/snapd/blob/master/docs/interfaces.md [10:46] network-bind says "Can access the network as a server" [10:46] network [10:46] Can access the network as a client. [10:46] Auto-Connect: yes [10:46] network-bind [10:46] Can access the network as a server. [10:46] Auto-Connect: yes [10:46] ogra_: yeah. and this program isn't a server. === hikiko|ln is now known as hikiko [10:46] ogra_: it's just making an outgoing connection [10:46] so if your binary actually only does client connections network should be enough [10:47] ogra_: that's what i thought [10:47] aks jdstrand once he is up :) [10:50] hmm, and now i can't replace the snap because (whenever i do snap install or snap remove) i get 'snap "bhttp" has changes in progress' [10:52] well, look at snap changes (and "snap change $number" for details) ... [10:53] hmm, and that's because i aborted a "remove" operation that seemed to be taking forever. i did "snap abort" and i still can't do "snap remove" [10:53] iirc you can also "snap abort" [10:53] give it a bit, it works async ... [10:53] ogra_: yeah, i tried that. but "snap remove" takes forever [10:53] might need some time to time out [10:53] ogra_: well, ok, i'll give it a few minutes [10:54] ogra_: do you think 5 minutes should be enough? [10:54] no idea :) [10:54] I've got a daemon that insists on not being run as root. Should my wrapper create the user and suid to it? Would that actually work? [10:55] no, it wouldnt [10:55] snap daemons always run as root [10:55] ah, looks like this might be the culprit: Aug 26 11:52:56 snazzy umount[9073]: umount: /snap/bhttp/x1: target is busy [10:55] are you in there with your shell ? [10:56] ok, so for now I need to patch unless they have hidden an option in there somewhere to allow it... [10:56] ogra_: no, but there's a process running the binary [10:56] ouch [10:56] stub, i fear so, yeah ... [10:57] Bug #1617275 opened: the home interface doesn't allow access to $HOME/snap-confine [10:58] ogra_: ok, success: i did "kill -9" on the defunct process, then aborted the remove operation and tried again and it worked. [10:58] ogra_: not a fantastic user experience though... :) [10:58] rogpeppe, definitely not, file bugs ;) [10:58] ogra_: will do; it'll be the third bug this morning :) [10:59] i'm at 5 for today :) chase me ;) [10:59] ogra_: https://www.google.co.uk/imgres?imgurl=http://t1.gstatic.com/images%3Fq%3Dtbn:ANd9GcTLjo_Nzd7eLrqGEG9ddoRlkUaAV7Q24KziKV2u5i6R-weg4Kgb&imgrefurl=http://books.google.com/books/about/Snappy_Little_Bugs.html%3Fid%3DrC3iGAAACAAJ%26source%3Dkp_cover&h=769&w=621&tbnid=URJ9QWARZ6Uq5M:&tbnh=160&tbnw=129&docid=bPGBY9Q323zj2M&itg=1&usg=__-RYGSWUfDcjgnJL1FRx3M0oDlwI= [11:00] ogra_: :) [11:00] haha [11:05] ogra_: https://bugs.launchpad.net/snappy/+bug/1617278 [11:05] Bug #1617278: remove hung forever due to defunct process [11:06] thx [11:06] Bug #1617278 opened: remove hung forever due to defunct process [11:08] ogra_: ok, it seems that only network and network-bind are needed. but what's weird is that the binary won't run at all (even to print a help message) without both of those plugs. [11:09] i guess your syslog should have some DENIED lines then [11:10] ogra_: this is the line i see: [11:11] Aug 26 12:10:40 snazzy kernel: [117476.140023] audit: type=1326 audit(1472209840.610:190): auid=4294967295 uid=1000 gid=1000 ses=4294967295 pid=10082 comm="bhttp" exe="/snap/bhttp/x3/bin/bhttp" sig=31 arch=c000003e syscall=49 compat=0 ip=0x5fbcb4 code=0x0 [11:11] rogpeppe: that's bind [11:11] zyga: syscall 49? [11:11] rogpeppe: it gets killed when it cannot use a syscall it tries [11:11] rogpeppe: yes, 49 is bind [11:12] mvo, seeing https://github.com/snapcore/snapd/pull/1684 ... should we also unseed the package (we explicitly have it in livecd-rootfs currently) [11:12] PR snapd#1684: many: drop ubuntu-core-snapd-units package, use release.OnClassic instead [11:12] zyga: hmm, i wonder why... i wonder if i can make it panic when it does that [11:12] rogpeppe: panic? [11:12] rogpeppe: you get SIGKILL from the kernel [11:12] zyga: also, is it right that the binary should just hang when it does that? [11:12] rogpeppe: it kills the offending thread [11:12] zyga: hmm, not the whole process? [11:12] rogpeppe: depending on the way your program works it will just hang [11:12] rogpeppe: nope [11:13] zyga: ok, that's probably the reason - it's a go program with multiple threads [11:13] rogpeppe: yeah, it's unpredictable what will happen [11:13] zyga: any way to configure different behaviour? [11:14] rogpeppe: not at present, I believe there's a bug about switching seccomp to return EPERM, we may be able to revive it [11:15] zyga: ah, i see now - the Go net package uses bind to probe to see if the kernel has ipv6 functionality [11:16] rogpeppe: https://lists.ubuntu.com/archives/snappy-devel/2016-April/001708.html [11:16] rogpeppe: that's correct, there's a separate bug about that, I've done the same investigation [11:16] zyga: so you can't have *any* Go program that imports /net without allowing the network-bind plug [11:17] zyga: i think i'll raise a Go bug about this [11:19] ah, there already is a bug: https://github.com/golang/go/issues/16789 [11:19] I'm looking for the bug on golang [11:19] er [11:19] snapd [11:20] I cannot find it [11:21] i think you can work around that by using a wrapper script with exec, so that systemd handles it differently [11:21] I'm sorry [11:30] Bug #1617264 opened: confusing error when installing snap when not logged in [11:42] hey folks, the icon guidelines link here is broken: https://developer.ubuntu.com/en/publish/creating-a-good-icon/ [11:42] ^^ where should I report that [11:42] and can anyone tell me what dimension a snaps icon should be? [11:45] what do you think is broken about it ? [11:45] (my icons all work fine still) [11:48] ogra_: good point about unseeding the scripts bits [11:49] * ogra_ is happy about every bit he can remove ... :) just working on getting rid of all the partitioning tools and initramfs-tools in ubuntu-core [11:50] (well, at least initramfs-tools-ubuntu-core ... seems apparmor has a hard dep on the generic initramfs-tools ) [11:59] Chipaca: https://github.com/snapcore/snap-confine/pull/120/files [11:59] PR snap-confine#120: Fix regression test when re-exec kicks in [11:59] Chipaca: care for a quick review of what we've just discussed? [12:00] zyga, three things [12:00] zyga, you're not stopping the .socket [12:00] zyga, and, you can start/stop n all together in a single line [12:00] Chipaca: I'm also not using it though I guess I can just stop it and be on the safe side [12:00] oh? [12:00] how :) [12:00] zyga, systemctl stop foo bar baz [12:00] ah [12:00] thanks, doing that [12:00] zyga, also also, systemctl stop snapd.{service,socket,refresh.timer} [12:01] but not sure if the bash these things run in has brace expansion [12:01] I've just spelled it out [12:01] fair [12:02] Chipaca: fixed, I also dropped sudo [12:02] look again please [12:06] zyga, question: instead of “umount $(ls -1d /snap/ubuntu-core/* | grep -v current | tail -1)”, wouldn't “ummount /snap/ubuntu-core/$(readlink /snap/ubuntu-core/current)” work? [12:07] Chipaca: perhaps, I didn't write this, jdstrand did [12:07] Chipaca: I'm happy to improve it when it is green again [12:07] ah ok [12:08] zyga, go for it (once green) [12:08] thanks [12:09] so i seem to have published my snap (https://myapps.developer.ubuntu.com/dev/click-apps/5803/feedback/) but i still can't "snap install" it. anyone know what I'm doing wrong? [12:09] is it just that i have to wait for a while for things to happen behind the scenes? [12:10] rogpeppe: is that snap published to the stable channel? [12:11] rogpeppe, what icon does the revision have next to it? a thumbs up, or a green box? === chihchun is now known as chihchun_afk [12:11] zyga: i think so. (i ran "snap release") [12:11] Chipaca: checking [12:11] Chipaca: a green box [12:11] rogpeppe, and what's the snap called? [12:11] Chipaca: bhttp [12:11] rogpeppe, if i can ask :-) [12:12] Chipaca: so that URL is personal? [12:12] rogpeppe, yes [12:12] Chipaca: is there a URL for the snap that I can share? [12:12] i get “You just tried to access a feature which you don't have permission to use.” [12:13] rogpeppe, what channels is it published to? [12:13] Chipaca: i'm not sure where there's a page i can search for snaps [12:13] Chipaca: stable [12:13] rogpeppe, i can confirm it isn't there :-) [12:13] rogpeppe, I was about to point you at https://snapfui.herokuapp.com/ but it's broken right now :-( [12:13] beowulf, ^ [12:14] Chipaca: what technique did you use to confirm that? [12:14] uappexplorer :) [12:14] * rogpeppe googles that [12:14] rogpeppe, "snap find bhttp" [12:14] rogpeppe, is it ARM-only? [12:14] or i386 or something other than amd64 or all [12:14] Chipaca: no, it's amd64-only [12:14] hmmm [12:14] Chipaca: AFAIK [12:15] Chipaca: i didn't set any architectures [12:15] Chipaca: one mo, i'll send you a screenshot [12:15] rogpeppe, you could wait until somebody with more access than i takes a look, or you can send me a screenshot :-) [12:15] rogpeppe, does it have a little green box next to the version in the UI ? [12:15] rogpeppe, m v o should be able to see things, but he's a busy person [12:15] ogra_, yep [12:15] hmm [12:17] PR snapd#1768 opened: many: automatically restart all-snap devices after os/kernel updates [12:17] Chipaca: http://imgur.com/a/5YksP [12:18] it *looks* like it's published to me, but maybe i'm missing some crucial piece of info [12:18] ogra_, hey there, sorry, this is very out of context now, but where you replying to me above? (the broken link) [12:19] rogpeppe, that should work [12:19] nessita, are you around? [12:19] mattyw, oh, i missed you meant a link, i thought you meant the instructions themselves [12:20] mattyw, not sure where to file it, davidcalle or dholbach might be able to point oyyu to the right place [12:20] Chipaca: right, but... you verified that it's not actually there, right? [12:20] rogpeppe, right [12:20] rogpeppe, hence me poking nessita [12:20] Chipaca: ok, i give up for now. i need to do some actual work :) let me know if you work anything out, please? [12:21] rogpeppe, if you scroll to the very bottom of the page there is also an "unpublish" button ? [12:21] Chipaca, indeed I am [12:21] ogra_: you think i should unpublish so that i can try to publish again? [12:21] nessita, ohi [12:22] nessita, given http://imgur.com/a/5YksP [12:22] rogpeppe, no, i just want to know if that button is there [12:22] ogra_: yes, it's there [12:22] nessita, which is https://myapps.developer.ubuntu.com/dev/click-apps/5803/ [12:22] nessita, what's missing, that snap can't find it? [12:22] Chipaca, let me check [12:22] nessita, thank you [12:23] version 0 looks suspicius, but "in paper" shouldn't be an issue [12:23] * nessita checks more [12:23] tyhicks: https://github.com/snapcore/snap-confine/pull/119 [12:23] PR snap-confine#119: Make snap mount directory configurable [12:23] tyhicks: I didn't change the apparmor profile yet but I'm about to [12:28] Chipaca, not finding anything obvious, but I can't find it in CPI either, debugging more [12:29] ogra_: I can haz logs? what does it break? [12:30] the issue is this is hard to fix when I can't upload myself to that PPA and people leave pretty early [12:32] hi guys [12:33] i was looking for ubuntu disk image writer on snap [12:33] there are nor a lot of apps on this repo [12:33] cyphermox, you got a bunch of bugs ;) [12:34] Saeron, early days :-) [12:34] Saeron, it would be a nice one to snap up, i agree [12:34] can someone tell me if anyone can make a snap of this app [12:34] i men ihave to be the developper? [12:34] Saeron, yes, anyone can make a snap of that app [12:34] the problem is i am not actually using ubuntu [12:35] and i cant find this app on manjara thats what i want a snap for it [12:35] cyphermox, but in general i think having some tool that adds an additionyl 25MB RAM requirement is not really a good idea, is there no way to do it in ncurses or dialog instead of python3 ? it currently actually doubles our minimal ram reqs. [12:35] ogra_: there shouldn't be that much; we build images with nearly the same stuff (probably a few revs behind in livecd-rootfs, but that's it) in our PPA with no issues whatsoever [12:36] no, there isn't. We'll have to see about the RAM usage. I'm shocked that it's as much as you say [12:36] so exits the posibility of a good soul make a snap from this app? [12:36] cyphermox, on the dragonboard i get two console-conf binaries each occupying 25MB RSS ... i noted that in one of the bugs [12:36] Saeron, are you looking for the gtk version? [12:37] yes i am looking for the main ubuntu version [12:37] cyphermox, but thats totally not unusual for python on arm ... it is about 10x slower and eats a lot more ram thgan i.e. C or perl [12:37] Chipaca: yes [12:37] Saeron, i could snap it, but over the weekend, not now [12:38] Chipaca: its all right, is not at urgent need is just i like a lot dat app and i think is a good one [12:38] ogra_: where did you open the bugs? [12:38] Chipaca: thx, is gonna be awesome to have dat app on manjaro :) [12:38] :-) [12:38]