=== AmarokNelg_ is now known as AmarokNelg [06:06] morning [07:26] good morning! :) [07:27] * zyga looks at a mountain of email [07:30] zyga: hey [07:31] hey :) [07:31] happy new year [07:32] likewise :) [07:33] zyga: did you enjoy your break? [07:35] mborzecki: yes, it almost feels weird to go back after so long [07:35] i would call it a shock, rather than just being weird :) [07:46] good morning [07:46] happy 2018 [07:52] good morning kalikiana! [08:10] mornings! [08:11] pstolowski: hey [08:19] good morning koza [08:31] * zyga reboots for some updates [08:44] mvo: hey [08:44] Issue snapcraft#1837 closed: Can't run snapd on Codenvy 5.22.0 [08:47] hey mborzecki! how are you? [08:47] mborzecki: did I miss anything interessting this morning :) ? [08:47] (unsurprisingly) the morning has been rather slow [08:48] i'm poking around yocto support, slow progress however [08:48] * mvo nods [08:48] mborzecki: yocto> nice to hear [08:49] promised this guy https://forum.snapcraft.io/t/yocto-rocko-core-snap-kernel-panic/3261/11 to look into it once i get back :) [08:49] * mvo nods [08:50] Issue snapcraft#1834 closed: Feature Request: NGINX plugin [09:03] PR snapd#4427 opened: snap: fix snap find " " output [09:22] zyga not around? [09:22] mborzecki: heh [09:24] hey mvo :) long time no see [09:24] how are you doing? [09:29] zyga: https://github.com/snapcore/snapd/blob/master/cmd/snap-confine/mount-support.c#L589 afaiu /root is assumed to always exist then? [09:29] mborzecki: yes, there's a lot of directories we just assume [09:29] mborzecki: for /root we could skip it if absent [09:30] i'm asking as it's not here in poky (yocto build), at least with rocko brach (the last release) [09:30] hey zyga ! good to see you, happy new year [09:30] frankly, i don't recall if it used to be there or now, never needed to do/keep anything around in /root [09:34] ▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄ ANOTHER EXPLOIT HAS BEEN FOUND IN WEECHAT ONE THAT SENDS YOUR LOGS TO THE DEVS IP ADDRESS!!jmnaxjlu: ubot9 hurricanehrndz verterok ▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄ [09:36] * kalikiana coffee break === __chip__ is now known as Chipaca [09:44] hey Chipaca [09:44] zyga! [09:44] how's things? [09:44] Chipaca: snappy :) [09:45] Chipaca: still waking up/getting into coding spirit (code reviews now) [09:45] how are you doing? are you feeling better? [09:45] hey Chipaca! welcome back, great to "see" you [09:45] mvo: o/ [09:45] zyga: i caught a lot of sun, and a little head cold, but overall feeling good. you? [09:46] Chipaca: I'm doing great, disappointed for not having much snow this xmas [09:47] mvo: likewise, how're you doing? [09:47] PR snapd#4428 opened: daemon: return "bad-query" error kind for store.ErrBadQuery [09:47] zyga: snow is nice but i'm happy with my choice :-) [09:48] Chipaca: did you travel somewhere warmer for your holidays/ [09:48] Chipaca: in general great, today in particular not so great, caught a stomach bug yesterday and still feel weak from that. but hacking beats lying in bed and being bored :) [09:49] mvo: i'll preted not to know half of your hacking today is from an unconventional seat [09:49] lol [09:50] zyga: visiting friends on the costa d'azur [09:50] Chipaca: stdout flushes manually, stderr flushes on output ;-) [09:50] * zyga stops thinking about IO jokes [09:50] Chipaca: ah, lovely [09:51] Chipaca: I've visited some friends from Catalonia but we chose to go to the capital of coal in poland (katowice) instead :) [09:51] oooh [09:51] i've only run trains through the places you visit [09:54] mvo: is the client side of the bad query kind coming, or not needed? [09:58] Chipaca: I can add it, I think it makes sense, I was mostly doing it for roberts glib based client implementation [09:59] (where it is not needed AIUI) - but for symmetry it does make sense [09:59] mvo: if we're not going to use it, it's just bitrot waiting to happen [10:00] mvo: (otoh all the kinds could use a reshuffle to not be defined twice...) [10:00] dunno [10:00] * Chipaca does not bring new wisdom into the new year [10:00] Chipaca: lol [10:00] Chipaca: all good, I will push the client side and make a nice(r) error message for `snap find "()!§$§%§%*"` [10:01] Chipaca: something like "error: cannot list snaps: bad query, please use less special chars" [10:01] mvo: "please use latin letters or digits" maybe [10:02] oh you just reminded me i probably need to change this error message before submitting it as a PR [10:02] zyga: yeah, thats better [10:02] * zyga looks at the intense fog outside [10:02] ErrAppEscapes = errors.New("snap is unusable due to fuckery") [10:02] Chipaca: *cough* this is a family channel! [10:02] mvo: or arabic digits ;-) [10:03] mvo: /o\ sorry [10:03] zyga: maybe ascii letters/digests [10:03] * mvo hugs Chipaca [10:03] Chipaca: go and stand in the corner now ;;-) [10:03] mvo: yeah [10:03] * Chipaca hugs mvo from the corner [10:04] we have a lot of PRs [10:04] zyga: my fault [10:04] mvo: how is that a fault? :) [10:04] mvo: thank you for all the nice fixes [10:04] zyga: I was doing bug triage and there is soooo many low hanging fruits [10:04] can I get a review on https://github.com/snapcore/snapd/pull/4315 [10:04] PR #4315: cmd/snap-update-ns: add execWritableMimic [10:05] it's from _november_ and has one review [10:05] * zyga boots subsequent VM for the round of updates [10:06] how's travis/linode/spread behaving? [10:06] I haven't tried anything new yet, I see lots of red PRs though [10:06] so maybe it's still having a hangover [10:09] zyga: one of the things we realised towards the year end was that the interfaces-many was taking a relatively long time to run [10:10] as in 10 minutes [10:10] that's pretty slow, is it CPU bound on apparmor compiler? [10:10] zyga: probably not [10:11] zyga: because dropping the unlock/lock around things dropped the time down a lot [10:11] iirc down to 4 minutes? somehting like that [10:11] Chipaca: oh? that's curiou [10:11] Chipaca: around what specifically? [10:12] zyga: around … and this is from memory … the security backend setup loop? something like that [10:13] zyga: basically it means moving to an actual database for state should not be postponed much longer [10:13] Chipaca: and that locking serializes state and this is expensive? [10:13] zyga: yep [10:14] zyga: when done in a loop like in that test, yes [10:14] keep in mind because of what that test does, it's almost a worst case [10:14] lots of changes to serialize [10:15] Chipaca: lots as in "KBs of text" [10:15] Chipaca: I think the real DB will take some time to arrive [10:15] Chipaca: 'actual database' do you have anything particular in mind? [10:15] Chipaca: it's a major change [10:16] zyga: yes i know [10:16] I told the same, given everything it's a post 18.04 afai would guess [10:16] mborzecki: gustavo had thoughts on this, there was somehting ike an in-memory all-go db [10:16] Chipaca: is it possible that the cost is in fsync ops and not in anything else? [10:16] pedronis: hello :-) [10:16] BoltDB I think [10:16] pedronis: just trying to make sure this isn't lost [10:16] is what Gustavo mentioned [10:17] zyga: go's profiler pointed to json encode [10:17] Chipaca: hello, me is slowly rebooting :) [10:17] mhm, i'm asking cause we moved to lmdb at some point in mender [10:17] zyga: but it _might_ be io disguised as that [10:49] PR snapd#4429 opened: tests: add regression test for LP 1681739 (snap var interpolation) [10:49] I still have one more week of vacation :) [10:50] PR snapd#4405 closed: taskrunner/many: KnownTaskKinds helper [10:51] PR snapd#4428 closed: daemon: return "bad-query" error kind for store.ErrBadQuery [10:55] PR snapd#4420 closed: cmd: clarify "This leaves %s tracking %s." message [10:57] PR snapd#4403 closed: asserts/signtool: support for building tools on top that fill-in/compute some headers [10:57] PR snapd#4422 closed: packaging/arch: disable services when removing [11:02] Son_Goku: showoff [11:02] eh, I was sick most of my vacation [11:02] this might be my only "enjoyment" week [11:04] Son_Goku: I think everywhere I've lived, if you're sick during a vacation it counts as sick leave and not vacation [11:04] I filed it as vacation, then became sick after [11:04] at that point, it doesn't matter [11:05] I got sick during my trip to my parents' home [11:05] Son_Goku: i understand you filed it as vacation, but if you get sick during your vacation, you can then claim it back (and take your vacation when you're not sick) [11:05] this is, in argentina and the eu it works [11:05] i've not lived elsewhere [11:05] it's true here too [11:06] good to know [11:06] but as I have "unlimited PTO" and only 5 sick days, there's no point [11:06] Son_Goku: i understand if it's academic because of … yeah, those limits [11:06] Son_Goku: especially if sick days are not paid [11:06] sorry, five paid sick days [11:06] Son_Goku: and is the unlimited PTO also paid? [11:07] yes [11:07] if I take more than two weeks of sick leave, I'd probably be fired [11:07] ah fair enough :-) [11:07] not without some kind of doctor's notice, anyway [11:07] ah yeah the "claim it back" thing involves a doctors scrip [11:07] strip? [11:08] bit of paper [11:08] * Chipaca has no idea where he picked up that word [11:12] zyga: quick question about lp: 1708703 [11:12] Chipaca: strips of paper exist. although you might want to take the whole one with you to show to your employer rather than tearing of part of it ;-) [11:12] heh [11:13] kalikiana: pretty sure it's 'scrip' [11:13] kalikiana: but as i say, dunno where i picked it up [11:13] my english has [*] reference needed all over the place [11:13] citation* [11:14] scrip is technically correct [11:14] though I've not heard that word in common use [11:14] it's short for prescription [11:15] the sources of my english include, but are not limited to, north scottish fishermen, south american british hospital staff, and welsh [censored] [11:15] ¯\_(ツ)_/¯ family [11:17] interestingly enough in that use is a contraction of script itself there a contraction of prescription [11:17] * pedronis (is perusing a dictionary) [11:19] mvo: looking [11:20] zyga: mostly curious if you know more about it as you assigned it to yourself [11:20] mvo: yes, I remember now [11:22] zyga: tell me more please [11:22] mvo: looking at the code, when we enable / disable we setup security [11:22] mvo: then we look at snaps that were affected and also setup those [11:23] mvo: then that can fail "silently" when we cannot get snap state for any of those snaps [11:23] mvo: now I'm not sure if this is happening here [11:23] mvo: let me try [11:23] mvo: it looks ok but ... [11:25] zyga: I tried it today and can still reproduce it (with the mir snaps) [11:25] mvo: did you try that in a VM with core snap? [11:25] mvo: or on classic? [11:26] classic [11:26] mvo: aha, let me see master quickly now, just adding some logging [11:27] ok [11:29] mvo: logging in tests says it's not broken, may be broken later, trying with real snaps now [11:30] zyga: ta [11:31] mvo: hmm, mir-kiosk is not in the store? [11:31] ah, edge [11:32] yes [11:33] ha [11:33] indeed [11:33] the reason is simple, autoconnect [11:34] for some reason i thought we were interpolating variables in app commands [11:34] mvo: when the mir-kiosk-apps is disabled it gets disconnected [11:34] turns out we aren't [11:35] weren't we going to do that? [11:35] mvo: then when it is enabled it thinks that nothing happened because of autoconnect [11:35] * zyga looks [11:37] zyga: iirc it was still showing as connected when I tried [11:38] mvo: yes, there's something buggy going on [11:38] mvo: it should say it got disconnected but didn't [11:42] * mvo nods [11:42] mvo: oddly repo.DisconnectSnap returns [] so something is very fishy there [11:50] mvo: so on enable r.plugs[] and r.slots for that snap is empty [11:50] mvo: so something is wrong in the repo, digging deeper [11:50] ok [11:50] thanks [11:50] pstolowski: ^ FYI [11:53] zyga, is this re 1708703? [11:53] yes [11:53] I see the bug now [11:55] ok, for a moment I thought I broke something in recent refactorings, but I guess that's an older issue [11:55] yes, it looks like an older issue [11:55] interesting how it survived all the changes :/ [12:01] fixed :) [12:02] now for some tests [12:04] zyga, thank you [12:15] * pstolowski lunch [12:16] PR snapd#4430 opened: overlord/ifacestate: fix disable/enable cycle to setup security [12:16] pstolowski, mvo: ^ [12:21] * mvo hugs zyga [12:30] zyga, when are you going to give me snapd selinux backend? :) [12:30] Son_Goku: no idea, but if I start now maybe in December? [12:30] better than never, so worksforme [12:30] Son_Goku: I haven't discussed what our todo list is yet [12:51] morning [12:54] hey sergiusens :) [13:06] morning sergiusens [13:06] and happy new year [13:08] PR snapd#4429 closed: tests: add regression test for LP 1681739 (snap var interpolation) [13:19] * kalikiana going to take a lunch break in ~15 [13:29] mborzecki: in https://paste.ubuntu.com/26306252/, what is the 1st column? [13:30] Chipaca: just an index in the data set [13:30] PR snapcraft#1838 opened: tests: fix broken rust test snap [13:32] Chipaca: same dataaset, ubuntu-16.04-64 only this time: https://paste.ubuntu.com/26306338/ [13:32] mborzecki: hmmm [13:33] mborzecki: what does the duration of the suite mean? [13:33] mborzecki: because the suite takes 58s and in the suite a bunch of tests take 50s, it sounds like an accouting error [13:33] oh wait [13:33] Chipaca: is it measuring wall time or accumulated time from each node? [13:33] Chipaca: it's taken from travis folds [13:33] mborzecki: my mistake sorry [13:34] tests/main/completion vs tests/completion :-) [13:35] meh, it's getting dark already [13:35] my monitors are too bright [13:35] and have poor controls for adjusting [13:36] zyga: redshift? [13:36] Chipaca: i think it's built into gnome now [13:36] * zyga looks for that [13:36] zyga: yes (but don't you need to enable it?) [13:37] Chipaca: I'm also complaining about short daylight cycle here :( software cannot fix that [13:37] zyga: i'm pretending not to notice [13:37] enabled :) [13:38] thank you for reminding me about it [13:38] * kalikiana bbl [13:39] Chipaca: on the other hand I moved my systems to a darker green tinted wallpapers/color schemes and I like that a lot [13:40] man, apt needs a 'find' alias now :-( [13:41] zyga: about teleconsole on fedora, `snap run --shell teleconsole` should not fail with 'execv failed: No such file or directory', right? [13:42] mborzecki: right [13:42] mborzecki: it looks like it's trying to exec /usr/lib/snapd/snap-exec [13:42] vs libexec [13:42] (just a guess) [13:42] I haven't rebooted F27 after updates [13:42] 47% installing [13:42] but it works for 'hello' snap [13:43] classic doesn't do redirection [13:43] I think I know what's going on [13:43] we disabled classic on fedora [13:43] so you did the symlink [13:43] but the code that runs classic stuff had hardcoded assumptions on dir layout [13:43] if you print execed programs you will see (probably) it running the wrong directory [13:44] Chipaca: #4389 and #4392 are probably PRs you could review [13:44] PR #4389: overlord/snapstate: override Snapstate.UserID in refresh if the installing user is gone [13:44] PR #4392: many: refresh with appropriate creds [13:44] mborzecki: as a quick test add a symlink /usr/lib/snapd -> /usr/libexec/snapd === __chip__ is now known as Chipaca [13:46] zyga: right, works now [13:46] mborzecki: cool, that should be as simple fix [13:46] *a [13:51] * zyga removes 34,028 messages from launchpad from his inbox [13:53] zyga: hm we can't exactly mess with rootfs and fedora guys would probably be unhappy if we symlink, the fix would need to be automatically cleaned up when snap exits [13:55] no no, the symlink is just a test [13:55] we don't need it [13:55] but we now know exactly what is wrong [13:55] it's a simple bug in cmd.go [13:55] or cmd_run.go [13:58] zyga: https://paste.ubuntu.com/26306484/ like this? :) [13:58] yes [13:58] looks excellent [13:58] thank you! [13:58] mborzecki: look at how this behaves when classic + reexec is in use [13:59] mborzecki: but this looks ok for fedora [14:04] zyga: https://paste.ubuntu.com/26306521/ [14:05] and teleconsole works now, yay [14:05] I think the test is wrong [14:05] if you are reexeed you want: [14:05] path to core snap in the distro + core libexec dir [14:06] e.g. /var/lib/snapd/snap/core/current/usr/lib/snapd/snap-exec [14:06] which combines a fedora-specific snap mountdir + core layout that contains snap-exec [14:06] you can perhaps simplify [14:06] hmm right [14:06] by using the same trick that cmd.go does [14:06] find your own executable [14:06] and then run snap-exec as your peer [14:07] though because snap is not in the same directory as snap-exec or snap-confine, this may be more cloudy [14:07] I'd vote for a direct test + explicit join [14:08] mhm [14:08] looks like we have a way forward, i'll open a PR later [14:09] thank you :) [14:09] been fun chasing it [14:09] still now idea why strace hangs occasionally [14:10] anyways, i'm off to pick up the kids [14:19] so, the completion tests do take 50s per [14:20] they can be made significantly faster by dropping a delay and/or by bumping a send rate [14:20] these two things exist because without them the tests were flaky [14:20] should I try dropping them to see if they continue being flaky? [14:29] Chipaca: worth a try, I wonder if we could do _something_ to become reliable [14:32] re [14:33] PR snapd#4421 closed: daemon: add new polkit action to manage interfaces [14:34] PR snapd#4427 closed: snap: fix snap find " " output [14:46] * zyga break for small taxes and paperwork [14:47] good morning team! [14:58] kyrofa: hi! I played with the nextcloud snap a bit over the holiday. I was wondering if you thought about a content interface (readonly?) for it so that snaps on the system could access the files. eg, a minidlna snap could consume nextcloud content for serving movies, pictures and/or videos [14:59] kyrofa: and happy new year :) [15:02] elopio good morning [15:04] Happy New Year everyone! [15:04] Morning jdstrand :) [15:05] kyrofa hey [15:05] jdstrand, I considered it, but it's possible for someone to use the removable-media interface for data, and then the snap isn't sharing anything useful [15:06] Hey there sergiusens, nice break? [15:08] kyrofa: with the nextcloud box, it stores its data not in /media by default since it formats the drive and installs itself on it [15:08] Saviq: is there an easy way to make a multipass disk bigger, 2.1GB seems awfully small for the initial size [15:08] jdstrand, yeah, by default it uses $SNAP_COMMON [15:09] popey: on launch, pass `--disk 5G` or so [15:09] kyrofa: I was thinking of creating a systemd unit to bind mount stuff into /media and use removable-media for a minidlna snap, but thought maybe things could be easier for people [15:09] Saviq: but not after I made it? [15:09] popey: 2G is what the default size is of the cloud images [15:09] kyrofa: anyway, food for thought [15:09] popey: I think someone wrote on the forums how to do it after launch [15:09] it's not trivial since you have to resize the FS [15:10] found it, thanks [15:10] so I'm not sure we'll ever built that functionality into multipass (but we'll allow you to change the default initial size) [15:10] I'll destroy and start again :( [15:11] kyrofa: also, is the nextcloud-client expected to be fully functional? (excepting auto-start) [15:11] popey: we might bump the default anyway, it's qcow so it's not stealing your storage straight away [15:12] jdstrand, indeed, although I'll admit I don't use it due to the autostart thing so I wouldn't be surprised to learn that it has issues. Does it? [15:12] kyrofa: (nextcloud-client snap that is. I kept ending up with defunct processes and things not syncing (not 100% sure things not syncing was the snap's fault and not mine cause it was still early days in my playing) [15:12] ) [15:13] Hmm, that sounds odd [15:14] jdstrand, kyrofa: hey :) [15:14] happy new year zyga :) [15:14] Happy new year, zyga! How was break? [15:14] kyrofa: exhausting :) too long I would say [15:14] Hahaha [15:14] kyrofa: I'm happy to be back :) [15:15] how are you guys doing? I heard US has move to ice age [15:16] *moved [15:16] kyrofa good, yourself? [15:17] zyga I want to be on a break for longer! [15:17] Yeah me too [15:17] sergiusens, short, but good [15:17] sergiusens: for how long were you away? [15:19] zyga just the week in between the two big holidays [15:22] ogra_: i think some files were moved inside os.snap, without updating the kernel plugin [15:22] ogra_: can you give your opinion on this? [15:22] ogra_: https://bugs.launchpad.net/snapcraft/+bug/1740882 [15:22] Bug #1740882: Missing initrd.img-core in os.snap