=== chihchun_afk is now known as chihchun [06:00] morning [07:17] mvo: hi, pushed a small shellcheck fix to #6331 hope you don't mind [07:17] PR #6331: systemd: allow only a single daemon-reload at the same time [07:19] mborzecki: thanks for this! [07:20] mborzecki: nice fix [07:21] mvo: shoould have gotten it right when i suggested the change :/ [07:21] mborzecki: no worries [07:21] PR snapd#6335 closed: tests: fix "No space left on device" issue on amazon-linux [07:28] great start of the year: http://seclists.org/oss-sec/2019/q1/54 [07:31] Hello [07:32] zyga: hey [07:37] PR snapd#6334 closed: dirs, interfaces/builtin/desktop: system fontconfig cache path is different on Fedora [07:47] PR snapd#6342 opened: use major() and minor() in instead of MAJOR and MINOR… === pstolowski|afk is now known as pstolowski [08:04] morning [08:07] Hey hey [08:07] I’ll be around shortly, just finishing coffee [08:32] PR snapd#6338 closed: cmd/snap: right-align revision and size in info's channel map [08:36] interesting thing with major/minor [08:37] mvo: the approval email arrived now (both of them :) [08:41] zyga: oh, let me see [08:42] zyga: got them now [08:44] thank you! [08:45] zyga: I approved the first one and will reject the other two [08:47] thank you === chihchun is now known as chihchun_afk [09:15] brb === chihchun_afk is now known as chihchun === sparkieg` is now known as sparkiegeek [09:25] mvo: hi, anything blocking #6331 ? [09:25] PR #6331: systemd: allow only a single daemon-reload at the same time [09:34] pedronis: not really, was just running some tests to see what a sensible number of runs is to trigger the bug (when I revert the fix). but thats not blocking it [09:35] PR snapd#6331 closed: systemd: allow only a single daemon-reload at the same time [09:35] PR snapd#6339 closed: cmd/snap: only auto-enable unicode to a tty [09:36] mvo: I was interested in the answer about test for 6339, I don't see John around yet tough [09:36] is he off? [09:36] pedronis: probably just starting a bit later, I don't think he is off [09:37] pedronis: I would also really like to poke a bit at the displayChannel revLen, sizeLen thing to see if there is a nice way to disentangle this a little bit [09:37] mvo: heh [09:37] mvo: doesn't sound like a task you should take on :) but you are the boss [09:37] pedronis: I know! OCD I guess [09:38] pedronis: I will need to learn a lot :) [09:38] PR snapd#6343 opened: overlord/configstate/configcore: support - and _ in cloud init field names [09:39] mborzecki: nice one - is this something we need for 2.37 ? [09:40] pedronis: how urgent is this change? ^^ [09:40] mborzecki: I think I missed the background for this one (sorry!) - where does the original report come from? [09:40] PR snapd#6344 opened: cmd/snap-confine: join freezer only after setting up user mount [09:41] mvo: it's from a mail thread [09:41] mvo: not sure there is one, there's a trello card though :) [09:41] mvo: ah, there you go [09:41] pedronis, mborzecki thanks! [09:41] mvo: I can forward you the thread [09:41] pedronis: how urgent? it looks like a nice drive-by for 2.37 [09:42] pedronis: I'm inclined to take it if we get reviews in time (happy to do one now) [09:42] mvo: about 6339, we lost tests indeed (at least afaict playing locally) [09:42] pedronis: meh, thats a bit sad [09:42] pedronis: so we need a followup [09:42] PR snapd#6345 opened: cmd/libsnap: pass --from-snap-confine when callingsnap-update-ns as user [09:43] mvo: is not urgent in the same that I don't think they can drop '-' for a long while, it would be good to have it in 18.04.02, which is supposed to happen Feb 7 ? [09:44] s/same/sense/ [09:44] pedronis: if 18.04.2 we better have it in, 2.38 is not going to be there in time [09:44] pedronis: or we could do a 2.37.1 for it but that seems wasteful if we can have it now. I do a review [09:44] mvo: mborzecki: anyway it needs a spread test [09:44] (now we should be able to have one) [09:49] PR snapd#6346 opened: osutil: add helper for loading fstab from string [09:50] mvo: I forwarded you the mail for context [09:51] PR snapd#6347 opened: many: allow snap-update-ns to write user mount profile [09:54] pedronis: thanks [10:01] Chipaca: hi, #6339 was merged but afaict we lost a bit of testing about the unicode formatting cases? it seems only CanUnicode tests consider the true case [10:01] PR #6339: cmd/snap: only auto-enable unicode to a tty [10:09] PR snapd#6348 opened: snap: split alignment calculation and display for channels [10:09] Chipaca: otoh I removed the blocked on #6280 [10:09] PR #6280: cmd/snap: make 'snap warnings' output yamlish [10:10] pedronis: ah, i was just replying to the first of those [10:10] pedronis: There's a test with --unicode=always, and there's a test that --unicode=auto does the right thing [10:11] mvo: not a fan of the structuring of #6348, still prefer what I proposed to John [10:11] PR #6348: snap: split alignment calculation and display for channels [10:13] mvo: i should have the spread test for #6343 up shortly [10:13] pedronis: maybe I misunderstood your original suggestion. in general I like to split data gathering from the display processing, I feel this makes things easier to read. but maybe that is not your concern? [10:13] PR #6343: overlord/configstate/configcore: support - and _ in cloud init field names <⛔ Blocked> [10:14] mvo: I proposed to encapsulate much more into a struct [10:14] mborzecki: oh, nice. if its shortly we can take that one too. I just reviewed 6343 [10:14] mvo: mhm, saw that, thanks! [10:14] mvo: I think John got what I proposed he can comment further, pick it up [10:16] pedronis: sounds good, reading your comment again I see what you have in mind (I think). that sounds even better. like I said, mostly my OCD to split it a bit more but more encapuslation would be even better of course [10:16] mvo: anyway as I said, not sure you should be working on this :) [10:16] pedronis: do you think a spread test for 6343 is required? I see you added blocked, is that because of the missing spread test? [10:17] mvo: yes [10:17] pedronis: yeah, I shouldn't [10:17] pedronis: ok [10:18] pedronis: I will start working on 2.37 now, if 6343 looks good, I can cherry pick it, we need a 2.37~pre1 in any case to see if autopkgtest is happy [10:19] mvo: anyway, it needs work [10:19] I mean #6343 [10:19] PR #6343: overlord/configstate/configcore: support - and _ in cloud init field names <⛔ Blocked> [10:19] now that I'm looking at it [10:25] mborzecki: some comments to 6343 [10:27] pedronis: thanks [10:27] I posted some simple branches for qucik review [10:27] PR snapd#6349 opened: cmd/snap-update-ns: move XDG code to dedicated file [10:27] as well as one branch I will work with jdstrand on [10:28] jdstrand: could you please review https://github.com/snapcore/snapd/pull/6347 -- it is a part of larger pull request https://github.com/snapcore/snapd/pull/6341 broken out for reviewabiliy [10:28] PR #6347: many: allow snap-update-ns to write user mount profile [10:28] PR #6341: cmd/snap-update-ns: persist per-user mount namespace profile [10:34] PR snapd#6350 opened: cmd/snap-confine: don't preemptively create .mnt files [10:44] pedronis: just to be clear about your coment, you're suggesting to use the fields in pairs, i.e. cloud_name is set, then we'd use availability_zone regardless of it being set or not (or the value of availability-zone), did i get that right? === chihchun is now known as chihchun_afk [10:57] mborzecki: yes [10:57] ack [11:09] pedronis: updated, pushed a spread test too [11:10] it seems that only ubuntu and amazon linux images have instance data set properly, others either do not use cloud-init or instance data is empty/unset [11:10] PR snapd#6351 opened: tests: review/fix the autopkgtest failures in disco [11:11] mborzecki: that's fine, as long as we have at least ubuntu [11:12] also, gce sets both - and _ names, while amazon only sets _ ones [11:12] interesting [11:12] maybe a diferent version of cloud-init [11:13] suspect so [11:13] but I thought they didn't drop yet producing both [11:13] anyway [11:13] that's why I was very keen of having the spread test [11:14] iirc it's set at boot time, i suspect that the engine hands an image that has just _ but cloud-init sets _ and - for compatibility with whatever might be runnning in the image [11:16] mborzecki: you are saying there's no cloud-init on the amazon linux one? [11:16] but the file is there? [11:16] pedronis: there is [11:16] anyway I need to go have lunch [11:16] I'll look again and we can finesse this then [11:17] pedronis: i've just pushed updates to hotplug-connect branch [11:20] mixed that up, amazon uses -, not _, gce does both [11:49] zyga: hey, can I book a minute to discuss https://github.com/snapcore/snapd/pull/6281#discussion_r241718332 ? [11:49] PR #6281: interfaces/builtin: add Multipass interfaces [11:49] hey! [11:49] looking [11:49] greyback: I was thinking about using the kmod backend [11:49] greyback: where snapd does the modprobe fore you [11:49] *for [11:50] zyga: oh, this is new to me. [11:50] let me show you an example [11:50] please [11:50] https://github.com/snapcore/snapd/blob/master/interfaces/builtin/firewall_control.go#L156 [11:51] that simple, huh? nice [11:51] you just declare the modules you want when an interface is connected [11:51] presto [11:51] yep [11:51] sweet, I'll use that [11:51] thanks! [11:51] great, thank you :) [12:11] whee [12:11] I reproduced this dreaded issue: [12:11] https://www.irccloud.com/pastebin/WAm6nIiD/ [12:13] https://www.irccloud.com/pastebin/HfXBv3XJ/ [12:13] indeed! [12:13] mvo: hey, can you take a look at #6113 when you have some time? [12:13] there's no /var/snap in core16! [12:13] hmmm [12:13] PR #6113: overlord/ifacestate: handler for hotplug-connect task [12:14] pstolowski: I put it on my mlist [12:15] zyga: in which core 16? [12:15] Chipaca: inspecting [12:15] pstolowski: thank, will look again at thos hotplug PRs once done with 2.37 stuff [12:15] we just get core16 from edge [12:15] ah, actual core16 [12:15] it's failing randomly all day for me [12:16] pedronis: sure, thanks [12:16] zyga: hm, core16 should have that, let me double check [12:16] zyga: hmm [12:16] mvo: I have a snap that does not [12:16] zyga: https://github.com/snapcore/snapd/pull/6350#issuecomment-453075903 ? [12:16] PR #6350: cmd/snap-confine: don't preemptively create .mnt files [12:16] ah, i see you already noticed [12:16] zyga: which version? [12:16] zyga: I mean which rev and version? [12:16] https://www.irccloud.com/pastebin/fA0Okwbn/ [12:16] is that expected? 2018 september is kind of old [12:17] zyga: looking [12:17] confirmed [12:17] i386 core16 on edge has no /var/snap [12:17] zyga: I triggered a new build from edge [12:17] http://paste.ubuntu.com/p/yyq8FFKPnY/ [12:17] mvo: thank you [12:17] how come this ever passed? [12:18] this is the list of files in that snap [12:18] amd64 is fine tho [12:19] zyga, Chipaca the versions are all a bit inconsistent, I triggered a new build with that they should be all in sync again [12:19] thank you [12:19] was it built recently? [12:19] revision is kind of low [12:19] but perhaps/ [12:19] I'm trying to understand why it fails [12:20] mborzecki: +1 for #6343 with a comment about the spread test tough [12:20] PR #6343: overlord/configstate/configcore: support - and _ in cloud init field names <⛔ Blocked> [12:20] zyga: I looked at this earlier today as it failed in autopkgtest on s390x, I think while trying to make sure all versions are vaguely correct I promoted the wrong one [12:20] zyga: but looking at the versions there is no right one right now :( [12:21] aha so you were changing the published revision? [12:21] I wish snapcraft had publishing history like git [12:22] zyga: left a litle comment under #6349 [12:22] PR #6349: cmd/snap-update-ns: move XDG code to dedicated file [12:23] mvo: I don't understand why it only sometimes failed [12:23] zyga: mvo: pedronis: https://pastebin.ubuntu.com/p/Z2BJDqjVRS/ [12:24] Chipaca: :) [12:24] those all seem very recent [12:24] Chipaca: I like the idea, can nitpick on the syntax [12:24] Chipaca: nice [12:24] Chipaca: is it still valid yaml? [12:24] Chipaca: how about channels:\n\t\n\t\tlatest/stable: ... [12:24] this was a quick hack just to see what was released where :-) [12:25] mborzecki: probably :-) [12:25] zyga: yea, I expect some back and both on the options and the output [12:25] *back and forth [12:25] mborzecki: http://yaml-online-parser.appspot.com/ thinks it is [12:26] zyga: mvo: strangely enough the released-at is all today, but the version numbers are all over the place [12:28] Chipaca: yeah, I tried to publish a consistent set today [12:28] Chipaca: which did not work well, so the released-at is correct [12:28] Chipaca: and its such a nice feature! [12:28] heh [12:29] you can thank sabdfl for that one :-) [12:29] * mvo thanks him [12:29] * Chipaca thinks you can probably thank him for the whole thing, while you're at it [12:29] Chipaca: which reminds me, did you update the bug to fix-commited yet :) ? [12:29] * mvo thanks him some more [12:29] he did [12:29] I did :-) [12:34] mborzecki: does my comment on the spread test make sense? [12:39] pedronis: wonder how much it depends on the cloud provider, nonetheless, the file is there in 16.04+ instances, so ensuring the test runs actually somewhere sgtm [12:46] mborzecki: yea, but the cloud provider we nail with backend check [12:46] s/backend check/the backend constraint/ [12:46] pedronis: it's pinned to backends: [google] atm [12:47] yes, that's what I mean [12:47] so my hope is that google x ubuntu (classic) should work [12:47] at a minimum [12:48] pedronis: instance data is also set in core-16 images too, only 14.04 is missing out [12:48] mborzecki: ok, should also be on core18 [12:48] but anyway [12:48] let's indeed do all ubuntus where it seems to work [12:49] mborzecki: also the red tests is because of a typo apparently [13:02] mborzecki: updated https://github.com/snapcore/snapd/pull/6349 :) [13:02] PR #6349: cmd/snap-update-ns: move XDG code to dedicated file [13:02] zyga: thanks, will take a look [13:03] off to pick up the kids [13:07] zyga: yes, I plan to go through that and the others I've been asked to look at today/tomorrow [13:07] thank you [13:10] PR snapcraft#2434 closed: rust plugin: refactor to use the latest rustup [13:17] PR snapd#6352 opened: many: remove .user-fstab files from /run/snapd/ns [13:20] pedronis: would a spread test be enough to cover the rererefresh case? [13:20] :-| [13:20] Chipaca: we need to talk, I'm in a meeting [13:20] pedronis: ok [13:20] i'll go have lunch === ricab is now known as ricab|lunch [13:22] PR snapd#6353 opened: cmd/snap-update-ns: move existing code around, renaming some functions [13:23] I pushed one long but simple branch that just splits bigger file into focused chunks [13:50] hm, under spread, jq is available in core-18 images but not in core-16 [13:51] or am i missing something there? [13:51] mborzecki: it's installed as a snap under core afaik [13:51] you can find the installation in some prepare code I think [13:51] maybe the code checks core but not core 18 atm [13:51] or the reverse [14:04] and now travis seems to be slow to start spread jobs https://travis-ci.org/snapcore/snapd/builds/477815742?utm_source=github_status&utm_medium=notification heh [14:05] mborzecki: yeah, quite a bit of delay in spread right now :/ [14:06] (well travis to be fair) [14:15] zyga: https://forum.snapcraft.io/t/tried-to-install-quelea-presentation-software-from-chrome-linux-beta/9247?u=chipaca dunno if you're the right person to answer here but ogra seems to think so :) [14:16] Chipaca, well, sorry ... zyga already told me he has no idea about chromeos support ... i didnt forward that to the forum post [14:17] but yeah, zyga being the "foreign distro ambassador" in the team, i thought he might know :) [14:17] I don't have a chromebook anymore [14:17] and AFAIK only some recent chromebooks can run lxd [14:17] and thus run snaps [14:17] but I have 0 knowledge about how that works [14:18] or what's the nesting story [14:18] even then i bet it is doubtful becase of lacking kernel features [14:20] dunno :) [14:21] jdstrand: replied on the permission topic [14:21] jdstrand: it's really about go vs c logic [14:21] jdstrand: one thing we might do is to open the fd while still in C [14:21] jdstrand: but I don't really think that warrants the extra logic today [14:22] zyga: even replying "I don't know and this is why I don't know" would be helpful [14:24] mvo: so based on what you said, can we merge https://github.com/snapcore/snapd/pull/6294 now? [14:24] PR #6294: packaging/ubuntu: build with golang 1.10 [14:26] zyga: yeah, maybe early next week but in theory we could do so *now* [14:27] zyga: missing on that PR: move the travis unit tests to run with 1.9 instead of 1.6 [14:27] zyga: I can push that change [14:28] Chipaca: oooh, indeed [14:28] Chipaca: if you can please do [14:36] can't wait to get rid of the 1.6 gofmt wrappers that I have now [14:38] mborzecki: next you'll be wanting generics or whatever they're getting called [14:38] haha :P [14:39] Chipaca: wouldn't mind pattern matching in go 2 tho [14:39] are we getting pattern matching as well? i missed that bit [14:39] * cachio lunch [14:40] I think the improvements to error wrapping are probably the most interesting bit for us [14:40] Chipaca: afaik no [14:40] pedronis: we could use github.com/pkg/errors for this now [14:40] mborzecki: it's not about our errors usually [14:41] more about the std lib ones [14:41] especially networking ones [14:41] mhm [14:41] mborzecki: I'm talking about the contorsions our retry code does for example [14:41] btw. did they sort out the context for io interfaces? [14:47] mvo: about jq https://paste.ubuntu.com/p/YGPvF65wFJ/ [14:47] mborzecki: hm, interessting === ricab|lunch is now known as ricab [14:49] mvo: changes: https://paste.ubuntu.com/p/GphB7jj8Gd/ up to change 8 it's all prepare, change 9 is caused by prepare section of the cloud-init test [14:49] mvo, mborzecki we install jq in prepare [14:49] zyga: but it's removed later on [14:49] mborzecki: grep for jq in the tests/lib section [14:50] disable_refreshes needs it [14:50] hmm [14:50] i see now [14:50] zyga: right, but it's removed just below [14:53] indeed [14:53] no idea now [14:55] mborzecki: mvo: fwiw I just double checked, there is no jq in core18 itself [14:55] at least not in stable [14:56] pedronis: not in edge either [14:56] waiting for vanilla core18/16 instances to boot and will check there [15:07] hm no clue [15:07] PR snapcraft#2435 opened: Appstream desktop [15:13]