[00:01] Man... I've spent all day commenting out tests trying to find which one is actually interfering with another. I need a drink [00:12] built: solus-runtime-gaming_0.0.0_all.snap [00:12] heh [00:14] kyrofa: https://ovh.themcphails.uk/index.php/s/LbKIquTSxMernmM [00:15] mcphail, you're too kind [00:15] ikey: so is this the "core" snap side of things? [00:15] right [00:15] just gonna be cheap and dirty and use 'snap pack' [00:16] but i think im gonna need to force some stuff like 'architectures' field, etc [00:17] built: solus-runtime-gaming_0.0.0_amd64.snap [00:17] bit better :p [00:18] This is very exciting [00:18] certainly a learning experience for me [00:18] i imagine the initial 0.0.0 is gonna be nasty a.f. [00:18] but then i can start optimising it and cleaning it up [00:19] and allow us to do builds with local overlay layers onto the image [00:19] i.e. so we can stage mesa properly, etc [00:19] and part of this is gonna be a cleanup process to unjank the image [00:19] as long as i got `snap pack` command now i'm golden [00:20] yes, I like that way of doing things [00:20] i strongly suspect we're gonna just rename linux-steam-integration to `LSI` and add an `lsi-exec` command, because I can think of a few games where this will help even outside steam.. [00:20] but one thing at a time :D [00:21] PR snapd#4184 opened: tests: adding new test for uhid interface [00:22] What does `snap pack` do that `snapcraft snap ` doesn't? [00:23] doesn't use dpkg/apt/etc [00:23] :P [00:23] i'm creating a base snap [00:23] i.e. not ubuntu [00:23] snapcraft is bricked for solus [00:23] because its basically ubuntu specific last we looked [00:28] ikey, yeah `snapcraft snap ` just runs `mksquashfs` with all the magical flags. It doesn't actually do anything with the dir [00:28] Although actually getting snapcraft ON there might be hard, unless classic snaps work [00:28] classic snaps work [00:28] classic "i think im ubuntu" not so much [00:28] this saves me a bit of work just having the command directly there [00:29] and yeah i saw its basically cleanup + meta + mksquashfs [00:29] basically im just tryna be hella lazy === jero is now known as Guest81164 === mup_ is now known as mup === mup_ is now known as mup === wgrant_ is now known as wgrant [05:59] morning everyone [06:02] o/ [06:04] hmm [06:04] - linode:ubuntu-16.04-32:tests/main/xdg-open-compat [06:04] - linode:ubuntu-16.04-64:tests/main/xdg-open-compat [06:04] I'm seeing those fail in 2.29 branches [06:26] ... [06:39] zyga-ubuntu: hey [06:40] ho [06:40] eating breakfast [06:40] PR snapd#4185 opened: interfaces/builtin/account_control: drop group filter from seccomp rules [06:40] neetd to look into those failing tests [06:41] yeah, noticed that sergiusens spread tests were failing because of those [06:41] commented [06:41] look at 4185 [06:42] ha, good catch, do you know if I could feed this rule to snap-seccomp locally to see if it compiles? [06:43] yes [06:43] I think it will compile [06:43] it just needs to run to see what happens [06:43] compile vs run [06:44] hm there's tests/main/interfaces-account-control i'll run it to see what will happen [06:45] ok [06:45] morning mvo [06:45] hey zyga-ubuntu [06:45] mvo: we have a problem with one spread test, it fails very very often across many PRs (including 2.29) [06:45] tests/main/xdg-open-compat [06:47] zyga-ubuntu: ok, I have a look. do you happen to have the error at hand? [06:47] no, sorry, running it now to see [06:50] mvo: https://s3.amazonaws.com/archive.travis-ci.org/jobs/299323286/log.txt there's some error log here [06:52] mborzecki: hm, Access Denied - anyway I will find one if its common :) [06:52] 49 PRs, woah [06:52] I think it's review sprint time [06:52] mvo: many are approved and need 2nd review [06:53] https://s3.amazonaws.com/archive.travis-ci.org/jobs/299180299/log.txt?X-Amz-Expires=30&X-Amz-Date=20171109T065333Z&X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAJRYRXRSVGNKPKO5A/20171109/us-east-1/s3/aws4_request&X-Amz-SignedHeaders=host&X-Amz-Signature=b2c9efa5b8cf4fe1b35132841d988eeaa70be47b26b4d84a68d849221c1b553d [06:53] mvo: ^ failure [06:54] + apt download snapd-xdg-open=0.0.0~16.04 [06:54] WARNING: apt does not have a stable CLI interface. Use with caution in scripts. [06:54] E: Version '0.0.0~16.04' for 'snapd-xdg-open' was not found [06:54] that's it [06:54] aha [06:54] seems simple [06:54] zyga-ubuntu: aha, I think that happens because snapd 2.28.5 is now in xenail-updates and we have the older version not available anymore [06:54] zyga-ubuntu: it should fail consistently though [06:55] https://packages.ubuntu.com/search?keywords=snapd-xdg-open&searchon=names&suite=all§ion=all [06:55] yes [06:55] magic, right? [06:55] maybe the non-xenial-updates version is cached on some images? [06:55] (just pulling ideas out ..) [06:55] zyga-ubuntu: thats how mborzecki name is pronounced ;) [06:55] zyga-ubuntu: a good question, or maybe not all mirrors are up-to-date or something [06:55] haha :) [06:55] zyga-ubuntu: definitely strange [06:55] close enough :) [06:56] mborzecki: I will learn it eventually :) [06:56] m-a-ch-e-y === chihchun_afk is now known as chihchun [06:56] mvo: curiously I didn't hit this locally yesterday, after running spread extensively [06:56] let's see if today's any different [06:57] offtopic: I found this very funny to read: http://www.sciencemag.org/careers/2016/01/how-read-scientific-paper :) [06:57] while I make some tea [06:57] zyga-ubuntu: yeah, but we need to fix it for 2.29 or autopkgtest in the distro will be unahppy :( [06:59] yes [06:59] ha, failed now [07:00] so maybe old package was garbage collected [07:00] zyga-ubuntu: heh, this how-to-read- article reminds me of https://www.youtube.com/watch?v=ILysMhm8lXs which is about taxes but otherwise similar [07:00] zyga-ubuntu: yes [07:00] only 2.28.5 is in the archive now [07:00] * zyga-ubuntu tries a simple fix [07:00] zyga-ubuntu: we need to thing what to do, maybe the only thing we can do is to kill the test :( [07:00] zyga-ubuntu: aha, cool. simple++ [07:01] running now [07:01] * zyga-ubuntu looks at tea [07:02] is there a way to mark it as expected failure (provided it fails consistently), like pytest.xfail or sth? [07:02] nope [07:14] mborzecki: sorry about my last answer, it was a mental shortcut [07:15] hm? [07:15] mborzecki: there's no direct way, we can echo "blah blah"; exit 0, or switch a test to manual: true to skip it [07:15] mborzecki: we can also patch spread, it's github.com/snapcore/spread after all [07:15] mborzecki: it's just that we traditinally tried to fix tests rather than improve the infra [07:16] zyga-ubuntu: I wonder why we did not notice 4182 ourselfs [07:16] mvo: it's curious [07:17] mvo: I ran snap-confine counltess times recently [07:17] zyga-ubuntu: does your fix for the download thing works? [07:17] and it did work [07:17] mvo: I made two tweaks to the policy that I haven't upstreamed yet but they are not for os-release [07:17] zyga-ubuntu: yeah, looking at the code we should check for os-release often and yet [07:17] maybe the real issue is more subtle and we're missing something? [07:18] # Allow reading the os-release file (possibly a symlink to /usr/lib). [07:18] /{etc/,usr/lib/}os-release r, [07:18] this is in the profile, but for snap-update-ns [07:18] not for snap-confine [07:18] PR snapd#4181 closed: interfaces/many: misc policy updates for browser-support, cups-control and network-status (2.29) [07:19] i'm trying to run linode:ubuntu-core-16-64:tests/main/interfaces-account-control and got `cp: error writing '/mnt/user-data/gopath/bin/fakestore': No space left on device` :/ [07:19] ha [07:19] mvo: it's silly [07:19] mvo: this why http://pastebin.ubuntu.com/25923441/ [07:20] fopen fails, no worries, return true [07:20] we don't even log it [07:20] it would only show up on core [07:20] autsch [07:20] ok [07:20] and when we run on core and we treat it as classic we do the normal thing [07:20] and because it's actually meant to work [07:20] nothing failed [07:20] we did "base snaps" on core since this code was introduced [07:20] if you know what I mean by this shortcut [07:21] we were treating core as classic systems, setting up separate namespaces, constricting them out of base snaps [07:21] mvo: as for your earlier question, no it's still running [07:21] not sure if stuck [07:21] http://pastebin.ubuntu.com/25923446/ [07:22] (feels broken) [07:23] zyga-ubuntu: yeah, what did you do to fix it? [07:23] mvo: I just pushed fix/xdg-open-compat [07:23] but it's not a real fix I think (it's not working) [07:23] I just changed the version [07:24] --dest=com.canonical.SafeLauncher [07:25] I wonder if we are missing a service file for the legacy object path [07:25] (not the one at io.snapcraft....) [07:25] aha [07:25] yeah, no such launcher in data/dbus [07:26] zyga-ubuntu: yeah, that will just pull in the empty transitional package [07:26] PR snapd#4186 opened: tests: disable xdg-open-compat test [07:29] mvo: yeah, let's kill this test for now [07:29] PR snapd#4187 opened: tests: fix xdg-open-compat [07:30] zyga-ubuntu: I pushed also a followup which just downloads the right deb from LP - I pushed both so that we have at least one working, we need to unblock 2.29, I fear we will be starved by travis not doing tests for us [07:30] i can also kill cgo'ed user lookup while fixing the groups [07:30] zyga-ubuntu: so when we push the fix into the open PRs we need to make sure we push only to the important ones we want for 2.29 [07:30] mborzecki: \o/ [07:30] mborzecki: +1 for that [07:33] mvo: +1 [07:34] ok, I'll re-focus on lxd issues now [07:35] mvo: hi, I got that error (apt snap-xdg-open) also on my last PRs, it seems consistent afaict [07:35] * pedronis breakfast [07:37] pedronis: yes, we are fixing it as we speak === JoshStrobl|Away is now known as JoshStrobl [08:11] PR snapd#4160 closed: cmd/snap-confine: add slave PTYs and let devpts newinstance perform mediation (2.29) [08:13] zyga-ubuntu: does 4146 needs further changes ? it looks like there are commits for 4144 which are not part of 4146? [08:15] PR snapd#4188 opened: osutil: replace cgo bits with non-cgo, vendered os/user [08:17] mvo: what's the status of #4049, seems old but nobody ever looked at it? [08:17] PR #4049: debian,vendor: import github.com/snapcore/squashfs and use [08:18] pedronis: its in a sad state, its not fully working and its unclear if the vendor approach is powerful enough, go get/govendor make some assumptions that are hard to workaround [08:19] pedronis: I need to get back to it but have not managed it yet, we may need to do something simpler and abandon this (simpler as in not using govendoring but copy code/git submodule or smilar [08:20] mvo: #4046 has two +1 but mentions spread-cron, so not sure what to do with it [08:20] PR #4046: release: snapd 2.28.5 [08:20] heh sorry [08:20] mvo: #4043 has two +1 but mentions spread-cron, so not sure what to do with it [08:20] PR #4043: cmd/snap-confine: update valid security tag regexp [08:20] oops [08:20] * pedronis is not sure to be actually awake [08:20] lol [08:20] * mvo hugs pedronis [08:20] * mvo hands pedronis a cup of coffee as well [08:21] mvo: I meant #4063 [08:21] PR #4063: tests: add new kernel refresh/revert test for spread-cron [08:22] mvo: also unsure what needs to happen with #4059 [08:22] PR #4059: tests: add test that checks core reverts on core devices [08:24] pedronis: I think this one can go in and needs the coresponding spread cron branch to get merged. let me do this now [08:24] PR snapd#4059 closed: tests: add test that checks core reverts on core devices [08:24] pedronis: uh, it did not actually have two +1 :/ [08:24] pedronis: sorry, I mixed this up with the previous one. its manual: true so no harm [08:26] mvo: no, 4146 can go in as-is [08:26] mvo: #4129 has also two +1 , made a small comment there though [08:26] PR #4129: wrappers: do not error on incorrect Exec= lines [08:26] mvo: 4146 has the same fix without the extra pretty printing [08:27] zyga-ubuntu: ok [08:31] mvo: #4173 has also two +1 , but not sure where you want to redo the disabling of auto-refreshes in tests [08:31] PR #4173: corecfg: validate refresh.schedule when it is applied [08:33] pedronis: I can fix that there (in a bit, need to go over the 2.29 PRs as we need a new 2.29.3 :( [08:33] sorry, I'm distracting you [08:34] mvo: otoh you should garden your old PRs a bit more (friendly nag) [08:36] pedronis: no worries about distracting. and yes, agreed sorry for this [08:40] mvo: should we close #4049 for now then? [08:40] PR #4049: debian,vendor: import github.com/snapcore/squashfs and use === pachulo_ is now known as pachulo [08:43] pedronis: if we close we need another reminder that we need to do this work soon, it felt off the agenda once before. but yeah, in its current form it is not close to be mergable [08:44] mvo: I can mark it blocked if you prefer that way [08:44] is there no matching topic? [08:45] pedronis: please mark it as blocked [08:46] pedronis: no topic it seems, thats sad [08:46] zyga-ubuntu: I have some questions for 4146, how confident are you in the changes? [08:46] zyga-ubuntu: there are actual review questions in there too :) [08:47] pedronis: a second review for 4146 would be great [08:47] (or someone else, maybe pstolowski ?) [08:47] mvo: marked as blocked, but yes best would be to have an upcoming topic and close it if the approach is non-viable [08:48] mvo: #4063 has two +1 but it is failing afaict [08:48] PR #4063: tests: add new kernel refresh/revert test for spread-cron [08:49] pedronis: agreed, I updated the bug [08:49] pedronis: I will look into the failure [08:49] mvo, sure, gladly [08:59] zyga-ubuntu: I looked at #4146 , seems alright, I have an unrelated (in the sense it was like that before) there though [08:59] PR #4146: interfaces: fix udev tagging for hooks (2.29) [08:59] unrelated question [09:00] Hey so whoever snapped Electrum: thanks a lot! [09:00] beer++ [09:00] pedronis: yes? [09:00] zyga-ubuntu: sorry, see PR [09:00] now [09:00] ah, ok [09:01] ah, great question, let me check [09:02] thanks for your review pedronis [09:02] zyga-ubuntu: please don't get me wrong, I want to merge it (and like a lot of it). I'm just worried about the churn [09:02] pedronis: so looking at 61a4a0e442ab329f10773098448eb931fe6c355b I'm not sure [09:02] jamie patched one specific instance, not any of the generic helpers === __chip__ is now known as Chipaca [09:03] I think we need jdstrand to answer that; I don't quite know how that works [09:03] zyga-ubuntu: maybe something to chat with him about, anyway the code was like that before [09:03] jdstrand: (and I'd love to know) [09:04] yes, definitely [09:04] so it's not a blocker for the PR [09:04] yay [09:04] #4122 needs a 2nd review [09:04] PR #4122: configstate: add support for configure-snapd for snapstate.IgnoreHookError [09:04] mvo, pedronis: I agree about finding a better way to work on spec objects; [09:05] to explain it quickly, I wanted to stay consistent and not introduce a huge argument change patch [09:07] PR snapd#4186 closed: tests: disable xdg-open-compat test [09:07] zyga-ubuntu: yeah, thats ok, thank you. I was not aware it was a common pattern [09:08] zyga-ubuntu: (I still dislike it but not something for the PR in question :) [09:10] mvo: yes, no disagreement there [09:11] zyga-ubuntu: in #4062 seems you promised a couple more comment tweaks ... also probably then mvo and pstolowski should look at it again [09:11] PR #4062: cmd/snap: warn when a snap is not from the tracking channel [09:11] wrong PR [09:11] sorry [09:11] zyga-ubuntu: I meant #4068 [09:11] PR #4068: interfaces/builtin: add support for content "source" section [09:12] pedronis: 4068 is not ready for landing, I'll add a "blocked" tag there, it's meant to land after most of the layout patches are ready [09:12] heh ok [09:12] #4100 needs niemeyer [09:12] PR #4100: add ssh-keys, ssh-public-keys, gpg-keys and gpg-public keys interfaces [09:15] pedronis, zyga-ubuntu I think my general comment about cleanSubPath still needs addressing, would be good to fix it with 4068 [09:16] unless I miss something [09:17] * zyga-ubuntu looks [09:17] interesting [09:17] anyway, I'll return to that after bulk of layout code is in [09:19] zyga-ubuntu: sorry for being annoying - did you idea about the lxd fix work out? [09:20] mvo: no, not yet :/ [09:20] mvo: I'm writing a test that reproduces it [09:20] at least will be a start for another look [09:21] zyga-ubuntu: cool, still keen to have it for .3 if its small and targeted [09:26] pstolowski: there are still open questions how the API should work in #4108 [09:26] PR #4108: repo: ConnectedPlug and ConnectedSlot types [09:26] pedronis, I know, i'm getting to it, got sidetracked by the snapctl and cookie issues [09:27] pedronis, can you take a look at lanes helper PR? [09:27] pstolowski: yes, I'm going through PRs cronologically (mostly) [09:27] thanks [09:27] doing only reviews this morning === LarreaMikel1 is now known as LarreaMikel === JanC_ is now known as JanC [09:30] hmm [09:30] [ 1280.959106] audit: type=1400 audit(1510219691.912:133): apparmor="DENIED" operation="open" profile="snap.lxd.lxc" name="/var/lib/snapd/hostfs/usr/lib/os-release" pid=29008 comm="snap-exec" requested_mask="r" denied_mask="r" fsuid=0 ouid=0 [09:31] (unrelated but something we probably want to fix) [09:31] if anyone wants to make a quick PR for lxd interface [09:34] i can take a look at it [09:35] i'm going through review anyway, so might as well take a look at something else :) [09:37] zyga-ubuntu, commented on #4146 [09:37] PR #4146: interfaces: fix udev tagging for hooks (2.29) [09:38] pstolowski: ack on one, not sure about other [09:38] I'm deep in lxd if you wnat to push small tweak there please do [09:41] zyga-ubuntu: do all snap-update-ns PRs need a jdstrand review ? [09:41] pedronis: not all but some surely do [09:41] pedronis: the one about designing the "writable mimic" is the one I need a review on [09:42] pedronis: the ones about secure-mk{dir,file} do as well [09:43] ok, I have a bunch of older/big ones to go through first though [09:43] PR snapd#4172 closed: interfaces/kmod: simplify loadModules now that errors are ignored [09:48] pedronis: 4122 is good from your POV, right? just needs a second review? [09:54] mvo: yes, needs 2nd review [09:55] ta [10:05] pstolowski: #4152 has my +1, needs a re-review by zyga-ubuntu [10:05] PR #4152: snapd: fix snap cookie bugs [10:05] so... we're moving exclusively to /etc/{passwd,group} for lookups? [10:06] mborzecki: or were we already there and I hadn't noticed? [10:06] pedronis, thanks! [10:06] because if this is what we want to do, we need to refactor some code [10:06] in particular the code that looks up user dirs [10:06] this is relevant to my current work, so i can whip up the refactor right now [10:06] Chipaca: something we discussed yesterday with niemeyer, the idea is to kill cgo code (or most of it at least) [10:08] i've pulled in non-cgo pieces of os/user as osutil/user for now and put it up for the review [10:08] right, but this could lead to inconsistencies, which is why i'm checking -- if this is what we want to do i need to refactor a bit of code [10:08] i can do it after we merge the osutil/user pr [10:09] as i understand that's the direction we want to go, zyga-ubuntu mvo ? [10:09] Chipaca: I think we wanted to be cgo free and decided to kill the c-goat on the altar to attain that [10:10] Chipaca: whatever you need / do I think we're not adding more cgo [10:10] zyga-ubuntu: my question is: is /etc/passwd and /etc/group going to be the canonical (heh) source of users and groups on all snappy systems [10:10] because that's the impact in practice [10:10] we're not even looking at extrausers/groups [10:10] Chipaca: no, (see extrausers) [10:10] Chipaca: it's more subtle [10:10] Chipaca: we're killing reasons to know [10:11] Chipaca: not picking means to know [10:11] zyga-ubuntu: sorry, i don't understand [10:11] the code in #4188 looks up things only in /etc, no extra [10:11] PR #4188: osutil: replace cgo bits with non-cgo, vendored os/user [10:11] pstolowski: about LaneTasks, I said something wrong I think about lane 0 and then corrected myself but it got missed, I think lane 0 behavior is wrong, probably easy to find out with a different test [10:12] from this conversation i suspect the lack of extra is oversight [10:12] easily fixable [10:12] Chipaca: I mean, as far as I understand it, is that we haven't said we'll use /etc/passwd exclusively (yet), more like we said for each of the reasons to lookup users/groups _so far_ we want to avoid that reason [10:12] but the approach of making files the only source of users, that's policy now? (fine by me) (still needs a refactor from me) [10:12] * zyga-ubuntu killed qemu to "refresh" kernel state [10:12] man this mount business [10:12] zyga-ubuntu: did we discuss this changes with jdstrand btw ? he added that stuff I think [10:12] fishy stuff [10:13] zyga-ubuntu: in core we will be creating users and groups we will then fail to find [10:13] pedronis: no, [10:13] pedronis: we plan to today [10:13] that was just yesterday [10:13] ok [10:13] so I maintain that that aspect at least is a bug [10:13] pedronis, ah, thanks, will fix [10:14] zyga-ubuntu: your sentence about avoiding looking up things clashes with the pr we have which has a bunch of code for looking up things [10:14] Chipaca: yeah, I probably miss something then [10:14] you might be thinking about us not wanting to list all users [10:14] Chipaca: there's a related PR #4185 where we remove group from seccom rules, if that does not fly because of security concerns, we'll probably need to look into #4188 [10:14]