| andschwa | I need some help. I'm looking for the global configurations so I can change SNAP_DATA and SNAP_COMMON for all my snaps (so I can install them to a separate drive). | 01:49 |
|---|---|---|
| andschwa | But my god, I've searched and searched. There's docs on what the environment variables are, but not how to set them for snapd. | 01:49 |
| andschwa | Should I really perhaps be looking at mounting my drive to /var/snap? | 01:50 |
| andschwa | This seems like a super straightforward thing to do. | 01:56 |
| andschwa | And I cannot find it. | 01:56 |
| andschwa | Is anyone here, or is everyone on "Rocket"? | 01:59 |
| andschwa | Is what I want to do against snappy's design or something? | 02:05 |
| andschwa | It doesn't seem like it, since it's all centered around environment variables set at runtime by snapd for a snap. | 02:05 |
| andschwa | So it seems almost explicitly setup such that you can change what snapd sets those variables to. | 02:05 |
| andschwa | Yet I cannot find it. | 02:06 |
| === chihchun_afk is now known as chihchun | ||
| mup | PR # closed: snapd#2807, snapd#3120, snapd#3260, snapd#3372, snapd#3398, snapd#3484, snapd#3502, snapd#3556, snapd#3573, snapd#3616, snapd#3617, snapd#3621, snapd#3625, snapd#3635, snapd#3642, snapd#3697, snapd#3719, snapd#3720, snapd#3727, snapd#3734, snapd#3748, snapd#3750, snapd#3755, | 04:25 |
| mup | snapd#3759, snapd#3760, snapd#3773, snapd#3777, snapd#3779, snapd#3780, snapd#3781, snapd#3795, snapd#3797, snapd#3804, snapd#3805, snapd#3807, snapd#3810, snapd#3812 | 04:25 |
| mup | PR # opened: snapd#2807, snapd#3120, snapd#3260, snapd#3372, snapd#3398, snapd#3484, snapd#3502, snapd#3556, snapd#3573, snapd#3616, snapd#3617, snapd#3621, snapd#3625, snapd#3635, snapd#3642, snapd#3697, snapd#3719, snapd#3720, snapd#3727, snapd#3734, snapd#3748, snapd#3750, snapd#3755, | 04:26 |
| mup | snapd#3759, snapd#3760, snapd#3773, snapd#3777, snapd#3779, snapd#3780, snapd#3781, snapd#3795, snapd#3797, snapd#3804, snapd#3805, snapd#3807, snapd#3810, snapd#3812 | 04:26 |
| abeato | ogra_, do you know which is the current status of https://forum.snapcraft.io/t/updating-bootloader-assets-in-the-gadget-snap ? | 06:24 |
| === JoshStrobl is now known as JoshStrobl|zzz | ||
| mup | PR snapd#3759 closed: add spread test for wayland <Created by jdstrand> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/3759> | 07:39 |
| zyga-suse | good morning | 07:41 |
| zyga-suse | sorry for being late, I was the last one at breakfast, along with my sleepwalking daugther | 07:41 |
| zyga-suse | mvo: how are you doing today? | 07:44 |
| mvo | zyga-suse: hey, good morning. doing fine, how are you? | 07:45 |
| zyga-suse | pstolowski: hey, I think there's something you want to look into | 07:45 |
| zyga-suse | mvo: perhaps ^ before release | 07:45 |
| zyga-suse | mvo: downgrading from master to stable fails if you have any changes in flight that involves hooks | 07:45 |
| zyga-suse | because of the new hooks that don't have handlers | 07:46 |
| mvo | zyga-suse hm, interessting. what does the error look like? can you pastebin the session? | 07:47 |
| pstolowski | zyga-suse, hey | 07:47 |
| zyga-suse | mvo: I _may_ I ran into it over weekend | 07:48 |
| * zyga-suse scrolls back | 07:48 | |
| mvo | zyga-suse: thank you | 07:48 |
| zyga-suse | mvo: essentially this looks like the "CE cannot rollback" problem we had | 07:48 |
| mvo | zyga-suse: I'm mostly curious about the exact effect, if we really cannot rollback that is a problem | 07:49 |
| zyga-suse | mvo: aha, maybe that wording was too strong, I think the given change will fail but rollback as a whole will work (still just guessing) | 07:49 |
| mup | PR core#38 closed: Add another pi-config option <Created by sergey-borovkov> <https://github.com/snapcore/core/pull/38> | 07:50 |
| mup | PR core#54 closed: Use io.snapcraft.Launcher instead of com.canonical.SafeLauncher <Created by mvo5> <https://github.com/snapcore/core/pull/54> | 07:50 |
| mvo | zyga-suse: yeah, if that is the case its bad but not a blocker for the release. still something we want to fix | 07:51 |
| mup | PR core#38 opened: Add another pi-config option <Created by sergey-borovkov> <https://github.com/snapcore/core/pull/38> | 07:51 |
| mup | PR core#54 opened: Use io.snapcraft.Launcher instead of com.canonical.SafeLauncher <Created by mvo5> <https://github.com/snapcore/core/pull/54> | 07:51 |
| pstolowski | zyga-suse, what do I do exactly to reproduce? | 07:51 |
| mvo | zyga-suse, pstolowski: please keep me updated :) | 07:51 |
| zyga-suse | pstolowski: wait, let me find it | 07:51 |
| zyga-suse | mvo: one more thing, I merged that PR with apparmor probing code | 07:52 |
| zyga-suse | mvo: I re-designed it after your comment | 07:52 |
| zyga-suse | mvo: please have a look if you like that | 07:52 |
| mvo | zyga-suse: you merged it and then re-designed it? | 07:55 |
| zyga-suse | mvo: no, I read your comment, re-designed and merged | 07:55 |
| mvo | zyga-suse: ok, I have a look. but I would prefer to look before it gets merged (two +1 and all that) | 07:56 |
| zyga-suse | mvo: it's very similar to your suggestion and I had +1 from jamie so I merged it but I'll wait for re-affirmation next time | 07:57 |
| * zyga-suse still looks for that hook info | 07:57 | |
| zyga-suse | darn, where was that... | 08:02 |
| zyga-suse | pstolowski: sorry I cannot find it | 08:06 |
| zyga-suse | I think it was on IRC but you were already off | 08:06 |
| zyga-suse | but I cannot see it in my backlog | 08:06 |
| zyga-suse | pstolowski, the situation is as follows, start installing a snap using master, stop snapd and revert to stable | 08:07 |
| pedronis | I don't see many solutions here except changing the handling of hooks slightly in 2.26 | 08:10 |
| zyga-suse | I think being more graceful, as if the task was done in 2.26 would be good | 08:11 |
| zyga-suse | essentially you cannot rely on the hook | 08:11 |
| zyga-suse | since old snapd may run it | 08:11 |
| zyga-suse | and if you want to rely, use assumes | 08:12 |
| pedronis | is the revert of core itself that explodes? | 08:12 |
| zyga-suse | I don't remember, I think it was not the core but then again, it may be | 08:12 |
| zyga-suse | and in that case it would be unfortunate | 08:12 |
| mvo | zyga-suse: I put some comments into the PR 3808, nothing criticial but worth considering IMO | 08:14 |
| zyga-suse | sure, | 08:14 |
| mvo | zyga-suse: don't get me wrong, I appreciate the work on this feature and I know its frustrating to wait for reviews :) | 08:16 |
| zyga-suse | mvo: the nil result is something I copied from Chipaca's PR | 08:18 |
| zyga-suse | mvo: I can undo that, I was just surprised that you can call methods on typed nil just fine | 08:18 |
| zyga-suse | mvo: and I did it here | 08:18 |
| mvo | zyga-suse: right, its fine, I was also surprised about it (haven't seen that pattern) and that is one good reason to have it in code review to spread the knowledge etc :) | 08:19 |
| * zyga-suse got a telemarketer call now | 08:19 | |
| mvo | zyga-suse: do you remember what PR it was where Chipaca used this nil check pattern? curious to see how/why it was used there too | 08:19 |
| mvo | zyga-suse: meh, I hate those | 08:19 |
| zyga-suse | with a 500GB 3G/LTE data plan | 08:19 |
| zyga-suse | mvo: no, but maybe Chipaca remembers, it was very recent | 08:20 |
| zyga-suse | at 23euro / month | 08:20 |
| * zyga-suse is considering that | 08:20 | |
| zyga-suse | they also bundle some crap laptop | 08:20 |
| mvo | Chipaca: do you think you could have a look at 3795? it probably also needs a review from niemeyer but if we are happy with the code that will help I think (we need gustavo mostly for the concecpt behind it) | 08:22 |
| mvo | Chipaca: its also short :) | 08:22 |
| zyga-suse | mvo: replied | 08:23 |
| zyga-suse | mvo: let me know if you want me to change the nil pointer thing | 08:23 |
| zyga-suse | man, clouds == crap network | 08:24 |
| pstolowski | zyga-suse, you mean install a deb with master, then revert to debs from the distro? | 08:26 |
| pstolowski | (pardon my ignorance, but I'm not sure if it should involve some images from edge etc.) | 08:27 |
| zyga-suse | pstolowski: I think just install edge and go to stable | 08:29 |
| zyga-suse | pstolowski: and have the right snap install pending | 08:29 |
| pstolowski | k | 08:40 |
| * mwhudson waves | 08:40 | |
| zyga-suse | hey mwhudson | 08:43 |
| Chipaca | zyga-suse: mvo: good morning! | 08:49 |
| Chipaca | *technically* it's a national holiday here | 08:49 |
| zyga-suse | Chipaca: ohhh, nice | 08:49 |
| zyga-suse | so what are you doing here? | 08:49 |
| Chipaca | zyga-suse: having breakfast, and also i've been trying to snap a rather gnarly application | 08:49 |
| zyga-suse | +100 | 08:50 |
| zyga-suse | I love trying to snap things | 08:50 |
| zyga-suse | to see how that feels like | 08:50 |
| zyga-suse | and to see what issues I run into | 08:50 |
| Chipaca | it's looking like i can't make this one work without either bases or layouts | 08:50 |
| Chipaca | and that assumes layouts lets me replace /lib | 08:50 |
| Chipaca | zyga-suse: mvo: which PR were you two discussing? | 08:54 |
| zyga-suse | Chipaca: remember when you showed me a PR where you did methods on nils? | 08:57 |
| zyga-suse | Chipaca: and I was surprised this is doable in go | 08:57 |
| Chipaca | yes | 08:58 |
| mup | PR snapcraft#1512 opened: pluginhandler: clean error for BasePlugin.run{,_output} <Created by kalikiana> <https://github.com/snapcore/snapcraft/pull/1512> | 09:00 |
| zyga-suse | mvo: ^ | 09:01 |
| ogra_ | abeato, no idea, perhaps ask in the thread ? | 09:04 |
| pedronis | afaik it's been postponed | 09:05 |
| abeato | ogra_, ok, will do | 09:05 |
| mvo | Chipaca: context is my review of 3808, most specically https://github.com/snapcore/snapd/pull/3808/files#r135463240 | 09:06 |
| mup | PR snapd#3808: apparmor,release: add better apparmor detection/mocking code <Created by zyga> <Merged by zyga> <https://github.com/snapcore/snapd/pull/3808> | 09:06 |
| mvo | Chipaca: and was curious in what context you use it. in the context of this pr I was mostly wondering why not just return by-value but mostly I wanted some context as this is/was a new pattern for me | 09:07 |
| mup | PR core-build#11 closed: remove cruft from the writable-paths <Created by mvo5> <https://github.com/snapcore/core-build/pull/11> | 09:07 |
| mup | PR core-build#11 opened: remove cruft from the writable-paths <Created by mvo5> <https://github.com/snapcore/core-build/pull/11> | 09:08 |
| zyga-suse | mvo, Chipaca: please let me know if you want me to change that code, otherwise I'll move on to other work | 09:10 |
| mvo | meh, it looks like our update to github.com/snapcore/snapd/vendor/golang.org/x/sys/unix broke powerpc | 09:10 |
| * mvo looks | 09:10 | |
| zyga-suse | aww, sucks | 09:10 |
| zyga-suse | mvo: btw, golang 1.9 had some support changes for that arch | 09:10 |
| zyga-suse | they dropped some older HW | 09:10 |
| pedronis | mvo: we updated it because of the golang/x/crypto update, I don't even think we needed it before | 09:11 |
| pedronis | fwiw | 09:11 |
| * pstolowski amazed by mvo's code review | 09:13 | |
| zyga-suse | Chipaca: hmm, why doesn't logger.Noticef print stuff? | 09:14 |
| pstolowski | mvo, thank you for taking time & proposing the changes | 09:14 |
| pstolowski | ! | 09:14 |
| mvo | pedronis: aha, thanks, that is good to know, I will investigate, its a bit of a bummer | 09:14 |
| zyga-suse | Chipaca: but it does later | 09:14 |
| zyga-suse | 2017/08/28 11:13:52.385942 daemon.go:275: started snapd/2.27.3+git744.gd34c3ddc6 (series 16; classic; devmode) opensuse/20170823 (amd64) linux/4.12.8-1-default. | 09:14 |
| zyga-suse | that's the first message I see | 09:14 |
| mvo | pstolowski: your welcome, my pleasure | 09:14 |
| zyga-suse | but before that I call noticef + println in another spot and only the println shows up | 09:14 |
| mwhudson | zyga-suse: 1.9 changed stuff around ppc64 big endian, which has never been an ubuntu architecture | 09:14 |
| mvo | zyga-suse: go-1.9> things we should land? | 09:14 |
| mvo | zyga-suse: in master I mean? | 09:14 |
| zyga-suse | mwhudson: ah, good | 09:14 |
| mwhudson | zyga-suse: this failure is about powerpc | 09:14 |
| mwhudson | i.e. 32-bit | 09:14 |
| zyga-suse | aha | 09:15 |
| * zyga-suse shuts up then | 09:15 | |
| mwhudson | which stopped being an ubuntu architecture in ... zesty? | 09:15 |
| mwhudson | it ought to be possible to add gccgo / powerpc support to sys/unix, i never figured out how to do it thouh | 09:15 |
| mwhudson | (also didn't care overly much) | 09:16 |
| mvo | mwhudson: powerpc we probably still need to support because of 16.04, the failures look slightly nasty: http://paste.ubuntu.com/25416387/ | 09:16 |
| * zyga-suse sees SetLogger | 09:16 | |
| mwhudson | mvo: x/sys/unix just doesn't support powerpc at all | 09:16 |
| mvo | mwhudson: aha, great that you know more about this, so what are our options? can we avoid sys/unix somehow? | 09:16 |
| mwhudson | mvo: i don't know which bit of crypto is using it | 09:16 |
| Chipaca | mvo: sitting in the top of the tree, try: find -type f -name \*.go -print0 | xargs -0 perl -wne 'BEGIN{$n=0}; if ($n!=0) {$n=0; print "$ARGV: $f$ARGV: $_" if /if $v == nil/ }; if (/^func \((\S+) \*/) {$n=1; $f=$_; $v=$1}' | 09:17 |
| Chipaca | zyga-suse: where doesn't logger.Noticef print stuff? | 09:17 |
| mwhudson | mvo: support for other arches with gccgo has been added, so presumably the same can be done for ppc | 09:18 |
| zyga-suse | woah | 09:18 |
| zyga-suse | that's nice | 09:18 |
| Chipaca | zyga-suse: it might be trying to print stuff before it's initialised? | 09:18 |
| ogra_ | mvo, i just noticed that the initramfs-tools-ubuntu-core deb is behind the git tree, androidboot is missing and there was no deb build after the unit tests branch landed ... if i push that to the PPA today do i disturb any release ? | 09:18 |
| zyga-suse | Chipaca: I patched an init() function in a module to call it | 09:18 |
| zyga-suse | Chipaca: I understand that during startup we actually init the logger | 09:18 |
| Chipaca | yes | 09:18 |
| zyga-suse | Chipaca: and I got the short end of the ordering stick | 09:18 |
| Chipaca | probably | 09:18 |
| zyga-suse | hmmm | 09:18 |
| * zyga-suse wants the user warning framework | 09:19 | |
| zyga-suse | I could really use it now | 09:19 |
| Chipaca | zyga-suse: is this in debugging snapd? | 09:19 |
| zyga-suse | yes | 09:19 |
| zyga-suse | well | 09:19 |
| Chipaca | zyga-suse: just print it? | 09:19 |
| mvo | ogra_: thanks for checking, things should be fine today | 09:19 |
| zyga-suse | I want something more sticky | 09:19 |
| zyga-suse | jdstrand requested that we log that we are on partial apparomr | 09:19 |
| ogra_ | mvo, ok, then i'll upload | 09:19 |
| zyga-suse | "prominently" logged | 09:19 |
| zyga-suse | Chipaca: I'll just print it with a comment | 09:20 |
| pedronis | so the failing test now is passing, so probably a bunch of other tests might start failing, will see | 09:20 |
| Chipaca | mvo: that pattern is used in our code in a few places where the expected use of a something includes it sometimes being nil | 09:21 |
| ogra_ | so i built a wine snap last night ... | 09:22 |
| mvo | Chipaca: thanks, interessting, it escaped me so far | 09:22 |
| ogra_ | and since wine currently only does i386 i actually had to build for i386 ... | 09:22 |
| Chipaca | mvo: so rather than having “artifact:=default; if thing != nil {artifact = thing.GetArtifact()}”, you move the check into GetArtifact | 09:23 |
| ogra_ | and i must say i have never had such a painful experience with snapcraft :/ | 09:23 |
| ogra_ | (trying to build for i386 on amd64 host) | 09:23 |
| pedronis | Chipaca: mvo: go does it by static typing, but it exists as pattern since smalltalk (maybe even earlier), afaik it is liked/disliked ever since though | 09:24 |
| Chipaca | ogra_: you know you can ship i386 binaries in an amd64 snap, yes? | 09:24 |
| Chipaca | maybe that wasn't your issue :-) | 09:25 |
| ogra_ | Chipaca, but when buiolding from source you still need a i386 build env | 09:25 |
| Chipaca | anyway, i'll shut up now, seeing as how i'm not really working | 09:25 |
| mvo | Chipaca, pedronis: thanks, it makes perfect sense was mostly wondering a bit when/how to use it. thanks for your input on this :) | 09:25 |
| Chipaca | ogra_: “apt install build-essential:i386”? | 09:25 |
| ogra_ | and i dont want the 79234846 build deps wine needs installed on my host ... | 09:25 |
| Chipaca | ah | 09:25 |
| ogra_ | so i use cleanbuild | 09:25 |
| ogra_ | and then things explode | 09:26 |
| ogra_ | even using a plain lxd container as i386 system doesnt work properly | 09:26 |
| ogra_ | (there is an unfixed bug in libc) | 09:26 |
| mwhudson | pedronis: smalltalk is a bit different though, messages sent to nil are all ignored, iirc? | 09:26 |
| * zyga-suse suffers terrible network | 09:28 | |
| zyga-suse | just 2 days left | 09:28 |
| zyga-suse | I'll be home by Wednesday | 09:28 |
| pedronis | mwhudson: well, I was more referring to people adding implementation of things to nil | 09:30 |
| pedronis | which you can | 09:30 |
| mwhudson | ah | 09:31 |
| pedronis | (unless I'm totally misremembering this) | 09:31 |
| pstolowski | zyga-suse, - Run refresh hook of "core" snap if present (internal error: no registered handlers for hook "refresh") | 09:31 |
| pstolowski | zyga-suse, ^ is that what you saw when refreshing core from stable to edge and back to stable? | 09:31 |
| zyga-suse | yes | 09:34 |
| zyga-suse | that's what I saw | 09:34 |
| mup | PR snapd#3814 opened: interfaces: enable partial apparmor support <Created by zyga> <https://github.com/snapcore/snapd/pull/3814> | 09:34 |
| zyga-suse | pstolowski: is that a release blocker? | 09:34 |
| mvo | 3781 needs a second review, its very nice and not very long | 09:37 |
| pstolowski | zyga-suse, afaict we don't have this hook in release/2.27 | 09:39 |
| pedronis | pstolowski: a blocker for 2.28 | 09:40 |
| * zyga-suse looks at 3781 | 09:40 | |
| mvo | Chipaca: 3748 has a conflict | 09:40 |
| pstolowski | pedronis, oh, for 2.28 yes | 09:40 |
| pedronis | or for 2.27 if we decide we need to do something there | 09:40 |
| pedronis | it seems this a problem that will show up again and again with new hooks | 09:41 |
| pstolowski | yes | 09:43 |
| zyga-suse | pedronis, pstolowski: are you suggesting a 2.27.5 with some future-proofing for hooks in general? | 09:44 |
| zyga-suse | ahead of 2.28 release? | 09:44 |
| pedronis | I don't know | 09:44 |
| pedronis | I'm mostly musing | 09:44 |
| * zyga-suse nods | 09:47 | |
| pstolowski | i'm not sure. i need to fully understand the problem. hook machinery is a little bit deceptive with the 'Optional' flag. clearly the registration of hook handlers is critical | 09:48 |
| zyga-suse | optional feels like "may fail" | 09:49 |
| zyga-suse | but we need to know it exists | 09:49 |
| zyga-suse | which poses a problem for each added hook | 09:50 |
| pstolowski | i have a feeling that the only way around this is to make a missing handler non-critical | 09:50 |
| pedronis | if it's optional and the hooks doesn't exist in the snap, we don't care if the hook is registered or not | 09:50 |
| pedronis | internally | 09:50 |
| pedronis | but I don't remember the internal details to know if this is easy or not to tweak that way | 09:50 |
| pedronis | it probably breaks abstractions | 09:51 |
| === chihchun is now known as chihchun_afk | ||
| === chihchun_afk is now known as chihchun | ||
| pstolowski | uh oh we have a logic like this in place already, but i think it has a bug | 10:22 |
| === chihchun is now known as chihchun_afk | ||
| === chihchun_afk is now known as chihchun | ||
| === chihchun is now known as chihchun_afk | ||
| * tbr slaps chihchun_afk around a bit with a large and slimy trout. Dude, turn off your nick-changes. That annoys the crap out of people. | 10:38 | |
| zyga-suse | whee | 10:41 |
| zyga-suse | let me try it on ubuntu now | 10:42 |
| zyga-suse | mvo: I'm almost ready with 2.28 branches :) | 10:42 |
| zyga-suse | mvo: how are we doing with release freeze/fork | 10:42 |
| ogra_ | zyga-suse, oh, did you tell pstolowski about the "refresh back to beta" image we found on the weekend ? | 10:43 |
| ogra_ | (i didnt file a bug) | 10:43 |
| zyga-suse | ogra_: yes, I was wondering where I saw that | 10:43 |
| ogra_ | pstolowski, http://paste.ubuntu.com/25416803/ | 10:43 |
| zyga-suse | ogra_: was that on IRC? I couldn't find the discussion | 10:44 |
| zyga-suse | ogra_: woot, thank you :) | 10:44 |
| ogra_ | this is when moving to edge (for a test) and then trying to go back to beta | 10:44 |
| zyga-suse | mvo: ^ given this I would say we need to fix that for 2.28 | 10:45 |
| pstolowski | ogra_, yeah, that's the issue i'm looking on. nb, you can of course snap revert core | 10:46 |
| ogra_ | a revert would switch channels ? | 10:46 |
| mvo | zyga-suse: sounds like something we need to 2.28, yes. I understand pstolowski is looking into it(?) | 10:47 |
| pstolowski | ogra_, no, just mentioning, it won't fail that way | 10:47 |
| ogra_ | right | 10:47 |
| pstolowski | mvo, yes, i think i've a fix | 10:47 |
| ogra_ | it only fails when moving to edge and trying to go back though ... | 10:47 |
| ogra_ | (i doubt it affects the beta/candidate or stable channels yet) | 10:48 |
| zyga-suse | ogra_: and by extension it will fail when CE will test rollback from 2.28 | 10:48 |
| ogra_ | right | 10:48 |
| fgimenez | hey mvo, looks like CE QA found an issue with 2.27.4, i've just forwarded you the message | 10:55 |
| mvo | Chipaca: http://paste.ubuntu.com/25416853/ is probably something we need to fix for 2.28 | 10:55 |
| mvo | fgimenez: thanks, let me have a look | 10:56 |
| zyga-suse | fgimenez: can this be public somewhere please? | 10:56 |
| zyga-suse | mvo: also FYI, not for 2.28 (needs research first) but important as it breaks containers: https://bugs.launchpad.net/snapd/+bug/1712930 | 10:57 |
| mup | Bug #1712930: snap-confine: mounts happen in the wrong order <snapd:In Progress by zyga> <https://launchpad.net/bugs/1712930> | 10:57 |
| fgimenez | zyga-suse: not sure better ask them, they are posting to snap-update-verification@lists.canonical.com, maybe you can subscribe? | 10:57 |
| zyga-suse | fgimenez: I mean, I could do that but then again we have https://forum.snapcraft.io/t/in-progress-snapd-2-27-4/572 that everybody should post to | 10:58 |
| mvo | zyga-suse: hm, is this just changing the order of things (ideally switching two lines). or is there more work involved here? | 10:59 |
| fgimenez | zyga-suse: yeap, i already mentioned them about the forum and the release threads but don't know about how (or if) they make their test results public, they can explain better for sure :) | 11:03 |
| zyga-suse | mvo: not sure if this is just two lines :) | 11:05 |
| zyga-suse | mvo: could be trivial or very complex, I didn't look yet | 11:06 |
| mvo | ok | 11:06 |
| mup | PR snapcraft#1513 opened: lifecycle: outdated step should raise SnapcraftError <Created by kalikiana> <https://github.com/snapcore/snapcraft/pull/1513> | 11:06 |
| mup | PR snapd#3815 opened: tests: install test-snapd-complexion on core <Created by mvo5> <https://github.com/snapcore/snapd/pull/3815> | 11:18 |
| mvo | Chipaca: trivial reproducer -^ | 11:18 |
| zyga-suse | hmmmmmmm | 11:22 |
| * zyga-suse is unhappy about https://travis-ci.org/snapcore/snapd/builds/269141620?utm_source=github_status&utm_medium=notification | 11:22 | |
| pedronis | mvo: I think it's an holiday for John | 11:22 |
| zyga-suse | is that just another bug in apparmor profile loading for tests | 11:22 |
| zyga-suse | or a real issue | 11:22 |
| zyga-suse | indeed | 11:22 |
| mwhudson | mvo: i wrote up that dput linter wrapper thing I talked about before https://gist.github.com/mwhudson/616499edb1191bd99c987bbbd8781ce9 | 11:22 |
| zyga-suse | he is off | 11:22 |
| mvo | pedronis: oh, thanks | 11:22 |
| mvo | Chipaca: weeeh, sorry - you are off today? I won't bother you anymore | 11:22 |
| mvo | mwhudson: nice | 11:23 |
| zyga-suse | mvo: question, will we re-load snap-confine's apparmor profile before configuring core? | 11:23 |
| === yofel_ is now known as yofel | ||
| mup | PR snapd#3816 opened: hooks: do not error out when hook is optional and no hook handler is registered <Created by stolowski> <https://github.com/snapcore/snapd/pull/3816> | 11:45 |
| pstolowski | zyga-suse, mvo ^ this should fix the problem with hooks, but I only tested with unit test | 11:45 |
| mup | PR snapd#3817 opened: tests: wait more and more debug info about fakestore start issues <Created by pedronis> <https://github.com/snapcore/snapd/pull/3817> | 11:51 |
| mup | PR snapcraft#1510 closed: ci: disable the travis deploy stage for docs <Created by sergiusens> <Merged by sergiusens> <https://github.com/snapcore/snapcraft/pull/1510> | 11:54 |
| mup | PR snapcraft#1508 closed: schema: version should have a max length of 32 <bug> <Created by kyrofa> <Merged by sergiusens> <https://github.com/snapcore/snapcraft/pull/1508> | 12:12 |
| cachio | mvo, did you see the email I fw? | 12:52 |
| mvo | zyga-suse, spineau, jdstrand: so I think I have an idea about the problem with "sudo network-manager.nmcli d show eth0" getting denies on 2.27 - we enabled seccomp argument filtering for socket. and the profile has "# socket AF_NETLINK - -" and I assume that nmcli uses the netlink socket | 12:52 |
| mvo | cachio: indeed I did and after a small panic I started debugging | 12:52 |
| cachio | mvo, ohhh, any finding? | 12:53 |
| mvo | cachio: yes, maybe, I think I need jdstrand to help me, but I suspect we need to enable socket AF_NETLINK at least for nmcli, maybe for more. | 12:54 |
| cachio | mvo, is this caused because the last change done for 2.27.4? | 12:55 |
| zyga-suse | mvo: there's a new set of netlink interfaces | 12:55 |
| zyga-suse | mvo: maybe those are sufficient there | 12:55 |
| zyga-suse | mvo: I wish we knew the values of other arguments but I suspect I can easily check that locally | 12:55 |
| zyga-suse | mvo: (if it is netlink or not) | 12:55 |
| zyga-suse | mvo: AFAIK it used KOBJECT_UEVENT but perhaps I'm wrong | 12:56 |
| spineau | mvo: like mmcli? that's the first one I can think of | 12:56 |
| zyga-suse | mvo: note that jamie is off today, I'll check it out after checking out one other thing | 12:56 |
| zyga-suse | mvo: if you can strace it on your machine and pastebin that it would be sufficient | 12:56 |
| spineau | zyga-suse: mvo: I'm still connected to our test device, if I can help you by providing more logs, just ask | 12:59 |
| zyga-suse | spineau: thank you, if you know how to run strace under snap confinement (not trivial) then please do that | 12:59 |
| zyga-suse | spineau: you need to edit the seccomp/apparmor profiles, allow extra stuff and run strace you get from the host somehow | 13:00 |
| zyga-suse | spineau: don't worry if you don't want to go through those hoops | 13:00 |
| spineau | zyga-suse: unfortunately, I'm still a snap newbie | 13:00 |
| mvo | zyga-suse: strace is not working for me, hangs | 13:01 |
| mvo | zyga-suse: when I enable AF_NETLINK things work | 13:01 |
| zyga-suse | mvo: strace needs some love, ok | 13:01 |
| mvo | zyga-suse: socket AF_NETLINK - - | 13:02 |
| mvo | zyga-suse: if that is part of the profile things work | 13:02 |
| zyga-suse | mvo: note that we don't have any interfaces that add that | 13:03 |
| zyga-suse | mvo: we have "socket AF_NETLINK - NETLINK_AUDIT" and similarly NETLINK_CONNECTOR | 13:03 |
| zyga-suse | we need to know which one is really required | 13:03 |
| zyga-suse | mvo: and perhaps consider an approach (new release of network-manager or a snap-id specific quirk) | 13:03 |
| willdeberry | so it seems that everything was passing, but since I have been rebasing just to keep things fresh, travis build now fails: https://github.com/snapcore/snapd/pull/3812 | 13:26 |
| mup | PR snapd#3812: Expose bluez interface on classic OS <Created by willdeberry> <https://github.com/snapcore/snapd/pull/3812> | 13:26 |
| zyga-suse | mvo: offtopic | 13:34 |
| pedronis | niemeyer: will you create a tracker topic in the forum for the current PRs? | 13:34 |
| zyga-suse | mvo: I updated my x250 to zesty | 13:34 |
| zyga-suse | mvo: and ... the update removed xorg-input-all | 13:34 |
| zyga-suse | mvo: that was curious to debug | 13:34 |
| zyga-suse | mvo: apparently people ran into this too and I found lots of forum posts and what not | 13:35 |
| zyga-suse | mvo: so now things work | 13:35 |
| zyga-suse | mvo: when you don't have xorg-input-all and its deps you have GDM/lightdm with *no* keyboard interaction at all | 13:35 |
| zyga-suse | not even alt-ctrl-f2 to get out | 13:35 |
| niemeyer | pedronis: I'm on the fence about it.. I want to make sure we take it as an actual sprint if it's added, and given all the work on the release we may not be able to do it | 13:36 |
| zyga-suse | aha, so I know what is wrong with the spread setup, nothing, it looks like a very odd and curious apparmor behavior | 13:36 |
| niemeyer | pedronis: Perhaps we can kick it off once the release is clean? | 13:36 |
| pedronis | niemeyer: ok, just wondering because I start to have a bit of hard time finding/remembering what to review | 13:36 |
| niemeyer | pedronis: Aha, I see | 13:37 |
| niemeyer | pedronis: Let me create one then | 13:37 |
| mup | PR snapd#3818 opened: interfaces: fix network-manger plug <Created by mvo5> <https://github.com/snapcore/snapd/pull/3818> | 13:37 |
| mup | PR snapd#3819 opened: hooks: do not error out when hook is optional and no hook handler is registered (2.27) <Created by stolowski> <https://github.com/snapcore/snapd/pull/3819> | 13:38 |
| mvo | 3260 needs a second review, tests are green now | 13:38 |
| mvo | zyga-suse: 3818 is maybe what we need to fix the nmcli issue, an early look is appreciated, I mostly opened it to get spread tests running | 13:39 |
| zyga-suse | mvo: ack | 13:43 |
| pedronis | snapd#3817 needs quick a review (it's about the fakestore tests that were sill failing sometimes for me in autopkgtests) | 13:43 |
| mup | PR snapd#3817: tests: wait more and more debug info about fakestore start issues <Created by pedronis> <https://github.com/snapcore/snapd/pull/3817> | 13:43 |
| zyga-suse | re-started something else and looking | 13:43 |
| niemeyer | Forgot to ask.. did we have any other issues on the testing infra after the Spread changes I did on Friday? | 13:43 |
| ogra_ | wow, cool ... | 13:43 |
| ogra_ | seems my wine snap really works with all exe installers i have lying around ... and all install just go fine to SNAP_USER_DATA | 13:44 |
| ogra_ | *all installs | 13:44 |
| mvo | ogra_: thats very neat, wine is a nice snap to have | 13:45 |
| ogra_ | well, needs --devmode | 13:45 |
| ogra_ | (or --classic, but i didnt feel like making a classic snap) | 13:45 |
| ogra_ | it wants to read / and /boot | 13:45 |
| mvo | makes one wonder why that :) | 13:46 |
| ogra_ | but it it really cool that you can just install wine 2.15 next to the deb one | 13:46 |
| ogra_ | because it can put .desktop files in place and such | 13:46 |
| ogra_ | for apps it installs | 13:46 |
| mvo | yeah, also wine might be perfect for tracks, i.e. being able to install 2.0, 2.1, etc | 13:46 |
| ogra_ | yeah | 13:46 |
| ogra_ | i only did it out of curiosity ... but there is a snapcraft source tree now that someone can take and make something real out of it | 13:47 |
| ogra_ | our forum doesnt work with the sippped wine IE :/ | 13:47 |
| ogra_ | *shipped | 13:47 |
| * mvo nods | 13:48 | |
| ogra_ | "your browser is to old" :P | 13:48 |
| ogra_ | snapcraft.io itself and ubuntu.com work fine | 13:49 |
| mup | PR snapd#3817 closed: tests: wait more and more debug info about fakestore start issues <Created by pedronis> <Merged by pedronis> <https://github.com/snapcore/snapd/pull/3817> | 13:53 |
| zyga-suse | mvo: re, thank you for checking it is UEVENT | 13:55 |
| zyga-suse | let me look at something related and I will review this | 13:55 |
| mvo | zyga-suse: no real rush, I want to see the test results first, maybe its actually not working | 13:57 |
| zyga-suse | I commented | 13:58 |
| mvo | zyga-suse: aha, indeed, silly me, you are right. I will fix | 13:59 |
| zyga-suse | thanks! | 13:59 |
| zyga-suse | I checked other interfaces | 14:00 |
| zyga-suse | they all use permanent snippet but for slot | 14:00 |
| zyga-suse | for plug I think it's fine to keep this on connected | 14:00 |
| mup | PR snapcraft#1514 opened: docs: add github processed templates <Created by sergiusens> <https://github.com/snapcore/snapcraft/pull/1514> | 14:07 |
| === JanC is now known as Guest96910 | ||
| === JanC_ is now known as JanC | ||
| mup | PR snapd#2807 closed: snap: add new `snap switch` command <Created by mvo5> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/2807> | 14:11 |
| mup | PR snapd#3820 opened: spread: don't set HTTPS?_PROXY for linode <Created by zyga> <https://github.com/snapcore/snapd/pull/3820> | 14:25 |
| zyga-suse | mvo: ^ trivial one | 14:26 |
| zyga-suse | tested locally | 14:31 |
| zyga-suse | mvo: looking at the ordering issue now | 14:33 |
| mvo | zyga-suse: great, thank you! | 14:33 |
| mvo | zyga-suse: I'm waiting for tests and I think 3818 will need a review from jamie as well, right? | 14:34 |
| zyga-suse | yes | 14:34 |
| zyga-suse | note that jamie is off today | 14:34 |
| niemeyer | pedronis: https://forum.snapcraft.io/t/review-sprint-3/1880 | 14:43 |
| pedronis | niemeyer: thanks | 14:49 |
| niemeyer | pedronis: No problem.. tweaked the format slightly as apparently the recent changes in Discourse modified some of the old icons we were using, but still same thing | 14:52 |
| mup | PR snapd#3821 opened: tests: new regex used to validate the core version on extra snaps ass <Created by sergiocazzolato> <https://github.com/snapcore/snapd/pull/3821> | 14:52 |
| * zyga-suse moved indoors, too cold | 15:08 | |
| === JoshStrobl|zzz is now known as JoshStrobl | ||
| soee | hi, i have installed an app via snap and have this: snap-confine has elevated permissions and is not confined but should be. Refusing to continue to avoid permission escalation attacks | 15:45 |
| soee | any idea what is wrong ? | 15:46 |
| zyga-suse | hey | 16:00 |
| zyga-suse | soee: let me try to help you | 16:00 |
| zyga-suse | soee: can you start by running "snap version" and pasting the result | 16:00 |
| Chipaca | mvo: ah! so on core we need to not do the snippets (and source complete.sh) | 16:01 |
| * Chipaca disappears again | 16:01 | |
| cachio | pedronis, hey, any idea why this error could be happening runing tests on staging ? | 16:27 |
| cachio | error: cannot fetch snap signatures/assertions: circular assertions are not expected: account (canonical) | 16:27 |
| pedronis | cachio: sounds it's not using the right kind of snapd | 16:41 |
| pedronis | maybe reexec doing something wrong | 16:41 |
| pedronis | cachio: fwiw I used a correctly built snapd against staging on Friday and things worked | 16:42 |
| pedronis | snapd built from master | 16:42 |
| cachio | pedronis, how did you setup the tests? | 16:43 |
| cachio | pedronis, which variables did you use? | 16:43 |
| cachio | I want to replicate that into the sread-cron tests against staging | 16:44 |
| pedronis | cachio: afair the test setup should do the right thing, there are two part to this, you need a snapd built with withtestkeys or withstagingkeys | 16:44 |
| pedronis | cachio: then you need SNAPPY_USE_STAGING_STORE=1 | 16:44 |
| cachio | pedronis, I have that | 16:45 |
| cachio | export SPREAD_REMOTE_STORE=staging | 16:45 |
| cachio | it is traduced into SNAPPY_USE_STAGING_STORE=1 | 16:45 |
| pedronis | and DEB_BUILD_OPTIONS='nocheck testkeys' dpkg-buildpackage -tc -b -Zgzip | 16:46 |
| pedronis | the testkeys should do the other part | 16:46 |
| pedronis | I need a bit more details where this exactly is breaking? breaking early or in some specific test? | 16:47 |
| pedronis | anyway I need to go have dinner | 16:48 |
| cachio | pedronis, on those tests https://travis-ci.org/snapcore/spread-cron/builds/268521976#L1224 | 16:52 |
| cachio | there are many tests failing for he same reason | 16:52 |
| pedronis | cachio: I'll look in my morning | 17:11 |
| cachio | pedronis, sure, tx | 17:11 |
| cachio | I am taking a look right now too | 17:11 |
| pedronis | notice that some are other platforms | 17:13 |
| pedronis | I mean != ubuntu | 17:13 |
| pedronis | I don't think it makes sense to run those tests against staging | 17:13 |
| pedronis | like that failure is for opensuse | 17:14 |
| pedronis | I don't know if it knows how to build a staging snapd | 17:14 |
| cachio | ok | 17:16 |
| cachio | pedronis, i'll change that too, thanks | 17:16 |
| pedronis | cachio: I mean until we understand more I would skip anything not ubuntu for staging tests | 17:16 |
| cachio | pedronis, sure | 17:20 |
| pedronis | the main user of these tests would be store after a staging deployment | 17:21 |
| pedronis | I think they have been too flaky to use them tough :/ | 17:21 |
| cachio | pedronis, yes, agree | 17:24 |
| cachio | most of them have been failing for a while | 17:24 |
| cachio | pedronis, I think staging should be used just to validate some specific stuff | 17:25 |
| cachio | pedronis, so it is easier to keep it in green | 17:25 |
| zyga-suse_ | hmm | 17:44 |
| zyga-suse_ | - Run install hook of "core" snap if present (internal error: no registered handlers for hook "install") | 17:44 |
| zyga-suse_ | https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-xenial-snappy-dev-image/xenial/i386/s/snapd/20170828_165347_9cd1b@/log.gz | 17:44 |
| zyga-suse_ | from upgrade/basic | 17:44 |
| === cachio is now known as cachio_afk | ||
| === JoshStrobl is now known as JoshStrobl|Work | ||
| stormmore | How difficult would it be to write a snap that can read /etc with making using --classic? | 19:14 |
| === cachio_afk is now known as cachio | ||
| mup | PR snapd#3821 closed: tests: new regex used to validate the core version on extra snaps ass <Created by sergiocazzolato> <Merged by sergiocazzolato> <https://github.com/snapcore/snapd/pull/3821> | 19:34 |
| niemeyer | stormmore: With classic or without classic? | 19:41 |
| niemeyer | pedronis, zyga-suse_, cachio: Review board now has details on who's reviewing or needs to review | 19:42 |
| cachio | niemeyer, nice | 19:43 |
| stormmore | niemeyer: preferably without | 19:43 |
| niemeyer | stormmore: That needs an interface for the specific need you have.. many are already available, but it may need to be created as well if it's not yet there | 19:47 |
| niemeyer | stormmore: With classic, it's just a matter of opening the files.. nothing fancy to be done | 19:48 |
| stormmore | niemeyer: whats a good place to see a list of interfaces. I did look there as that was what I was suspecting but I didn't find one that would give even read-only (not sure that is going to be sufficient for my needs yet but gotta start somewhere) access to /etc from their desc | 19:49 |
| niemeyer | stormmore: No interface will give access to all of /etc | 19:50 |
| niemeyer | stormmore: Some will give access to specific bits inside /etc, according to their described purpose | 19:50 |
| stormmore | niemeyer: ah that is why I am probably leaning towards making this a classic snap (at least initially) | 19:51 |
| niemeyer | stormmore: Yeah, it's a fine way to get started | 19:51 |
| stormmore | niemeyer: thanks, just wanted to see if there was a "better way" but looks like I was on the right track | 19:52 |
| mup | PR snapd#3120 closed: [WIP] interfaces/hooks: expose attrs to the interface API, snapctl enhancements (step #4) <Created by stolowski> <Closed by niemeyer> <https://github.com/snapcore/snapd/pull/3120> | 20:00 |
| zyga-suse_ | niemeyer: wow, that's nice | 20:08 |
| zyga-suse_ | WOW, way nicer than I expected :D | 20:08 |
| zyga-suse_ | really nice work :) | 20:10 |
| zyga-suse_ | niemeyer: I wonder what will happen when we have initial clashes in the team | 20:10 |
| niemeyer | zyga-suse_: Wow, thanks :) | 20:11 |
| niemeyer | zyga-suse_: You mean more than 26 reviewers? ;P | 20:11 |
| zyga-suse_ | well, if we get another zygmunt, will we call him z^1 ;D | 20:12 |
| niemeyer | zyga-suse_: We already have clashes.. look closely | 20:12 |
| niemeyer | and apparently we also have a bug.. something is wrong with #3260.. looking | 20:13 |
| zyga-suse_ | ah, indeed | 20:13 |
| * zyga-suse_ is happy that zygmunt is an uncommon word and his life is simple here | 20:14 | |
| cachio | niemeyer, do you know which is the job to sync prod with staging stores? | 20:19 |
| pedronis | cachio: there's no such job | 20:21 |
| cachio | pedronis, ok, is it a manual procedura? | 20:22 |
| pedronis | yes, we typically add relevant snaps to staging | 20:22 |
| pedronis | staging store and prod store are fairly indepedent | 20:22 |
| cachio | ok, so I should download/upload the snaps | 20:23 |
| cachio | pedronis, I'll try that | 20:23 |
| pedronis | anyway probably another good reason to cut down staging tests | 20:24 |
| cachio | pedronis, yes | 20:25 |
| cachio | pedronis, if it is not automatic, it does not scale | 20:25 |
| pedronis | not sure how to do that though, spread doesn't have more tagging than manual | 20:25 |
| pedronis | skipping support is also missing, our skips atm ad hoc | 20:25 |
| pedronis | one approach could be just have a different suite and symlinks | 20:27 |
| cachio | pedronis, why we are running tests against staging? | 20:29 |
| cachio | pedronis, if there is not something new to test on there the best we can do is to remove those execution from spread-cron as you suggested | 20:30 |
| pedronis | cachio: ? the idea was to run them each time one of the store services gets a new staging deploy ? | 20:30 |
| pedronis | that's what triggers runs | 20:31 |
| pedronis | to check that changes to the store don't break the contract with snapd | 20:31 |
| pedronis | that was the idea | 20:31 |
| cachio | pedronis, makes sense | 20:31 |
| pedronis | yes, it's a sound idea, the executation hasn't been amazing so far though, also probably too many not relevant tests | 20:32 |
| cachio | so, we just need to keep the databases synchronized | 20:32 |
| pedronis | and also the store has changed, so we need different checks | 20:32 |
| pedronis | cachio: ? | 20:32 |
| pedronis | don't think that's going to happen | 20:32 |
| cachio | pedronis, I mean, perhaps the solution is to add a job to keep both in sync | 20:34 |
| pedronis | we can add something that keeps some subset of snaps in sync | 20:34 |
| pedronis | but in general afaik there's no plan to keep them generally in sync | 20:34 |
| cachio | pedronis, otherwise always when we do a change in a snap we need to make that same change in staging | 20:35 |
| cachio | is it correct? | 20:35 |
| pedronis | that's how we did so far, yes | 20:35 |
| cachio | pedronis, ok | 20:36 |
| cachio | pedronis, what do you suggest, disable the staging jobs? or continue doing that sync manually? or start doing something to keep it on sync | 20:37 |
| pedronis | cachio: I think we need to take a step back and setup a meeting with the stakeholder interested in that | 20:38 |
| pedronis | it's not about what *I* suggest | 20:38 |
| pedronis | *stakeholders | 20:39 |
| cachio | pedronis, ok, makes sense, how are the interested stakeholders? | 20:39 |
| cachio | who | 20:39 |
| cachio | celso? | 20:39 |
| pedronis | I can ask tomorrow | 20:40 |
| cachio | pedronis, sure | 20:40 |
| cachio | let's talk about that tomorrow | 20:40 |
| niemeyer | Okay, figured the issue and fixed.. board updated again, with correct data. | 21:16 |
| === matteo is now known as matteo` | ||
| === matteo` is now known as matteo | ||
| === matteo is now known as matteo_ | ||
| === matteo_ is now known as matteo| | ||
| === matteo| is now known as matteo | ||
| mup | PR snapd#3760 closed: Allow snap-confine to be confined even with --disable-apparmor <Created by mwhudson> <Closed by mwhudson> <https://github.com/snapcore/snapd/pull/3760> | 22:30 |
| mup | PR snapd#3260 closed: cmd/snap: implement userd command as replacement for snapd-xdg-open <Created by morphis> <Merged by niemeyer> <https://github.com/snapcore/snapd/pull/3260> | 23:36 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!