/srv/irclogs.ubuntu.com/2021/09/09/#snappy.txt

=== E_Eickmeyer is now known as Eickmeyer
mupPR snapd#10758 opened: image/image_linux.go: add newline <Simple 😃> <Skip spread> <Created by anonymouse64> <https://github.com/snapcore/snapd/pull/10758>03:10
mborzeckimorning05:00
mardy'morning!05:43
mborzeckiwhat a mess with boltdb, upstream was forked once, then since coreos became part of rh, it got moved again06:04
mborzeckisome distros have packages of the intermediate location (at github/coreos), but not the new one at etcd06:05
mborzeckiand ofc go get github.com/coreos/bbolt fails because github redirects coreos to etcd-io for this repo go get github.com/coreos/bbolt06:05
mborzeckihttps://paste.ubuntu.com/p/nHbSR4V5qK/06:05
mardymborzecki: about your suggestion for the systemdTooOldError: should I actually make it public (and probably call it TooOldError, since we have the package prefix anyway)?06:26
mborzeckimardy: probably doesn't need to be public, if you really want to check if it's that specific error you can have `func IsSystemdTooOld(err error) { _, ok := err.(*systemdTooOldError); return ok }` ?06:28
mardymborzecki: your suggestion would actually allow us to use systemd.EnsureAtLeast() here too: https://github.com/snapcore/snapd/blob/master/usersession/userd/privileged_desktop_launcher.go#L98-L11006:28
mardybut that requires the error to be public, I guess06:28
mardymborzecki: ah, ok, if you think that IsSystemdTooOld() is better, I'll go with that06:29
mborzeckihmm broke remodel, fixed remodel06:53
pstolowskimorning07:04
mardypstolowski: hi!07:14
mvogood morning pstolowski and mardy and mborzecki 07:16
mborzeckimvo: pstolowski heya07:16
mupPR snapd#10688 closed: configcore: add read-only netplan support <Created by mvo5> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/10688>07:21
mupPR snapd#10757 closed: build-aux: stage libgcc1 library into snapd snap <cherry-picked> <Created by mwhudson> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/10757>07:21
mardymvo: hi!08:12
mborzeckieven dropped a little reactor of remodel tasks, meh that code was nasty, still is but a just a little less08:18
mwhudsonmvo, mborzecki: how often does the snapd snap get updated in stable, out of curiousity?08:38
mvomwhudson: in theory every 4 weeks, in pratice it's a lot more volatile because we recently got a lot "dot" releases for customers08:39
mwhudsonmvo: ok08:39
mvomwhudson: why (just out of curisoity)08:40
mwhudsonmvo: somehow that libgcc_s.so.1 thing is making snapd autopkgtests fail08:40
mwhudsonhttps://autopkgtest.ubuntu.com/packages/s/snapd/impish/amd6408:41
mvomwhudson: oh, we will do a new 2.52 soon, I can also upload out-of-band releases to impish if that is needed (but in a meeting right now so let's chat in a bit what I can do to help :)08:41
mwhudsonmvo: it's really the snapd snap, not the package in impish (afaict)08:41
mvomwhudson: uh, ok08:42
mwhudsonat least that's what https://bugs.launchpad.net/snapd/+bug/1943077 says08:45
mupBug #1943077: snapd fails to autopkgtest on mksquashfs, which is looking for libgcc_s <update-excuse> <snapd:New> <openssh (Ubuntu):New> <squashfs-tools (Ubuntu):New> <https://launchpad.net/bugs/1943077>08:45
mupPR snapd#10725 closed: o/hookstate: require snap-refresh-control interface for snapctl refresh --proceed <Needs Samuele review> <Refresh control> <Created by stolowski> <Merged by stolowski> <https://github.com/snapcore/snapd/pull/10725>09:11
mborzeckimvo: thanks for updating snapcore/bolt, i've updated https://github.com/snapcore/snapd/pull/10754 and we'll see how it fares now09:41
mupPR #10754: tests/lib/prepare-restore: make sure that package build is done without network access <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/10754>09:41
mupPR snapd#10759 opened: tests: be more robust against a new day stepping in <Created by mardy> <https://github.com/snapcore/snapd/pull/10759>09:47
mvomborzecki: \o/09:48
mvomborzecki: our of curiosity, was a "go get -u" enough for tihs?09:48
mborzeckimardy: cannot approve your pr sorry 🙂 you're missing a reference to https://www.youtube.com/watch?v=NaGLVS5b_ZY09:49
mborzeckimvo: heh no, had to go get -u github.com/snapcore/bolt@master otherwise it just picked another tag with v1.3.x09:49
mvomborzecki: oh nice, good to know!09:50
laneyquick one (I'm ok filing a bug if there isn't one). seeding is failing on a new impish install like this: https://paste.debian.net/1211056/ - is this known, and can anyone provide me with a workaround for now? :-)10:01
mborzeckilaney: does /etc/apparmor.d/abstractions/base exist?10:02
laneymborzecki: erm WTF10:04
laneylaney@sherwood:/etc/apparmor.d/abstractions$ ls -l10:04
laneyls: cannot access 'base': No such file or directory10:04
laneyand... -????????? ? ?    ?       ?            ? base10:05
mborzeckilaney: hmm https://packages.ubuntu.com/impish/amd64/apparmor/filelist there is /etc/apparmor.d/abstractions/base, borked fs?10:05
laneylooks like something weird happening here10:06
laneylet me poke, thanks for the pointer10:06
mborzeckimvo: so not it's looking for: https://github.com/canonical/go-sp800.90a-drbg10:08
mborzeckiapparently it's needed by secboot, but it's listed in go.mod, so why earlier go mod vendor did not fill that dependency?10:10
mvomborzecki: that is interessting, I have no idea why it's missing from go mod :/10:10
mvomborzecki: maybe conditional somehow via buildflags?10:11
mborzeckimvo: well it's listed there as an indirect dependency, but i'm not sure why go mod vendor has not fetched the repo10:11
mardymborzecki: :-D10:11
mupPR snapd#10760 opened: o/snapstate: support ignore validation flag on install/update <â›” Blocked> <Needs Samuele review> <validation-sets :white_check_mark:> <Created by stolowski> <https://github.com/snapcore/snapd/pull/10760>10:57
mborzeckimvo: ok, so i got something about that build problem, we have a buch of `//go:generate` which invokes go run, or runs a script (update-pot) which in turn runs go install without all the flags11:17
jdstrandamurray: hey, mentioning this as a reminder to folks of why we might not autoconnect something that most snaps don't need even if on the surface it seems like it wouldn't be a problem: https://ubuntu.com/security/CVE-2021-3612 (joystick interface)11:22
mardymvo: can you merge this? https://github.com/snapcore/snapd/pull/10653 Or should I rather retry the build, since if failed on so many machines (looks like I hit a multitude of unlucky situations :-D)11:34
mupPR #10653: mount-control: step 1 <Needs Samuele review> <Created by mardy> <https://github.com/snapcore/snapd/pull/10653>11:34
amurrayhey jdstrand :) - that is a great example - I recall asking you that very question once and you explained it really well at the time11:47
mupPR snapd#10761 opened: o/snapstate: optimize conflicts around snaps stored on conditional-auto-refresh task <Refresh control> <Created by stolowski> <https://github.com/snapcore/snapd/pull/10761>11:47
mupPR snapd#10762 opened: o/servicestate: Update task summary for restart action <Created by mardy> <https://github.com/snapcore/snapd/pull/10762>12:27
mupPR snapd#10763 opened: tests: pre-cache snaps in classic and core systems <Simple 😃> <Created by sergiocazzolato> <https://github.com/snapcore/snapd/pull/10763>12:57
* pstolowski lunch13:34
jdstrandamurray: oh, heh, I forgot we discussed that, hehe. Yeah, any device is kernel attack surface. I can't recall a CVE in the joystick driver before this one, but I'm always like "but what if there was one?" :)13:43
* cachio afk14:05
mupPR snapd#10741 closed: interfaces/block-devices: support to access the state of block devices <Skip spread> <cherry-picked> <Created by woodrow-shen> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/10741>14:07
ijohnson[m]@bboozzoo hey I was wondering would it "simplify" things if we just imported github.com/etcd.io/bolt directly in our go.mod, then used the "replace" directive on specific distros that don't have a package for that module to redirect our import of that module to one that is installed locally via distro pkg? I don't recall the semantics around replace working with modules that declare themself as a differently named module14:13
mborzeckiijohnson: yeah, that could work, we may need to patch some things on fedora maybe?14:15
mborzeckiworth a try anyway14:15
ijohnson[m]yeah I was just wondering if that would be easier to add/remove that go.mod line than to mess around with the fork and things14:15
mborzeckiijohnson: also, despite having go.mod inside the repo, we can build without modules for now, tbh i don't see how go upstream will make building packages across various linux distros withouth either patching go.mod to update versions automatically or allowing people to disable modules altogether14:19
mborzeckiespecially as some distros are really adamant to vendoring random dependencies14:20
mborzeckiijohnson: (cc mvo) i've pushed a patch to https://github.com/snapcore/snapd/pull/10754 please take a look, once it lands LP builds will hopefully be fine again14:28
mupPR #10754: packaging, tests/lib/prepare-restore: build packages without network access, fix building debs with go modules <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/10754>14:28
ijohnson[m]@bbo14:28
ijohnson[m]gah14:29
mvomborzecki: nice, thank you14:29
ijohnson[m]@bboozzoo thanks will take a look14:29
ijohnson[m]I like fluffychat as a snap matrix client but it doesn't seem to do username autocomplete via the tab key which is unfortunate14:29
mborzeckiijohnson: i use element web interface, and sometimes gnome client called 'fractal'14:30
ijohnson[m]I used to use element but I wanted to try out something else, element works pretty well though14:31
ijohnson[m]I'll have to give fractal a try though14:31
ijohnson[m]ah nice well fractal does seem to do username autocomplete with tab key so that's nice, but fluffychat looks a bit nicer IMHO14:34
mardypstolowski: you are the servicestate expert, I guess: do you know what I should do in the unit tests, so that snapstate.Get() will return a snap? (it returns err here)14:37
pstolowskimardy: sure. snapstate.Set(...), you will find plenty of that in other unit tests14:38
pstolowskimardy: sometimes some tests have helpers that create snap.yaml on the disk *and* entry in the state, but often it needs to be done manually14:39
mardypstolowski: thanks, I'll try with snapstate.Set() first14:40
mupPR snapd#10758 closed: image/image_linux.go: add newline <Simple 😃> <Skip spread> <Created by anonymouse64> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/10758>14:43
ijohnson[m]@cachio hey is the snap run failure you were mentioning only happening on impish + debian sid or is it happening on other systems too?14:46
mupPR snapd#10742 closed: tests: update nested tool - part1 <Run nested> <Created by sergiocazzolato> <Merged by sergiocazzolato> <https://github.com/snapcore/snapd/pull/10742>15:13
ijohnson[m]@mvo can you land https://github.com/snapcore/snapd/pull/10732? the failures are unrelated15:21
mupPR #10732: tests/lib/nested.sh: split out additional helper for adding files to VM imgs <Simple 😃> <Run nested> <Created by anonymouse64> <https://github.com/snapcore/snapd/pull/10732>15:21
mvoijohnson[m]: sure15:21
ijohnson[m]thanks15:21
mupPR snapd#10732 closed: tests/lib/nested.sh: split out additional helper for adding files to VM imgs <Simple 😃> <Run nested> <Created by anonymouse64> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/10732>15:23
miguelpirescachio hey, can you help me with the translation linting spread test? https://github.com/snapcore/snapd/pull/10756/files#diff-5720a79fd79b88cebcc680e6914186981e330df7bc933261176bed1f42a1fb1d15:44
mupPR #10756: cmd/snap: only log translation warnings in debug/testing <Created by MiguelPires> <https://github.com/snapcore/snapd/pull/10756>15:44
miguelpiresI installed language-pack-pt to reproduce the issue so I guess that's what's missing from the spread env. But in the same test there's a test with "LANG=de", not sure why one works and the other doesn't15:48
ijohnson[m]/nick ijohnson|lunch16:43
ijohnson[m]well that didn't work16:44
ijohnson[m]/nick ijohnson|lunch16:45
ijohnson[m]hmm so only element seems to know about /nick it seems :-/16:45
ijohnson[m]huh so actually all this time, element was just changing the matrix display name in the matrix bridged room, that never actually propagated to IRC at all16:47
ograijohnson[m], so no lunch ofr you then ... 16:56
cachiomiguelpires, hi, sorry for the delay, I was having lunch16:57
cachiolet me take a look16:57
cachioijohnson[m], I saw that on debian sid and also in uc20 tests when I run beta validation16:58
cachiomiguelpires, so, in which system is it failing?17:02
cachiobecause I don'r see error in the execution logfs17:03
cachioijohnson[m], have you seen that one?17:08
cachiohttps://paste.ubuntu.com/p/9nJv3W46M7/17:08
miguelpirescachio no worries, thanks for helping me with this. It's not failing but I expected it to. It fails locally because I have the language pack with a incorrect translation and I assume the spread env doesn't have it. But the same spread test also expects assumes that LANG=de works, so I'm not sure17:11
cachiomiguelpires, what should can do there is to add "exit 1" before the line echo "Lint translated commands" 17:13
cachioand yo uexecute spread -deubg17:13
cachioso then you will have a debug session and you can run the line and see the output17:14
cachioI see that you are redirecting the output to /dev/null 17:15
cachiobut the best you can do there is to debug the line17:15
cachiodoes it make sense?17:15
miguelpiresYes, I'll try that then. Thanks! 17:17
cachioI would try without redirecting to /dev/null the output 17:24
cachioso NOMATH receives the input17:24
cachiomiguelpires, still don't know why it fails locally17:25
cachioperhaps could be because another error?17:25
ijohnson[m]@cachio I've only seen the snap-run failure on sid and impish, I haven't seen it on uc20, do you have logs of the uc20 failure ? 17:27
cachioijohnson[m], yes, one second17:27
ijohnson[m]@cac17:27
ijohnson[m]@cachio the issue with uc20-create-partitions is known it just randomly happens sometimes17:28
mupPR snapd#10763 closed: tests: pre-cache snaps in classic and core systems <Simple 😃> <Created by sergiocazzolato> <Merged by sergiocazzolato> <https://github.com/snapcore/snapd/pull/10763>17:28
cachioijohnson[m], https://paste.ubuntu.com/p/9jBtBNfnng/17:31
ijohnson[m]thanks looking now17:31
cachioit is a log from a rpi417:31
ijohnson[m]hmm I can't reproduce that in a uc20 amd64 VM17:36
ijohnson[m]I wonder if it's arm specific17:36
cachioijohnson[m],  let me check other logs17:37
cachioijohnson[m], in amd64 is working properly17:39
cachiospecifically in pi417:40
ijohnson[m]@cachio I can't reproduce that issue on my arm64 rpi418:03
ijohnson[m]https://pastebin.ubuntu.com/p/Kr8SgXpfNZ/18:04
ijohnson[m]granted that's edge snapd not beta18:04
ijohnson[m]let me try with beta channel snapd18:04
ijohnson[m]same onbeta channel too18:05
miguelpires@cachio I used -debug like you mentioned and there were only a couple of language-pack installed. After installing all, the linting worked and caught the bad translation so it's working as  Thanks =)18:12
cachiomiguelpires, nice18:29
cachiomiguelpires, so18:30
cachioto install dependencies you cant do apt install 18:30
cachiobecause that won't work on other oss than ubuntu or debian18:31
cachioyou need to add that here https://github.com/snapcore/snapd/blob/master/tests/lib/pkgdb.sh#L53718:32
cachioand also the fedora package in pkg_dependencies_fedora function 18:32
cachioand the same for amazon, opensuse and arch18:33
cachioso this will be installed when the test start18:33
cachioijohnson[m],  I use image from stable but update snapd to beta18:35
cachiorefresh18:35
ijohnson[m]hmm18:35
ijohnson[m]I'm not sure what image that was on18:35
ijohnson[m](for my run)18:35
cachiohttps://storage.googleapis.com/snapd-spread-tests/images/pi4-20-beta/pi.img.xz18:35
cachiothis is the image 18:36
cachiothis is where it fails18:36
cachiopi 10718:37
cachiocore20 108418:37
cachiopi-kernel 33918:37
cachiosnapd 1326918:37
ijohnson[m]cachio: oh wait my device was on uc18 not uc2018:37
cachiothis is the last image for 2.5218:37
ijohnson[m]I'll have to reflash with uc20 to test18:37
cachiook18:37
ijohnson[m]cachio: so is it reproducible just running that snap run command ?18:37
ijohnson[m]or is there more setup needed to make it fail with the segfault?18:38
cachioijohnson[m], just running -> snap run --strace="invalid" test-snapd-sh.sh -c 'echo hello'18:41
cachioso having that snpa installed18:42
cachioshould be enough18:42
ijohnson[m]ok, that's what I was trying but obviously I wasn't hitting the bug as I was actually on uc18 not uc2018:42
cachiohehe18:42
cachionp18:42
ijohnson[m]I'll try again with a uc20 image later, I need to work on 360s now18:43
mborzeckiheh, sid still failed in https://github.com/snapcore/snapd/pull/10754/ the package built successfuly but prepare failed anyway :/19:10
mupPR #10754: packaging, tests/lib/prepare-restore: build packages without network access, fix building debs with go modules <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/10754>19:10
mborzeckiehh it's because i disabled the generator, so somehow the package still ended up with unknown version, even though mkversion is ran earlier :/19:12
* cachio afk19:33
=== rzr is now known as rZr

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!