[01:32] <mup> PR snapcraft#2426 opened: tests: use fixed version for idna in plainbox <Created by sergiusens> <https://github.com/snapcore/snapcraft/pull/2426>
[05:16] <brlin> Good afternoon, from UTC+8
[05:54] <brlin> I noticed that `snapcraft init` adds a recipe that asserts `base: core18`, is that indicate that it is now recommended to use core18 as the base instead of core16?
[06:22] <mborzecki> morning
[06:22] <mvo> mborzecki: hey
[06:23] <mborzecki> mvo: can you take a look at https://github.com/snapcore/snapd/pull/6297 ?
[06:23] <mup> PR #6297: data/selinux: fix syntax error in definition of snappy_admin interface <SELinux> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/6297>
[06:23] <mvo> sure
[06:23] <mborzecki> mvo: we'll need to cherry pick that patch to 2.36
[06:24] <mvo> mborzecki: ok - should we run sepolgen-ifgen as part of the tests?
[06:25] <mborzecki> mvo: yes, that's a good idea, i'll add it to the selinux PRs
[06:25] <mvo> mborzecki: I merged this one now
[06:25] <mborzecki> mvo: thanks
[06:26] <mvo> and cherry-picked, so all good
[06:26] <mvo> thank you!
[06:26] <mborzecki> mvo: great, thanks!
[06:26] <mup> PR snapd#6297 closed: data/selinux: fix syntax error in definition of snappy_admin interface <SELinux> <Created by bboozzoo> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/6297>
[06:27] <mborzecki> and added a card :)
[06:37] <mborzecki> another trivial PR https://github.com/snapcore/snapd/pull/6298
[06:37] <mup> PR #6298: packaging/{fedora,opensuse}: own /var/lib/snapd/cookie <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/6298>
[06:38] <mup> PR snapd#6298 opened: packaging/{fedora,opensuse}: own /var/lib/snapd/cookie <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/6298>
[06:54] <zyga> Good morning
[06:54] <mborzecki> zyga: hey
[06:55] <mvo> hey zyga
[07:10] <zyga> :-)
[07:23] <zyga> sorry, a bit sleepy still
[07:23] <zyga> I stayed up late and fixed my TV
[07:24] <zyga> mborzecki: trivial removal of dead code: https://github.com/snapcore/snapd/pull/6293
[07:24] <mup> PR #6293: cmd/snap-confine: remove unused sc_discard_preserved_mount_ns <Per-user mount ns  🐎> <Created by zyga> <https://github.com/snapcore/snapd/pull/6293>
[07:25] <mborzecki> looking
[07:27] <mup> PR snapd#6295 closed: systemd: start snapd.autoimport.service in --no-block mode <⚠ Critical> <Created by mvo5> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/6295>
[07:32] <zyga> mvo: what's the state of the release?
[07:33] <zyga> ah
[07:33] <zyga> the gai_strerror thing
[07:36] <mvo> zyga: working on it
[07:36] <mvo> zyga: some bits have landed, but more bits needed
[07:36] <mborzecki> damn, ls -Zd has different output on centos and fedora
[07:36] <zyga> cannot allocate memory in dnf makecache? https://www.irccloud.com/pastebin/2TrCykuR/
[07:36] <zyga> mborzecki: what's the difference?
[07:37] <mborzecki> zyga: centos drwxr-xr-x. niemeyer google-sudoers unconfined_u:object_r:home_root_t:s0 .
[07:37] <mborzecki> fedora unconfined_u:object_r:user_home_t:s0 .
[07:42] <mvo> sil2100: hey, good morning! one final core18 pr (hopefully the last) https://github.com/snapcore/core18/pull/103
[07:42] <mup> PR core18#103: static: add missing systemd symlinks <Created by mvo5> <https://github.com/snapcore/core18/pull/103>
[07:52] <zyga> mvo: did you see the test failure there?
[07:52] <zyga> is that expected?
[07:55] <mvo> zyga: well, yes, I have not had time to look into it :( but tests are unhappy there
[07:55] <zyga> it doesn't seem related to the nature of the change in the pull request
[07:56] <mvo> zyga: exactly, looks like something innternal
[07:56] <mvo> zyga: so snapcraftctl iirc
=== pstolowski|afk is now known as pstolowski
[08:03] <pstolowski> hey
[08:11] <sil2100> mvo: morning! Looking in a minute
[08:12] <mvo> sil2100: thank you
[08:14] <zyga> hello Pawel
[08:28] <mup> PR snapd#6299 opened: travis: short cicruit failures in static and unit tests suites task <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/6299>
[08:28] <mborzecki> zyga: ^^
[08:29] <zyga> looking
[08:29] <pedronis> mvo: morning, sorry I'm a bit late,  what's the status of things?
[08:29] <zyga> thank you!
[08:29] <zyga> mborzecki: yes yes yes :)
[08:29] <pedronis> mvo: I saw some of the PR are landed now
[08:32] <mvo> pedronis: yeah, progress
[08:32] <mvo> pedronis: some things still pending though
[08:35] <mborzecki> heh, the code formatted with new gofmt is incorrectly formatted with the 1.6 one
[08:36] <zyga> mborzecki: yeah
[08:36] <zyga> I opened a PR to highlight that before
[08:36] <zyga> sucks :/
[08:36] <mborzecki> i'm thinking, maybe we could go get gofmt (the tool) with 1.6 instead
[08:37] <pedronis> mvo: are you waiting for me to look at #6243 ?
[08:37] <mup> PR #6243: systemd: allow only a single daemon-reload at the same time <Created by mvo5> <https://github.com/snapcore/snapd/pull/6243>
[08:37] <zyga> mborzecki: hmm?
[08:38] <mborzecki> zyga: go get github.com/golang/go/src/cmd/gofmt, this would pull the latest one
[08:38] <zyga> aha
[08:38] <zyga> I see
[08:38] <zyga> as long as people on older releases are happy to integrate this with their workflow, yeah
[08:38] <mvo> pedronis: its a bit of a target of opportunity - if you think its too much we can just leave it for later
[08:38] <mvo> pedronis: it seemed its now relatively clean and the win is relatively high
[08:38] <pedronis> mvo: my main comment there, thinking a bit further, is that the description is misleading, we do daemon-reload in other places?
[08:39] <pedronis> mvo: if to fix it we really need to serialize daemon-reload then that is not enough, no?
[08:39] <mvo> pedronis: let me see, I think you are right
[08:39] <pedronis> so it would need more work, and doesn't feel a .3 thing
[08:40] <mvo> pedronis: hm, in the overlord we really only do the daemon reload there afaict - but yeah, lets push it out
[08:40] <pedronis> let me double check
[08:40] <pedronis> mvo: we do reloads in wrappers which is used indirectly by overload for example
[08:40] <pedronis> sorry, overlord
[08:41] <mvo> pedronis: aha, its the only explicit place. indeed
[08:41] <pedronis> also interfaces systemd backend does it
[08:42] <pedronis> let me put this chat in the PR
[08:43] <mvo> pedronis: thank you, yeah - we need to investiagte, serializing all daemon reloads would be a bit painful but I'm not into the bug enough to know
[08:44] <pedronis> mvo: done, thanks for marking as blocked, yea, needs more thinking I fear
[08:47] <pedronis> mborzecki: so you were right and it seems on arch we end up with a snapd that uses the cgo resolver,  I thought we had reasons for not wanting cgo at all, do you know if can stop arch using that? not an immediate thing
[08:47] <pedronis> mvo: ^
[08:47] <zyga> cross distribution work is fun
[08:47] <zyga> you get to see *all* the buys
[08:47] <zyga> bugs
[08:50] <pedronis> mvo: so we are left with 6287
[08:51] <mvo> pedronis: there is one more that I'm working on
[08:52] <mvo> pedronis: we have snapd.seeded.service now on core18 at startup. this has a "Requires=snapd.socket" which means that when "snapd.socket" gets stopped (even if its not active) snapd.seeded.service is also stopped
[08:52] <mvo> pedronis: during the seeding the code so far naively restarts things (including snapd.socket) now it needs to be smart and only restart when snapd.socket is actually active
[08:54] <pedronis> mvo: I'm not following
[08:54] <mvo> pedronis: sorry, let me be less terse :)
[08:54] <pedronis> what breaks? :)
[08:54] <mborzecki> pedronis: hm maybe there's a build tag we could use, other than that we can use GODEBUG= switch
[08:55] <sil2100> mvo: approved if anything!
[08:55] <mvo> pedronis: we have the snapd.seeded.service. it has a "Requires=snapd.socket". this means in systemd terms that the when snapd.seeded starts snapd.socket is started by snapd. the converse is also true - systemctl stop snapd.socket will also stop snapd.seeded even if snapd.socket was not active before
[08:56] <mvo> pedronis: what breaks is that during the seeding boot snapd.seeded stops and that will (AIUI) make systemd start units with the after=snapd.seeded ordering
[08:56] <sil2100> mvo: will you guys perform some validation testing on the core18 snap now?
[08:56] <mborzecki> pedronis: hm i think we could also do that in the code, set a preferred resolved on a dialer
[08:56] <mvo> pedronis: but at this point its too early, the seeding is not done yet, we just got our snapd.socket unit and that is started now (via a restart which is the problem)
[08:56] <mvo> sil2100: yeah, once all the pieces are in cachio will run tests
[08:57] <sil2100> (still no merge-powers sadly)
[08:57] <mvo> sil2100: we also get tests from cert (cwayne team)
[08:59] <sil2100> mvo: yeah, I know the automation from certification, but I also saw some tests coming from cachio last time
[09:02] <pedronis> mvo: there is no snapd.socket in core18 itself?
[09:02] <pedronis> we write it from snapd?
[09:02] <zyga> pedronis: correct
[09:03] <mvo> pedronis: correct
[09:04] <pedronis> mvo: so we should know what to do, no? we are just being naive?
[09:04] <pedronis> I'm still missing if there's a deep problem or a shallow one
[09:04] <pedronis> (I suppose that's what systemd does to you)
[09:05] <zyga> pedronis: I think we could look at the way snapd stats up a little so that the relevant code is in our primary repository and so that core18 just consumes that
[09:05] <mvo> pedronis: we are just naive
[09:05] <zyga> pedronis: and to spend some more time on making things more robust and better tested
[09:06] <zyga> pedronis: but in general things feel okay
[09:06] <pedronis> zyga: ?
[09:06] <pedronis> it's already mostly the case
[09:06] <pedronis> there is little code in core18
[09:06] <pedronis> about this
[09:07] <zyga> pedronis: yes but it in core18 repo, it will need to be in core20 repo and in other repos eventually, we could just have core18 and core16 eventually consume a deb that is built out of snapd source package
[09:07] <pedronis> it's mostly the baggage of systemd unit keywords (they often do a bit more or less than one would exactly like)
[09:07] <pedronis> zyga: sounds like a code org issue
[09:07] <pedronis> not a what the code does issue
[09:08] <zyga> yes
[09:08] <zyga> I said we're mostly okay
[09:08] <pedronis> anyway it seems very premature before we got our v0 working
[09:13] <mup> PR snapd#6300 opened: travis: use more recent version of gofmt <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/6300>
[09:19] <mup> PR snapd#6298 closed: packaging/{fedora,opensuse}: own /var/lib/snapd/cookie <Simple 😃> <Created by bboozzoo> <Merged by bboozzoo> <https://github.com/snapcore/snapd/pull/6298>
[09:19] <mup> PR snapd#6301 opened: wrappers: only restart service in core18 when they are active <⚠ Critical> <Created by mvo5> <https://github.com/snapcore/snapd/pull/6301>
[09:22] <mvo> mborzecki: quick question for you - the systemd mount unit bug - we were wondering if the workaround must serialize all daemon-reloads. or if its enough to ensure that there is only a single mount unit started (and we wait until that start is finished) at the same time. i.e. however many daemon-reloads are ok as long as there is just a single mountunit starting. this question affects the nature of the workaround. if its just a single mountunit at
[09:22] <mvo>  the same time the workaround I pushed is probably sufficient. if we need to serialize all dameon-reload things are more complicated. do you have more insight into this?
[09:22] <mvo> (uh, that was long, sorry!)
[09:24] <mborzecki> mvo: i don't have more insight, i think i'd play it safe and serialize reloads & mounts
[09:24] <mvo> pedronis: fwiw 6301 is up and has the bits I mentioned
[09:24] <mvo> mborzecki: ok, so when we do the mountunit generation / daemon-reload /start nothing no other daemon-reloads?
[09:25] <mvo> mborzecki: that will be interessting code-wise :)
[09:25] <mup> PR snapd#6268 closed: overlord/ifacestate: helpers for serializing hotplug changes <Hotplug 🔌> <Created by stolowski> <Merged by stolowski> <https://github.com/snapcore/snapd/pull/6268>
[09:25] <mborzecki> mvo: yes, that'd the the super safe variant
[09:25] <mborzecki> mvo: otoh, your pr only serialized the mounts and seemed to work fine
[09:26] <mvo> mborzecki: yeah, this is why I was wondering
[09:26] <mborzecki> mvo: so we could start small, it's unlinkely we'd introduce a regression
[09:26] <mvo> mborzecki: if your script could be extended to springle random daemon-reloads but only a single mount at the same time and it is fine we might get away with the simpler version
[09:27] <mvo> mborzecki: fair point, feel free to report in the bug
[09:27] <mvo> mborzecki: eh, PR
[09:27] <mborzecki> mvo: i'll update the script and let you know
[09:27] <mvo> mborzecki: \o/
[09:27] <pedronis> mvo: so stopping something that is not running does something ?
[09:27] <pedronis> that's probably the bit I don't understand
[09:27] <zyga> pedronis: yes, stops related stuff AFAIK
[09:28] <mvo> pedronis: coreect
[09:33] <pedronis> mvo: I'm working on 6287
[09:33] <pedronis> btw
[09:33] <mvo> pedronis: ta
[09:37] <mup> PR snapd#6300 closed: travis: use more recent version of gofmt <Created by bboozzoo> <Closed by bboozzoo> <https://github.com/snapcore/snapd/pull/6300>
[09:46] <mborzecki> mvo: bad news, single mount & many reloads fails the same way
[09:47] <mvo> mborzecki: thanks, there was hope for a moment :)
[09:48] <mvo> mborzecki: its fine, makes things more complicated but at least we know now. thanks for checking
[09:48] <mborzecki> mvo: i've updated the script https://gist.github.com/bboozzoo/d4b142229b1915ef7cc0cf8593599ad9 in case yout want to double check the code
[09:48] <pedronis> ok, so that PR is indeed not enough
[09:48] <mvo> mborzecki: my brain is too mushy today :)
[09:48] <mvo> pedronis: yeah, slightly sad
[09:49] <mvo> mborzecki: but thanks for adding it, will probably be helpful in the future
[09:51] <zyga> mborzecki: nice work
[09:51] <mborzecki> mvo: fwiw, it's harder to reproduce now
[09:53] <mborzecki> mvo: i think i'd start with a simpler fix
[09:56] <mvo> mborzecki: thanks, yeah, it won't hurt and make things a bit better. we just need to make sure to document that its not the full fix
[09:56] <mborzecki> mhm
[10:02] <pstolowski> pedronis: do you have some time for a HO today?
[10:03] <pedronis> pstolowski: yes
[10:04] <pstolowski> pedronis: standup ho?
[10:05] <pedronis> pstolowski: yea, but I need 5 mins
[10:05] <pstolowski> pedronis: ok
[10:09] <mup> PR snapd#6301 closed: wrappers: only restart service in core18 when they are active <⚠ Critical> <Created by mvo5> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/6301>
[10:11] <mup> PR snapd#6302 opened: wrapprs: address review feedback from #6301 <Created by mvo5> <https://github.com/snapcore/snapd/pull/6302>
[10:12] <mup> PR core18#103 closed: static: add missing systemd symlinks <Created by mvo5> <Merged by mvo5> <https://github.com/snapcore/core18/pull/103>
[10:12] <pedronis> pstolowski: now works
[10:13] <pstolowski> ok
[10:30] <pedronis> mvo: I pushed changes/fixes to #6287
[10:30] <mup> PR snapd#6303 opened: cmd: add tests for lintArg and lintDesc <Simple 😃> <Created by zyga> <https://github.com/snapcore/snapd/pull/6303>
[10:30] <mup> PR #6287: httputil: retry on temporary net errors <Squash-merge> <Created by mvo5> <https://github.com/snapcore/snapd/pull/6287>
[10:30] <zyga> pstolowski: trivial PR with some unit tests for internal functions https://github.com/snapcore/snapd/pull/6303/files
[10:30] <mup> PR #6303: cmd: add tests for lintArg and lintDesc <Simple 😃> <Created by zyga> <https://github.com/snapcore/snapd/pull/6303>
[10:30] <pedronis> zyga: mborzecki: 6287 needs 2nd reviews btw
[10:31] <pstolowski> zyga: ok
[10:31] <Chipaca> morning peeps, sorry i'm late
[10:31] <pedronis> Chipaca: hello
[10:31] <mborzecki> pedronis: mvo: btw. do we take any steps to use go resolver on ubuntu?
[10:32] <Chipaca> zyga: wrt 6303, what were the l10n issues?
[10:37] <zyga> pedronis: ack
[10:37] <zyga> Chipaca: https://bugs.launchpad.net/snapd/+bug/1806761
[10:37] <mup> Bug #1806761: Trash output for LANG=ru_RU.UTF-8 (probably all non-ASCII languages) <snapd:In Progress by zyga> <snapd (Ubuntu):Confirmed> <https://launchpad.net/bugs/1806761>
[10:37] <mvo> mborzecki: iirc we use CGO_ENABLED=0 during the build
[10:38] <zyga> Chipaca: I think someone misunderstood the translator comment
[10:38] <mvo> mborzecki: aha, no, wait - just for selected bits like snap-exec
[10:38] <zyga> and added <foo>s with the s at the end :)
[10:39] <zyga> Chipaca: so my idea is
[10:39] <Chipaca> zyga: brb, need to order a bigger palm on amazon, for the facepalm i need to do
[10:39] <zyga> Chipaca: if the format is "<.*>s", cut the s and not log anything
[10:39] <zyga> Chipaca: right? :D
[10:39] <zyga> Chipaca: and fix the translator suggestions to make sure we say "don't add the frelling s"
[10:42] <mborzecki> mvo: pedronis: looking at net/conf.go, i think go gets confused with 'mymachines' listed as one of the hosts sources on arch
[10:42] <mvo> mborzecki: aha, it inspects nsswitch.conf ?
[10:42] <mborzecki> mvo: yes
[10:42] <mvo> mborzecki: smart!
[10:43] <mborzecki> mvo: it's parsing both nsswitch.cofn and resolv.conf, whenever it finds an option/source that's unknown it'll default to libc resolver first
[10:43]  * mvo nods
[10:44] <Chipaca> zyga: argument %q's %q should begin with < and end with >
[10:44] <Chipaca> zyga: ?
[10:44] <mborzecki> mvo: there's libnss_mymachines.so.2 in my system, owned by libsystemd pacakge
[10:45] <mborzecki> mvo: on a system which uses resolved it'll probably default to libc resolver too
[10:47] <zyga> Chipaca: sounds good
[10:47] <zyga> Chipaca: are you still off today?
[10:48] <Chipaca> no
[10:48] <zyga> mborzecki: yes
[10:48] <zyga> Chipaca: ah, splendid
[10:48] <zyga> Chipaca: shall we land https://github.com/snapcore/snapd/pull/6303?
[10:48] <mup> PR #6303: cmd: add tests for lintArg and lintDesc <Simple 😃> <Created by zyga> <https://github.com/snapcore/snapd/pull/6303>
[10:48] <Chipaca> but i didn't need to take the boys to school today so i turned off my 6am alarm … and then it was suddenly 10:15
[10:48] <Chipaca> zyga: not without a second review, I'm still smarting from a few others we slipped in too quickly
[10:51] <zyga> mvo: some comments on https://github.com/snapcore/snapd/pull/6287
[10:51] <mup> PR #6287: httputil: retry on temporary net errors <Squash-merge> <Created by mvo5> <https://github.com/snapcore/snapd/pull/6287>
[10:55] <mvo> zyga: thanks
[10:58] <mborzecki> Chipaca: maybe something simple to start with https://github.com/snapcore/snapd/pull/6299 ?
[10:58] <mup> PR #6299: travis: short cicruit failures in static and unit tests suites task <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/6299>