[02:28] PR snapd#5448 opened: tests: start using the new opensuse image with test dependencies [05:13] morning [07:02] mornings [07:04] pstolowski: hey [07:05] Hey ho [07:05] A bit late today, [07:06] mvo: I went on a small detour to simplify some code and add tests [07:06] I have some misc branches and then more meat [07:07] (Rice, soy, ...) [07:07] zyga: hey, thanks [07:08] I was working towards testing all of the patches properly [07:08] Out of the interface WIP one [07:09] zyga: neat [07:10] mvo: zyga: hey guys [07:12] hey mborzecki [07:13] mvo: i'm looking at sysctl manpage, have you checked that sysctl -p with as separate arg, not -p works? [07:14] mborzecki: the test will check, I have not, let me try that now [07:14] i have a vague recollection of something being iffy there, but that's from a few years back and with busybox [07:17] mborzecki: I did a quick test and it seems to be fine [07:17] mvo: ok :) [07:17] mborzecki: but the test will also fail if it is not [07:26] Can you check the confinement of a snap before you install it? I can't see anything in "snap info x" or another obvious snap command. [07:26] Gargoyle: try snap info --verbose [07:27] Cool. Ta. :-) [07:28] PR snapd#5447 closed: snap,interfaces: move interface name validation to snap [07:36] mvo: could you take a look at #5434? [07:36] PR #5434: overlord: introduce InstanceKey to SnapState and SnapSetup, renames [07:52] mborzecki: sure [07:52] mvo: ta === matteo| is now known as matteo [08:02] mvo: thanks for that! [08:06] PR snapd#5449 opened: snap: add helper for renaming slots [08:07] mvo: that's one of the helpers ^ [08:07] pedronis: could you please look at https://github.com/snapcore/snapd/pull/5443 - it has two reviews already but I wanted to ensure you ack it [08:07] PR #5443: interfaces: treat "snapd" snap as type:os [08:11] mvo: quick review on https://github.com/snapcore/snapd/pull/5446 [08:11] PR #5446: coreconfig: add support for `snap set core network.disable-ipv6` [08:15] PR snapd#5445 closed: tests: add artful for sru validation on google backend [08:16] zyga: aha, nice. yeah, initially I wanted to do ipv4/ipv6 but then a) YAGNI b) ipv4 is not as straightforward === chihchun_afk is now known as chihchun [08:19] * Chipaca waves [08:20] Chipaca: hey, now that you're here, would you mind taking a look at #5426? :) [08:20] PR #5426: client, cmd/snap: pass snap instance name when installing from file [08:20] mborzecki: no [08:21] mvo: it's fine not to do ipv4 [08:21] Discord have both beta and "bleeding edge" versions available, is it possible to get these wired up into the snap? [08:22] Gargoyle: depending on who's doing the snapping, those can be tracks or just the pre-existing risks [08:25] Not sure what that means! :/ [08:25] Gargoyle: if the people making the snap want to, they can do that with snaps [08:26] Gargoyle: they can offer a beta channel/track [08:26] Gargoyle: and an "edge" channel/track [08:26] Gargoyle: alongside stable, that is [08:26] and people installing the snap can choose which one to get [08:33] Gargoyle: you can look at the 'go' snap for (a rather extreme) example [08:34] Gargoyle: you can get anything from 1.6 to devel, in a snap === chihchun is now known as chihchun_afk [08:36] Chipaca: and by look we mean "snap info go" [08:36] zyga: or gnome software, or https://snapcraft.io/go (and click on 'all versions') [08:37] That's a lot of options! :D [08:37] Chipaca: I wonder if we will ever need to do paging on "snap info" ;-) [08:38] Chipaca: hum [08:38] Chipaca: why is 1.10 above 1.11? [08:38] mborzecki: I did a pass over #5434 [08:38] PR #5434: overlord: introduce InstanceKey to SnapState and SnapSetup, renames [08:38] Chipaca: do we need number sorting there? [08:38] pedronis: thanks! [08:38] (actually the ordering of tracks is a bit weird in general) [08:38] zyga: there is no order to tracks yet [08:39] zyga: we don't know how to sort them, last time we had a discussion with Chipaca and other store people, we need a sort order defined in the store [08:39] zyga: there probably will be at some point, but that point is still in the future :-) AFAIK at least [08:39] Chipaca: should we version sort them "just in case" [08:39] the result we get from the store are sorted [08:39] aha [08:39] I see [08:39] well [08:39] except they are lexically sorted atm [08:39] they are _ordered_, which is more important [08:39] re-sorted by us or by the store? [08:39] :) [08:39] no, that's the point [08:40] if snapd is sorting them, it shouldn't be [08:40] ok [08:40] snapd gets them with a defined order and should leave it alone [08:40] there's a store order (is just not a good one) [08:40] but we should't try to fix that [08:40] it's up to the store [08:40] (to answer whether snapd sorts them i'd need to look at the code) [08:40] agreed [08:40] OTOH, snapcraft.io and 'snap info' show a different order [08:40] so that's a separate issue i'm chasing down [08:41] (it just so happens to be a nicer order! grrr) [08:41] (for go, on snapcraft.io) [08:42] pedronis: yeah, the non-test code setting InstanceKey in either snapsup or snapst lives here atm https://github.com/bboozzoo/snapd/commits/bboozzoo/parallel-install-overlord === chihchun_afk is now known as chihchun [08:50] pedronis: ETOOMANYCHANNELS [08:50] er [08:50] mborzecki: ^ [08:50] :) [08:51] heh [08:54] pretty please https://github.com/snapcore/snapd/pull/5449 :) [08:54] PR #5449: snap: add helper for renaming slots [08:54] it's short and test code [08:59] mvo: thank you for the update, two more comments on 5446 [08:59] * zyga coffee [09:11] PR snapd#5450 opened: store: keep all files with link-count > 1 in the cache [09:18] pedronis: pushed a fix to #5434, i've tweaked the error message a little, added some tests [09:18] PR #5434: overlord: introduce InstanceKey to SnapState and SnapSetup, renames [09:27] mvo: one last comment on 5446, [09:29] dog emergency, need to go to vet === pstolowski is now known as pstolowski|vet [09:30] zyga: thanks for your comment. this is a bit tricky, if things get removed sysctl -p won't work. on removal the system default is used. we can make this explicit but it has the side-effect that there will always be a sysctl file created even with an empty config [09:32] PR snapd#5451 opened: interfaces: honor static attributes when reloading conns [09:37] pstolowski|vet: interesting, reading [09:40] zyga: I commented, I can rework the code accordingly [09:40] zyga: I think it makes sense, will do so in a wee bit [09:40] thank you, looking [09:43] mvo: I commented as well, let me know where you want to take this [09:45] mborzecki: replied on your comment on RenameSlot [09:47] mvo, not sure you have noticed, ppc64el and s390x daily core snap builds failed (looks like a gpg error) [09:51] zyga: thanks, I updated the code now, I think we are fine with EnsureDirState and a bit of extra smarts, I pushed a new revision, please let me know if I overlooked anything but I think we should be fine now [09:52] ogra_: you mean the builds in https://launchpad.net/~snappy-dev/+archive/ubuntu/edge/+packages?field.name_filter=snapd&field.status_filter=&field.series_filter= ? [09:53] mvo, https://launchpad.net/~snappy-dev/+snap/core/+build/267041 and https://launchpad.net/~snappy-dev/+snap/core/+build/267042 [09:53] ogra_: they are flaky on these arches but not consistently failing? or am I looking at the wrong place maybe? [09:53] ogra_: aha, the core snap, let me check [09:53] it is the first build error i have seen in a while [09:53] * mvo looks [09:53] noit flaky usually ... [09:53] ogra_,mvo: Those are transient failures due to known keyserver network chaos [09:53] *not [09:53] just retry [09:54] ah, great [09:54] thanks cjwatson [09:54] yeah, i noticed a ton of livefs mails too [09:54] AIUI somebody discovered that you could upload a gigabyte key to the keyserver network and have it kill stuff [09:54] or something along those lines [09:54] lol [09:54] *cough* [09:54] the livefs stuff is different and I'm investigating [09:54] ah, k [09:56] PR snapd#5428 closed: devicestate: fix panic in firstboot code when no snaps are seeded [10:02] mvo: looking [10:06] mvo: it is good [10:06] mvo: I suggested one more tweak to make it more clear why this happens [10:06] but logic wise it is correct [10:06] smart way of doing that btw :) [10:08] cjwatson: so gpg network is the new bittorrent? [10:08] hopefully not ... [10:19] zyga: sure thing, thank you [10:20] niemeyer: any chance we can bump the largest post size from 32000 to ~40000 on discourse. I'm trying to migrate a large doc to the site, but it's ~35K [10:22] Heya [10:22] Which doc is that? [10:23] https://github.com/snapcore/snapd/wiki/REST-API [10:23] it's the last one still on the wiki [10:24] well, last one linked on the front page of the wiki [10:24] popey: I think we should break it down a bit on the way in, even so we can read it reasonably [10:25] popey: IIRC I provided an outline of how to break it down in the thread where we discuss the doc site [10:25] that seems reasonable. [10:26] popey: We don't need to break it down every single section, but rather by area of interest or something similar [10:26] Robert seems to "own" that doc. I'd rather he broke it up and kept owning it [10:26] popey: Thanks for the work on that, btw [10:27] I'll drop Robert a mail as he's in another TZ [10:27] thanks [10:31] popey: np, and if you want I'd be happy to work with you on that migration.. it's been pending for too long [10:50] mborzecki, mvo: can you please look at https://github.com/snapcore/snapd/pull/5449 [10:50] PR #5449: snap: add helper for renaming slots [10:50] I have a pair of follow ups on top [10:52] hmm [10:52] journalctl fails to restart https://www.irccloud.com/pastebin/w7dISItk/ [10:52] have you guys seen that before ^ [10:53] journalctl failing to _restart_ ? [10:57] * zyga -> walk [10:57] zyga: yes, seen it before [11:20] mvo: Heya [11:20] mvo: One question on spread#61 [11:20] PR spread#61: client: fix dialOnReboot() if the remote does not reply [11:22] parallel installs will be fun to merge/resolve conflicts :/ [11:46] PR snapd#5452 opened: store, overlord/snapstate: introduce instance name in store APIs [11:46] pedronis: Chipaca: probably something for you guys ^^ [11:47] it's very likely this will conflict with changes i wanted to open for review after landing #5434 [11:47] PR #5434: overlord: introduce InstanceKey to SnapState and SnapSetup, renames [11:50] pedronis: you've seen part of the store changes 2-3 weeks ago, most of it is the same patches just updated to account for downloads + some extra adjustments [11:50] re [11:51] MATCH mystery again https://www.irccloud.com/pastebin/p8HCgyXK/ [11:53] mborzecki: well, I should probably land my other PR and then it will likely conflict [11:54] pedronis: no worries, i'll keep resolving the conflicts as they come [12:24] mvo: can you please review https://github.com/snapcore/snapd/pull/5449 quickly [12:24] PR #5449: snap: add helper for renaming slots [12:26] Chipaca: did you manage to come up with anything regading 'the mystery of MATCH' yesterday? [12:26] mborzecki: no, i didn't carry on digging [12:27] falling too far behind with warnings [12:27] ack [12:30] thank you mborzecki :) [12:30] * zyga looks at the diff from the upcoming branch and is very happy [12:33] PR snapcraft#2175 opened: ci: disable osx tests until a new pyyaml is released [12:34] zyga: sure, I have a look [12:35] zyga: match mystery> I think your idea to hexdump the output we got is interessting, maybe there is some whitespace mangling going on [12:36] mvo: travis hides certain characters in the output log, though i'm not sure if the same applies to the raw log [12:41] mborzecki: mvo: I did 'od -a' the file fwiw [12:41] and it was fine [12:41] (i thought i'd told you this yesterday0 [12:41] ) [12:43] why haven't i had lunch yet? [12:44] also, where has my morning gone [12:44] * Chipaca scrambles to get lunch [12:45] mvo: I honestly don't believe my theory anymore [12:45] mvo: I bet I will be surprised by the _actual_ issue [12:57] PR snapd#5449 closed: snap: add helper for renaming slots [13:01] joining [13:03] PR snapd#5453 opened: interfaces: tweak tests to have less repetition of "core" and "ubuntu… [13:06] mvo: ^ the small refactor for the disconnect/connect resolve with snapd awareness [13:06] the patch on top of this one is the main part of the WIP branch that was still pending [13:06] so very close with getting all of that out [13:06] I will prepare a branch that is derived from your integration branch where I have the new patch set (from master) and see how things are going [13:07] I didn't manage that last night, sorry [13:12] cachio: the Software crash, is that in 18.04? [13:12] Chipaca: artful iirc [13:25] zyga: hmm https://paste.ubuntu.com/p/CNqt2p6gj8/ [13:26] oh [13:26] weird [13:26] ah, sorry [13:26] I didn't read this righ [13:26] *right [13:26] this is a bug [13:27] caused by >1 instance of "change to this directory and then back" [13:27] without handling the error by going to the void [13:27] heh lucky me [13:27] is this new? [13:27] zyga: why is this a bug? [13:28] it's in /tmp [13:28] zyga: dk, tried to figure out why shellcheck in tests/unit/go does not work [13:28] Chipaca: because the correct behavior was (and it worked before) to go to /var/lib/snapd/void [13:28] zyga: ahhh [13:28] zyga: maybe /var/lib/snapd/void doesn't exist? [13:28] Chipaca: yes but the /tmp inside the snap is different and cannot contain what is on the host (it's a private tmp) [13:28] Chipaca: oh, that's interesting [13:28] does it mborzecki / [13:29] zyga: it does drwxr-xr-x 2 root root 4096 Jul 3 12:36 void [13:29] try it locally [13:29] the permissions are wrong [13:29] anyway, that's something to add to the queue [13:29] probably a fallout of something that happened recently [13:30] running it outside of tmp at least once makes the problem go away [13:30] this implies the problem is in the first part of the setup code [13:32] mvo: https://github.com/snapcore/snapd/pull/5446#issuecomment-402155147 [13:32] PR #5446: coreconfig: add support for `snap set core network.disable-ipv6` === pstolowski|vet is now known as pstolowski [13:34] zyga: yeah, just noticed and looking [13:36] zyga: fixed, it was just that the output of sysctl has some extra whitespace [13:37] zyga: did you tag all PRs that help with the iface renames as core18? [13:37] zyga: if so, I will merge and run them now against core18 [13:38] no, not yet, hold on [13:39] I didn't open all of them due to stacking [13:40] zyga: https://paste.fedoraproject.org/paste/4hQSsupdzQNcvvbwpZlhUA makes it go away [13:40] Chipaca, it is on 17.10 [13:40] just finished installing it [13:40] (because why not) [13:42] zyga: or https://paste.ubuntu.com/p/yWyRj3Pqqj/ no clue why fpaste is so slow today [13:42] cachio: and you start 'ubuntu software', and search for 'snapd', and the UI crashes? [13:43] zyga: why does it even bother chdir()ing in setup_private_mount() while it done so in sc_populate_mount_ns() already [13:46] zyga: no worries, I think I found the relevant ones [13:46] I haven't opened all of them [13:46] I actually wanted to ask you [13:46] well, tell you [13:46] zyga: aha ok [13:47] I'll create a branch from your working one but replace the older versions of the patches as I said in the standup [13:47] it's fine to test what you made, let's where it gets [13:47] but I want to have a clean picture of the patches that are not in master [13:47] (a list that shrinks rather than grows0 [13:50] zyga: ok, it might be easier to use core18-all-fixes-all-tests-2 [13:50] zyga: which is a clean master with just all the core18 PRs [13:50] sure, I'll use that [13:50] cachio: can #5391 land? [13:50] nice [13:50] zyga: and it also enables all core18 tests [13:50] PR #5391: tests: simplify econnreset test [13:53] pstolowski, yes [13:53] thanks [13:54] PR snapd#5391 closed: tests: simplify econnreset test [13:57] pedronis: thank you for the comments! I'm on that now [13:58] mborzecki: we need to rework more code I'm afraid, to get the right "way" of handling that error [14:00] zyga: i can open a PR with http://paste.ubuntu.com/p/XvBhhFSMk8/ which resolves the issue, the question is why it tries to chdir into the old cwd while doing that, since it's already being restored at the end [14:01] pedronis: zyga: core18 meeting [14:01] mborzecki: later [14:01] ack [14:07] PR snapd#5369 closed: overlord,interfaces,cmd: WIP early support for interfaces on core18 [14:26] PR snapd#5453 closed: interfaces: tweak tests to have less repetition of "core" and "ubuntu… [14:26] mvo: thank you for the review, one more for implicit operations this time ^ [14:26] well v [14:27] 5454 [14:27] PR snapd#5454 opened: interfaces: prefer "snapd" when resolving implicit connections [14:30] cachio: how do I reproduce the crash? [14:31] just open ubuntu software [14:31] cachio: done [14:31] and search "snapd" [14:31] cachio: done [14:31] cachio: … and the crash? [14:32] zyga: thanks, looking [14:33] Chipaca, did it return results? [14:33] for me, it closed the app [14:33] zyga: also, please share the run results if you run the full tests with your PRs against core18, maybe there is some stuff that I need to tweak [14:33] cachio: yes [14:33] cachio: with what version of snapd? [14:33] try again [14:33] I try with the 2.38.* [14:33] and with 2.33.1 [14:33] 2.28 sorry [14:34] cachio: this is snapd from the package, yes? [14:34] yes [14:34] cachio: I get 2.32.9+17.10 from the archive [14:34] cachio: where do i get 2.28 from? [14:34] mvo: yes, I will share the results, I need to spend some time to prepare the branch [14:35] 1 sec [14:35] cachio: ah, i can get 2.28.5 from artful/main (as opposed to artful-updates/main) [14:35] I have 2.28.5+17.10 [14:36] cachio: yes, taht's the one in artful/main (you've not updated) [14:36] mvo, zyga i recently see this (not sure if it was there before) when starting a core 16 edge image on my pi's https://paste.ubuntu.com/p/gGKBBSG4nB/ ... the system seems to behave normally but it prints in red during boot so it is pretty noticeable [14:36] cachio: apt update && apt upgrade should get you 2.32.9 [14:37] cachio: in any case, it worked ok with both [14:37] cachio: also with core from edge [14:37] so I don't know what I'm looking for [14:38] Chipaca, it is failing here [14:38] perheps it is a different version of ubuntu software [14:39] did you search for something else and then "snapd" [14:39] ? [14:39] Chipaca, I just got distro_install_package google-compute-engine [14:39] sorry, seg fault [14:40] ogra_: that's very interesting [14:40] can you provide journal data from that service? [14:40] Chipaca, [ 173.948719] pool[1815]: segfault at 0 ip 00007fdea83ec1f0 sp 00007fde6effc7e8 error 4 in libglib-2.0.so.0.5400.1[7fdea83b2000+111000] [14:41] Chipaca, gnome-software 3.26.1 [14:41] ogra_: thank you for reporting this [14:42] PR snapd#5455 opened: many: assorted shellcheck fixes [14:42] cachio: how much memory in your vm? [14:43] 2GB [14:43] cachio: can you try with 4? [14:43] sure [14:43] cachio: gnome-software is that same version here [14:45] cachio: it doesn't crash here with 2 either though [14:46] Chipaca, weird, I'll update snapd and retry [14:48]