[01:34] #blackarch === chihchun_afk is now known as chihchun [08:01] good morning [08:01] Pharaoh_Atem: looking :/ [08:01] PR snapd#3248 closed: tests: fail early in the spread suite if trying to run it inside a container [08:01] aha, not great [08:01] Pharaoh_Atem: I'll fix that in a sec, thanks for reporting [08:02] hey zyga_ ! [08:02] pstolowski: hello [08:05] Pharaoh_Atem: offtipic, can I propose rawhide packaging for inclusion into packaging/ [08:05] Pharaoh_Atem: it will be useful for CI soon === zyga_ is now known as zyga [08:15] zyga: we can't yet [08:15] morphis: because of patches/ [08:15] zyga: we still miss changes to get snap-confine building [08:15] morphis: we can add the actual patches too you know [08:15] zyga: but I have things locally and will submit PRs soon [08:15] aha ok [08:15] * zyga pushed the fix for shellcheck [08:15] now where's my github token :/ [08:16] unless we have initial CI I would like to held the .spec file back [08:16] as only then it has value inside the tree [08:17] zyga: but if you have a good idea of how we can solve the problem http://pkgs.fedoraproject.org/cgit/rpms/snapd.git/tree/0001-cmd-use-libtool-for-the-internal-library.patch solves, that would be awesome :-) [08:19] Pharaoh_Atem: https://github.com/snapcore/snapd/pull/3258 [08:19] PR snapd#3258: cmd/snap-confine/tests: fix shellcheck on recently added files [08:19] PR snapd#3258 opened: cmd/snap-confine/tests: fix shellcheck on recently added files [08:21] pstolowski: this needs a 2nd trivial review https://github.com/snapcore/snapd/pull/3234 [08:21] PR snapd#3234: overlord/snapstate/backend,interfaces/mount: move ns management code [08:21] ok [08:25] morphis, Pharaoh_Atem: is that shellcheck fix something you can easily carry in the tree? [08:25] (as a pathc) [08:25] patch [08:25] zyga: I think Pharaoh_Atem is fine as long as we have a PR or the PR merged upstream [08:26] PR snapd#3230 closed: tests: extend network-control spread test to cope with network namespaces [08:29] zyga: do you have a raspbian system at hand? [08:29] morphis— out of curiosity, for what? [08:29] Chipaca: snapd packages :-) [08:30] morphis: yes [08:30] well [08:30] on a pi1 [08:30] but I can (probably) move that to a pi2/3 easily [08:30] zyga: hm, would be nice to see the error message on the pi1 [08:31] error message about? [08:32] zyga: that we don't have a suitable core snap for the pi1 [08:32] I guess it will try to load the armhf one but not sure [08:32] PR snapd#3234 closed: overlord/snapstate/backend,interfaces/mount: move ns management code [08:33] morphis: I think it will just fail, let me try [08:33] pstolowski: thanks! [08:33] zyga: yeah [08:33] pi1 is so slow... [08:34] :-) [08:34] zyga— remember snap 1 on the bbb? [08:34] or weren't you around for that [08:34] snappy* [08:34] Chipaca: snap 1 as in 15.04 snapd? [08:34] snappy --help [08:34] zyga, yw [08:34] <30 seconds waiting> [08:34] Chipaca: I recall it was slow :D [08:34] was that in python or all go already? [08:35] snappy was python [08:35] heh [08:35] nice :D [08:35] and the slow was the main driver for the move to go [08:35] * zyga would love to see python code of snappy [08:35] zyga— no, no you wouldn't [08:35] but it's in git :-) [08:35] zyga— this was pre-overlord, so the fs was the transaction layer [08:36] was is terrible? [08:36] aaaaha :D [08:36] remember the 16.04 push to make the insanity go away? [08:36] morphis: let me update first, I'll install snapd shortly [08:38] zyga: let me upload the repo first [08:47] morphis: which repo is that? [08:49] zyga: deb https://mm.gravedo.de/raspbian/ jessie main [08:50] morphis: fyi, raspbian already has snapd [08:50] 500 http://archive.raspbian.org/raspbian stretch/main armhf Packages [08:51] zyga: stretch does, yes [08:51] 2.21-2 [08:51] zyga: that is currently in rc [08:51] so everyone out there is still using raspbian based on debian 8 [08:51] which doesn't have it [08:51] and a stable raspbian based on debian 9 is still months away [08:57] zyga: what you officially get from the pi foundation is jessie: https://www.raspberrypi.org/downloads/raspbian/ [09:00] zyga: are you running stretch then on your pi? === chihchun is now known as chihchun_afk [09:05] morphis: yes [09:05] morphis: well, not stretch, but raspbian + jessie/wheezy/stretch raspbian archives [09:06] sounds hacky :-) [09:12] zyga: however, if you add the repository you should get snapd 2.24 === chihchun_afk is now known as chihchun [09:15] zyga, have you seen the failures in master? https://forum.snapcraft.io/t/tests-broken-in-master/457 [09:16] pstolowski: no, looking now [09:16] pstolowski: can you please edit your post to use formatting [09:18] zyga_: once I did some more testing I will make a post in the forum to call for wider testing [09:20] morphis: ^ did you see that failure? update test hangs on debian [09:21] zyga_: which failure? [09:24] pstolowski: you have a link to the test run on in travis? [09:24] zyga, done; i thought I did by using ` , but apparently it's not enough [09:25] zyga, https://travis-ci.org/snapcore/snapd/builds/227867834?utm_source=email&utm_medium=notification [09:25] pstolowski: I think you need ``` [09:26] i'm not sure it's from the same travis run, but failed the same I think [09:27] pstolowski: so... [09:27] pstolowski: I think we rejected one fix that may be the cause here [09:27] it got stuck on Run configure hook of "core" snap if present [09:27] zyga, pstolowski: so this is running 2.21 and installing the core snap from stable [09:29] zyga, pstolowski: sounds very muhc like https://bugs.launchpad.net/snappy/+bug/1674193 [09:29] Bug #1674193: core snap's configuration hangs on debian | openSUSE | mainline kernel [09:31] morphis, indeed it's an upgrade from old version. but I wonder why it's failing now, i don't think it's a new test? [09:31] pstolowski: did we release a new core snap to stable recently? [09:31] or rather, any new conditions [09:32] dunno [09:32] zyga_: any idea? [09:32] morphis: yes [09:32] * zyga_ looks for the PR [09:33] https://github.com/snapcore/snapd/pull/3145 [09:33] PR snapd#3145: overlord/ifacestate: fix auto-connect during core snap transition [09:33] zyga_, pstolowski: ok, can reproduce the same problem on my debian install [09:34] look at https://github.com/snapcore/snapd/pull/3145/commits/71ce65eb95b3210515a8c81f4737e5c3d9bd18fb [09:34] so we rejected that PR [09:34] let me try to recall the rationale [09:34] https://github.com/snapcore/snapd/pull/3145#discussion_r110766622 [09:34] PR snapd#3145: overlord/ifacestate: fix auto-connect during core snap transition [09:34] zyga, looking at your last comment, you had a simpler variant merged? [09:35] so the question is, why did the slot not auto-connect after restarting into new snapd? [09:35] zyga: with a core snap from edge the problem doesn't exist [09:36] a $ snap install core fails and a $ snap install --edge core works [09:36] morphis: how about edge? [09:36] er [09:36] beta [09:36] zyga: what do we have currentl in stable? [09:36] morphis: no idea [09:36] morphis: the version field is useless [09:36] wonderful .. [09:37] ogra_: hey [09:37] zyga_, yo [09:37] ogra_: is the snapcrat feature that would let us give real and sensible verison available now [09:37] shall we take this conversation to https://forum.snapcraft.io/t/tests-broken-in-master/457? [09:37] ogra_: so that "snap info core" shows the snapd version there? [09:37] zyga_, not sure if it landed with the last drop [09:38] zyga_: beta works too [09:38] zyga_, if it did i have to do a few changes first ... the bug wasnt closed but perhaps thats just an oversight from sergio [09:38] ogra_: aha, thanks [09:38] morphis: ok, then when 2.25 is released this should be fixed [09:38] zyga: right [09:38] however that wont help us to get CI back working on master [09:39] zyga_: unless we do a $ snap install --beta core for debian during the test run [09:39] morphis: I think that's a reasonable compromise for now [09:39] let me do a PR [09:39] is federico working today? I'd like to know his opinion [09:39] morphis: please do, thanks! [09:41] pstolowski: yeah, let's summarize on the forum [09:41] we can paste this converstation [09:43] zyga_, pstolowski: https://github.com/snapcore/snapd/pull/3259 [09:43] PR snapd#3259: tests/upgrade: force install core snap from beta for debian [09:43] pstolowski: can you summarize the discussion in the forum? [09:44] PR snapd#3259 opened: tests/upgrade: force install core snap from beta for debian [09:49] morphis, will do [09:52] morphis: hmm [09:52] morphis: so one question [09:52] morphis: after 2.25 is out it won't change anything in debian [09:52] morphis: because debian is frozen, right? [09:52] morphis: so I'm trying to understand one thing there [09:52] morphis: the test installs some debs [09:52] morphis: are they not the same as snapd in edge? [09:52] morphis: (except for the name of the plug?) [09:53] morphis: what I'm trying to say is that perhaps, if we get this rigth, the change is not needed, no need to special-case debian [09:53] morphis: I think some code may be missing from master [10:00] morphis: master contains code that renames core-support to core-support-plug [10:00] morphis: so it should work [10:01] why? [10:01] better asked, why -plug as suffix? [10:01] morphis, zyga_ summarized the discussion in the forum, let me know if i missed anything [10:01] morphis: to keep plug and slot names distinct [10:02] zyga_: isn't is already imminent for a plug that it is a plug? [10:05] ogra_: in re yesterday's question about test script, I don't know [10:05] maybe? [10:05] the shellcheck thing has always been there, though [10:06] it passed in 2.24 [10:06] zyga_, morphis: https://paste.fedoraproject.org/paste/NmsFYMd2WEStihK1ATvizV5M1UNdIGYhyRLivL9gydE= [10:06] shellcheck test fails on 2.25 [10:06] Son_Goku, to late ... fix is alöready pending [10:06] :) [10:06] ogra_: is there an active PR? [10:06] (zyga works in his sleep :P ) [10:07] https://github.com/snapcore/snapd/pull/3258 [10:07] PR snapd#3258: cmd/snap-confine/tests: fix shellcheck on recently added files [10:07] excellent :D [10:07] well... not so excellent [10:07] what the hell happened to spread/travis? [10:08] oh, welp [10:08] Son_Goku: from what I experienced so far there are differences between shellcheck on Ubuntu and others [10:08] also the version plays a role [10:08] yeah, shellcheck runs are disabled on F24 and older [10:08] the 14.04 shellcheck is pretyt crappy ... not sure what versioon that was thopugh [10:09] because shellcheck < 0.4.4 doesn't actually support -x [10:09] my trees all install a 16.04 chroot when doing shellcheck [10:09] Fedora 24 and EPEL 7 have shellcheck 0.3.x, which can't actually run the tests :/ [10:09] is pedronis around today ? [10:10] manually upgrading a pi3 gets me "[/] Setup snap "core" (1829) security profiles (phase 2)" ... forever .... shell prompt never returns [10:11] (pretty much like described in one of the forum threads) [10:18] wut [10:18] this PR somehow breaks Fedora builds [10:18] Son_Goku: fixed [10:18] Son_Goku: sorry, the patch had a broken file inside by accident [10:19] take 4 [10:21] good, gobuild is fixed [10:21] now we're at C build [10:21] shellcheck passed... good [10:23] Son_Goku: yeah, sorry about the earlier patch [10:24] * zyga_ feels like he has cold :/ [10:24] zyga_, dude ... i tend to test stuff before commenting ... shellcheck didnt disagree ;) [10:24] but leave it ugly if you feel like ;) [10:25] ogra_: I think it may be different version then, I did fix all the things shellcheck reported [10:25] ogra_: I tested on F26 [10:25] ah [10:25] I wanted to have a patch that would be clean for fedora packaging [10:25] well, newer ubuntu versions might spill a warning there [10:25] btw, I think shellcheck would disagree about $FOO/"whatever" but I may be worng [10:26] warning? [10:26] yeah [10:26] about extra quoting? [10:26] zyga_: I think shellcheck would be fine with "$FOO/whatever", though [10:27] about possible interpretation as: "$TMP/$(basename " $0 ")" [10:27] well, for inner quotes, shouldn't you use a different quote for the inner one? [10:27] anyway, if it doesnt complain now, leave it [10:27] like single quotes for inner and double for outer [10:28] technically just quoting $0 is enough [10:29] Son_Goku: no, that's actually correct [10:29] Son_Goku: $() gives new quoting context [10:29] zyga_: ah [10:29] shell is confusing :) [10:29] Son_Goku: "$(echo "This is $(echo "ok")")" [10:29] yes [10:31] well, I'm about ready to commit my 2.25 package [10:31] +1 [10:31] unless there's any other patches I need besides PR3222 and PR3258 (though I don't need the latter) [10:31] Son_Goku: not sure, I don't know of any [10:32] well, I guess that's what testing updates is for :) [10:32] I've also created a snap-mgmt script that will be run on %preun on final removal [10:32] based on the snapd.postrm script [10:32] so that will need specific testing [10:36] zyga_, morphis: building: https://koji.fedoraproject.org/koji/taskinfo?taskID=19367859 [10:36] Son_Goku: awesome! [10:37] Son_Goku: thank you [10:38] building on all Fedora branches now [10:39] zyga_: there are some advantages to building for an upstream distro that moves fast :) [10:45] I hate armv7hl builds :( [10:45] they need to fully move over to the VMs on aarch64 hosts [10:45] those are so much faster... [10:57] Chipaca: hey [10:58] zyga_— hi [10:58] Chipaca: can you please review https://github.com/snapcore/snapd/pull/3259 [10:58] PR snapd#3259: tests/upgrade: force install core snap from beta for debian [10:58] Chipaca: master is currently broken because of htis [10:58] Chipaca: howerver plase note what I said above: [10:59] Chipaca: actually let me add that to the forum [10:59] zyga_— you said a lot of things above. Which is the note? [10:59] https://forum.snapcraft.io/t/tests-broken-in-master/457/6 [10:59] I want to fix master but I'm not sure this is the actual fix [11:00] and that we're not missing anything [11:06] zyga_, morphis: https://bodhi.fedoraproject.org/updates/FEDORA-2017-866e9643a8 (F24), https://bodhi.fedoraproject.org/updates/FEDORA-2017-2e4459fa03 (F25), https://bodhi.fedoraproject.org/updates/FEDORA-2017-74f7c7df46 (F26) [11:10] morphis, zyga_: please specifically test final removal of snapd *after* upgrading or installing v2.25 [11:10] source script: https://src.fedoraproject.org/cgit/rpms/snapd.git/tree/snap-mgmt.sh [11:11] install step: https://src.fedoraproject.org/cgit/rpms/snapd.git/tree/snapd.spec#n434 [11:11] scriptlet execution: https://src.fedoraproject.org/cgit/rpms/snapd.git/tree/snapd.spec#n568 [11:12] Son_Goku: ack, will do [11:13] Son_Goku: will do! [11:14] this is specifically for solving https://bugzilla.redhat.com/show_bug.cgi?id=1444422, so that's what this aims to fix [11:17] * Son_Goku hopes the patch load doesn't get any bigger on snapd... [11:17] I'm sort of cheating by using PR diffs, but the number of patches being applied to snapd is quite high now :( [11:28] fwiw i think we agreed that we should let you remove core if it was the last snap left, but it wasn't prioritised [11:29] Chipaca: yes, I think we agreed [11:29] (but my memory is bad so don't trust me alone on this) [11:29] Chipaca: I think it should be easy, right? [11:29] Chipaca: no other snap installs in progress [11:29] Chipaca: and no other snaps [11:30] zyga_— i think it wasn't super easy because of how isolated the bit that checks for core was, but i could be misremembering [11:30] in any case, it still isn't prioritised :-) [11:30] Chipaca: we could catch the context at the time the op is made [11:30] and pass a flag like "ok-to-remove-core' [11:32] zyga_— that sounds like the wrong way to do it [11:33] * Son_Goku grumbles about crappy hotel Wi-Fi [11:51] lunch! [12:01] good idea [12:04] zyga_, https://bugs.launchpad.net/snapd/+bug/1687608 ... you asked about more info about this in https://forum.snapcraft.io/t/device-gets-bricked-after-trying-to-upgrade-core-snap/453 .... i can reproduce it fine here if you want me to collect data [12:04] Bug #1687608: running "snap refresh core" on an UbuntuCore device locks you out until reboot === palasso_ is now known as palasso [12:11] re [12:11] ogra_: looking [12:12] zyga_: you had time to give snapd on raspbian a try? [12:12] ogra_: aha, can you tell me what is in snap changes? [12:12] morphis: no, I left the apt-get update && apt-get dist-upgrade -y running in the background [12:12] let's see [12:13] zyga_, http://paste.ubuntu.com/24498251/ [12:14] morphis, zyga_: could one of you guys create a topic on the forums announcing the snapd 2.25 testing for Fedora once it syncs out to updates-testing? [12:14] zyga_, note that my SD is worn out, thus change 13 failed (but rolled byck fine) ... change 14 is the actual one showing the issue [12:14] *back [12:15] interesting [12:15] I wonder why change 13 failed [12:15] how is your SD a factor? [12:16] zyga_, because uboot.env was trashed (Sd worn out) [12:16] aah [12:16] 2017-05-02T09:40:50Z ERROR cannot finish core installation, there was a rollback across reboot [12:16] Son_Goku: sure [12:16] right, ignore that one [12:16] it rolled back fine [12:16] ogra_: ok so after rollback what did you do? [12:17] zyga_, snap refresh core [12:17] 2017-05-02T10:04:41Z INFO Waiting for restart... [12:17] I'm trying to grok the log there [12:17] it was waiting for restart over and over [12:17] which also works fine but does never finish the "Setup snap "core" (1829) security profiles (phase 2)" bit [12:17] yeah [12:17] I think we have a bug there [12:17] let me look [12:17] yes, we do [12:17] it used to give me the console back before ... now it doesnt anymore [12:19] weird [12:19] so [12:19] we log that message when snapd itself saves in the state that it is waiting to restart [12:20] yeah, the message was there before ... but didnt spin forever [12:20] and that is not even saved in the state [12:20] it's just saved in memory [12:20] so we really need to restart [12:20] (though I suspect that it may be broken if we restart snapd rather then the machine) [12:20] the security profiles thing is saved in memory ? [12:20] ogra_: no [12:20] the "restarting" flag [12:21] the restarting flag isnt the issue [12:21] the task just does nothing until that flag is cleaned [12:21] so what is? [12:21] the issue is "Setup snap "core" (1829) security profiles (phase 2)" never finishing [12:21] or at least never returning the console access [12:21] ogra_: ... [12:21] ogra_: did you look at the code? [12:21] no [12:21] ogra_: the restaring flag guards that code from running [12:22] ogra_: that's why I mentioned it [12:22] well, the code obviously runs :) [12:22] ogra_: yes and bails out instantly because of that [12:22] http://paste.ubuntu.com/24498125/ shows even the spinner rotating between line 60 and 78 [12:23] yeah, the spinner is client side query [12:23] so... [12:23] ah, so you mean the reboot flag kicks in to early ? [12:23] no, I don't think it does [12:23] * zyga_ thinks [12:23] so [12:23] it's just this [12:23] we set the flag [12:23] and ask the system to reboot [12:24] right [12:24] but the reboot is not instant, it will be a while before it happens [12:24] which obviously works as expected [12:24] then lots of time passes [12:24] and then we reboot [12:24] and then it works OK [12:24] but the UX is not making it clear what is going on [12:24] that message has been there before since day one ... but it always left me interacting with the tty ... now it doesnt anymore and the spinner goes on [12:25] ogra_: you can ctrl-c and exit [12:25] ogra_: but I suspect shutdown may set nologin [12:25] before the spinner stopped at some point [12:25] no, ctrl-c is ignored [12:25] ogra_: yes but there was a bug there (hence phase 2) [12:26] morphis: no luck with your repo [12:26] morphis: apt-get doesn't like it [12:26] i think before it acvtually printed the "[/] Setup snap "core" (1829) security profiles (phase 2)" ... finished it and only then the reboot message showed up [12:27] at least thats what i remember [12:27] now both are there at the same time and the "[/] Setup snap "core" (1829) security profiles (phase 2)" seems to hold the console access [12:27] zyga_: what does apt-get say? [12:27] it will most likely complain because if the unknown key [12:28] E: Failed to fetch https://mm.gravedo.de/raspbian/dists/jessie/InRelease [12:29] raspbian !? [12:29] zyga_: that file exists, can you give me the full output? [12:29] ogra_: yeah [12:31] morphis: there's nothing else reall [12:31]