[00:59] ok, copy & paste of snap building was fine: https://github.com/sd-hd/termite-snap/runs/407725965?check_suite_focus=true [01:00] hmm, thanks === dgadomski is now known as dgadomski_ [11:06] uh 18C in the office [11:08] * zyga debugs lxd issue [11:21] and now it obviously doesn't fail [11:21] WTF [14:24] waaat [14:24] stgraber: hey [14:24] stgraber: not sure if you are traveling now [14:25] stgraber: on an amd64 system, "lxc launch ubuntu:xenial" launches ... i386 container [14:26] er "lxc launch ubuntu:18.04" [14:26] doesnt here [14:26] ogra: it does here once in a few runs [14:26] ogra: it's super weird [14:26] it started yesterday [14:26] https://www.irccloud.com/pastebin/FmxUFgc1/ [14:26] very strange [14:27] is there a way to pick the architecture explicitly [14:27] must be new indeed [14:27] lxc [14:30] zyga: I'm still in Cape Town [14:30] stgraber: aha [14:30] stgraber: was there a release of lxd into candidate lately? [14:31] stgraber: we're pulling candidate in all our tests [14:31] zyga: I know what change could have caused this, but it's not happening on my system here using master so it's unlikely to be something that hits every time [14:31] stgraber: it's random [14:32] hits every few times [14:32] maybe dictionary ordering [14:32] or something like that [14:32] zyga: what does `lxc image info ubuntu:xenial` get you? [14:32] the logic specifically sorts by arch, starting by primary before going to secondary, so shouldn't be random [14:32] checking [14:33] https://www.irccloud.com/pastebin/RRCUeVqC/ [14:33] architecture looks correct here [14:34] zyga: hit it here just now [14:34] well, on a remote server, CPT internet is too slow to do a test loop :) [14:34] :) [14:34] yeah [14:36] stgraber: if there's a chance to tweak it and patch candidate we'd love that [14:36] root@buildd01:~# for i in $(seq 20); do lxc image info ubuntu:xenial | grep Architecture; done | sort | uniq -c [14:36] stgraber: otherwise please give me a signal and I'll mask LXD tests for now [14:36] 5 Architecture: i686 [14:36] 15 Architecture: x86_64 [14:36] PR snapd#8047 opened: tests: detect LXD launching i386 containers [14:36] zyga: yes, it's a fix we had to rollout to fix all foreign arch (say armhf on arm64) that was broken, so we'll fix the fix and push it through quickly [14:36] haha, thank you :) [14:36] zyga: we obviously don't want folks to accidentally run their foreign arch instead of native :) [14:37] I hope I didn't ruin your weekend [14:37] just call it an easter egg ! [15:25] testing a fix now [15:25] zyga: nah, I'm at the company meetings this weekend [15:35] zyga: I think my fix is working, just finishing a test loop [15:35] yep, ran 1000 times and all x86_64 [15:35] sending upstream [15:37] zyga: https://github.com/lxc/lxd/pull/6773 [15:37] PR lxc/lxd#6773: shared/simplestreams: Fix inconsistent sorting [17:27] stgraber: super, looking [17:28] zyga: it's in candidate now [17:28] stgraber: yep, I just checked [17:28] re-starting our tests [17:28] thank you! [17:28] Will promote to stable after dinner once Jenkins is happy [18:18] stgraber: hmmm, I just reproduced it again... from candidate [18:18] ah, no sorry [18:18] hmm [18:18] actually [18:18] yeah [18:18] https://www.irccloud.com/pastebin/D0mfZLII/ [18:18] This was two hours ago [18:19] Right so on the old rev still [18:19] I ran a for loop trying it 1000 times so pretty sure this code path isn't racy anymore [18:19] ok [18:20] re-started [19:27] stgraber: did you publish to candidate? [19:27] it failed again at 18:57 CET [19:28] stgraber: so either I'm running unfixed revision or it is still broken [19:28] lxd (candidate) 3.19 from Canonical* installed [19:36] Do you know the revision? [20:31] stgraber: no, the test doesn't show it [20:32] I'll change it to do so [20:35] stgraber: https://github.com/snapcore/snapd/pull/8047 will show it soon [20:35] PR #8047: tests: detect LXD launching i386 containers <⚠ Critical> [20:53] stgraber: installed revision 13113 [20:54] stgraber: version 3.19 from latest/candidate [20:54] zyga: 13113 isn't an amd64 revision [20:54] stgraber: note that latest/edge is 13119 [20:54] zyga: 13115 is the amd64 one [20:54] doh! [20:54] indeed [20:55] so yeah, if testing on 32bit, you'll get i386 [20:55] we still run -32 version [20:55] if you weren't, that'd be a bug :) [21:15] corrected my checks, let's see [21:25] PR snapcraft#2893 opened: [legacy] meta: include environment in hook wrappers [21:40] PR snapcraft#2894 opened: Fix issue with multipass mount on win32 [22:12] stgraber: all good