=== chihchun_afk is now known as chihchun === chihchun is now known as chihchun_afk [03:13] PR snapd#6721 opened: tests: new profiler snap used to track cpu and memory for snapd and snap commands [06:43] Hello [07:00] * zyga was watching the news about Paris last night [07:00] a bit sleepy, sorry [07:06] zyga: it looks like the selinux-clean test is unhappy (in master and in the recent PR spread runs). is anyone on it or shall we disable it for now? [07:09] mvo: mborzecki is on it [07:09] we talked about it a lot yesterday [07:10] essentially it keeps finding bugs [07:13] ok [07:14] mvo: there's a fun bug in the kernel and in apparmor breaking disco [07:14] mvo: but apparently only in lxd [07:14] mvo: the security team was on it last evening [07:14] disco ships on Thu, fun [07:15] yes [07:16] everyone is well aware, lsm stacking landed [07:16] lxd started using shiftfs [07:16] this confused apparmor parser caching [07:16] and no profiles got loaded === pstolowski|afk is now known as pstolowski [07:16] mornings [07:16] hey pawel [07:17] pstolowski: good morning [07:17] zyga: woah, sounds like fun indeed [07:27] PR snapd#6722 opened: tests: set selinux-clean test to manual for now [07:34] woah, now travis fails in the static tests even oh boy [07:35] PR snapd#6673 closed: cmd,tests: forcibly discard mount namespace when bases change [07:36] oh fun [07:37] mvo: did go change in travis? [07:38] PR snapd#6723 opened: cmd/snap-confine: remove unused sc_open_snap_{update,discard}_ns [07:38] pstolowski, mvo: ^ trivial [07:38] done [07:45] zyga: you got it as well [07:45] thank you [07:45] aaah [07:45] I see [07:45] the go version fun [07:46] super mysterious as we did not change anything there did we? [07:47] * mvo looks [07:48] this looks like a travis issue, we request go1.9 [07:48] maybe they dropped it? [07:49] morning [07:49] hey mborzecki [07:50] mvo: hey [07:51] 56 PRs open, better than yday [07:51] mborzecki: yeah, its getting better except master is not happy [07:52] mvo: what's the problem with master? [07:52] mborzecki: but we are working on it :) some things could land once this is unbroken [07:52] mborzecki: the selinux-clean test is unhappy more often than not [07:52] mborzecki: also static checks are complaining right now [07:52] mvo: heh [07:52] PR snapd#6724 opened: travis: fix requested go version <⛔ Blocked> [07:52] mvo: #6718 should help a bit with selinux clean test [07:52] PR #6718: spread, tests: do not leave mislabeled files in restorecon test, attempt to catch similar files [07:52] mborzecki: static checks have a version mismatch and die very early [07:53] mborzecki: aha, nice, is that green? [07:53] mvo: and yesterday i noticed that snapd started calling runuser [07:53] is that automatic snapshotting? [07:53] mborzecki: I think so [07:53] mborzecki: 6718 ironically fails in selinux-clean [07:53] Chipaca probably knows more about that (runuser) [07:54] mvo: yeah, looked at it yday evening and iirc it was because of runuser [07:54] aha, ok [07:55] flag provided but not defined: -V, hmm [07:56] PR snapd#6724 closed: travis: fix requested go version <⛔ Blocked> [07:57] conspiracy theories in the forum [08:01] it looks like version mismatch - go tool asm -V does indeed not exist in 1.9 [08:01] so it looks like go vet is driving it the wrong way (or something is) [08:01] * mvo has a meeting now so won't look at this [08:03] mvo: where do we get vet from? [08:03] mborzecki: I don't know [08:04] mvo: go tool vet is part of golang distribution package, and afaik vet is just a frontend, presumably both should be from the same distro package [08:07] runuserwha? [08:07] g'morning [08:07] mborzecki: what's up with runuser? [08:07] Chipaca: morning, runuser, we use it to snapshot user's data right? [08:08] mborzecki: yes (and hopefully any user data manipulation someday) [08:08] mborzecki: why? [08:09] Chipaca: cool, we need to extend the selinux policy to allow some actions it does (setgid, writing to audit and so on) [08:10] pedronis: starting on coherence, do you have a link to the server-side to hand? [08:10] fwiw, the changes look trivial, i should have a PR up soon [08:10] Chipaca: one sec [08:10] mborzecki: i guess we only saw this now because of automatic snapshots? [08:11] Chipaca: yes, that would be my guess, wonder why it didn't come up in the travis job of the PR that added automatic snapshots [08:12] well, maybe the PR did not include the test back when it was opened [08:25] re [08:25] sorry, had to clean the office a little [08:31] Chipaca: sorry, need to have a quick errands, I can give you links after (in ~30 mins) [08:31] pedronis: no probs [08:39] PR snapd#6725 opened: travis: add some debug to run-checks [08:45] why on earth we use /home/gopath and since it's under /home we treat it like another homedir [08:46] but since it's not a homedir (there's no gopath user), none of the selinux transitions take place [08:46] I think I found the issue [08:46] with the status check [08:46] oh well [08:46] I wonder why this ever worked [08:47] mvo: what did you find? [08:49] zyga: PR up in 60s, then it will be obvious [08:50] thank you :) [08:52] PR snapd#6725 closed: travis: add some debug to run-checks [08:52] PR snapd#6726 opened: tests: do not hardcode go1.10 in travis [08:52] zyga: -^ [08:52] * mvo takes a short break [08:52] looking [08:53] interesting [08:53] mvo: there was a simliar bug in suse [08:53] that is, suse golang stack got updated during startup [08:53] but the environment variables were pointing to the old stack [08:53] with similar insane errors === chihchun_afk is now known as chihchun [09:06] policy update for runuser is up [09:06] PR snapd#6727 opened: data/selinux: allow snapd to execute runuser under snappy_t [09:26] Hi. I'm new to Ubuntu Core and I want to boot from it from a usb pendrive without using any device other than my notebook (so I have no device to SSH from) -- what is the easiest way to do this? I don't have VTx support on my CPU and wouldn't prefer using a virtual machine, partly because I have an idea of using ubuntu core as my main desktop system (browser and a terminal is all I need from it). [09:32] mborzecki: with 6727 can we get rid of 6722? [09:33] ilyaigpetrov: hey [09:33] ilyaigpetrov: I think core is not quire suited for that use case yet [09:33] mvo: yes [09:33] ilyaigpetrov: it is mainly usable as virtual machines in the cloud and single-purpose devices in the field [09:33] zyga: I want to learn new things this way [09:33] ilyaigpetrov: support for users and sessions is limited [09:33] ilyaigpetrov: you won't get a graphics session or a browser easily either [09:34] ilyaigpetrov: if you want to learn you can try getting the amd64 ubuntu core image and copy it to a USB drive [09:34] ilyaigpetrov: then instruct your device to boot from it [09:34] mvo: #6726 will probably fail on fedora/centos due to selinux, we should merge it anyway [09:34] PR #6726: tests: do not hardcode go1.10 in travis <⚠ Critical> [09:34] ilyaigpetrov: it will work, it's just not a device that one would use for daily stuff IMO [09:35] mborzecki: selinux question, when are semicolons required? [09:35] zyga: yes, I boot from the usb pendrive, enter my ubuntu SSO and it says I need to ssh into it, but I don't have any other notebook to log in via ssh from [09:35] sorry, that's the only way in [09:35] once you ssh you can set a password [09:35] zyga: I think to create a box based on raspberry Pi for kids to learn computers and coding, I want them being able to log in [09:36] zyga: is there any image customization I can do? [09:36] zyga: allow rules require it at the end, interfaces have a bunch of allow rules, each should end with ; already [09:36] ilyaigpetrov: I commend your desire to learn but I don't think core is the best use for that yet, for learing there are much better suited (especially for children) tailored distributions with graphical displays and keyboard / mouse support [09:37] ilyaigpetrov: what kind of customization? [09:37] mborzecki: indeed, so how about I combine 6726 and 6722 and then master should be green again. then we combine your selinux fixes and run it ~10 times or so and if all is green there we enable selinux-clean again? sounds reasonable? [09:37] zyga: replace config launcher on getty (I guess it's called getty) [09:37] ilyaigpetrov: no, that's not something you can do easily [09:37] ilyaigpetrov: and it will be undone by the system on the next update [09:38] *sigh* so sad [09:38] ilyaigpetrov: note that the user has no password [09:38] ilyaigpetrov: you must login via ssh first [09:38] ilyaigpetrov: then you can login interactively on another tty [09:38] (tty1 is special, rest should be regular) [09:38] ilyaigpetrov: core is designed for single use devices without users really [09:38] ilyaigpetrov: you don't expect each streetlight to have an user account, this is what core is aiming at [09:39] zyga: can I install a basic wayland compositor (or a mir server) and from another snap to install a firefox browser -- will they work together? [09:39] mvo: sounds good, then we can land 6727 and 6718 [09:40] I'm not sure I understand how desktop snap is used by other snaps [09:40] mborzecki: cool, let me prepare things [09:40] * zyga quick breakfast [09:40] ilyaigpetrov: there are no snaps that support user sessions on a core system [09:41] zyga: mm, but I have seen egmde confined snap [09:41] https://snapcraft.io/egmde-confined-desktop [09:41] I don’t know about that [09:41] some of my branches need a second review: 6418 (might be interessting for zyga), 6599 and 6603 would be great. also 6627. all remodel related but relatively small and self-contained. *please* :) [09:42] mvo: gladly! [09:42] In 15 minutes [09:42] zyga: thanks for you help, bond appetite [09:42] ilyaigpetrov: you can try core in qemu [09:42] On i386 you don’t need hardware support [09:43] zyga: I really like the idea of using core as my main system, even if virtual tty is all I have with a single user [09:46] ilyaigpetrov you need a 2nd machine then [09:47] ilyaigpetrov: i have a laptop which is running Ubuntu core. It's a touch limiting for normal use. [09:50] popey: I used to use gentoo confining myself to using vttys only for learning purposes, and I learn how to run a single window of a browser on naked xorg server, I think it was useful [09:50] It's certainly fun [10:00] back now [10:00] mvo: looking [10:02] mvo: note I will still address the snapd tools directory going stale [10:02] mvo: separately from this PR [10:06] mvo: https://github.com/snapcore/snapd/pull/6418#pullrequestreview-227099069 [10:06] PR #6418: many: allow core as a fallback for core16 [10:06] mvo: it should be ok to just push the commit suggestion buttons a few times :) [10:07] mvo: doing another one [10:08] mvo: about 6599, should AddAll just add edges? [10:08] mvo: or fail if edges are present? [10:09] sil2100: hey, is there anything in the latest core18 that may affect wifi? [10:10] mvo: https://github.com/snapcore/snapd/pull/6599#pullrequestreview-227102274 [10:10] PR #6599: snapstate,state: add TaskSet.AddAllWithEdges() and use in doUpdate [10:10] zyga, mvo: or do extra iteration to check first? [10:10] hey cwayne, long time no see :) [10:10] zyga: heya :) how's it goin? [10:11] cwayne: great, swamped under bugs and feature work but feeling positive today :) [10:12] zyga: I'd rather be swamped than have nothing to do :) [10:12] cwayne: I think we are not at any risk of having nothing to do :) [10:12] lol true! [10:16] cwayne: hm, I'd have to dig deeper, but I don't think so [10:16] We did release a new systemd yesterday, which is why a new core18 was triggered [10:17] sil2100: ok, will poke on this end, seeing some wifi failures on Pi's which we hadnt seen in awhile [10:18] Oh, hm [10:18] cwayne: for a moment I was worried that someone released the new netplan package, but no, at least not for bionic [10:19] And anyway, Matt said that he addressed the previous networking issue in the current netplan SRU anyway [10:19] Yeah this isn't the same issue we saw earlier with netplan for sure [10:19] Ok [10:20] mvo: https://github.com/snapcore/snapd/pull/6603#pullrequestreview-227104276 [10:20] PR #6603: snapstate: add new NoReRefresh flag and use in Remodel() [10:21] mvo: https://github.com/snapcore/snapd/pull/6627 has some unanswered questions [10:21] PR #6627: devicestate: deal correctly with the "required" flag on Remodel [10:21] doing a pass anyway [10:22] zyga: aha, ok, let me double check, maybe I missed something. thank you [10:22] cwayne: the new systemd SRU we released did address some network bugs, so who knows, maybe there's some regression there [10:22] cwayne: but the changes didn't seem highly likely to cause wlan-only issues [10:23] sil2100: ack, well poke around on our end [10:25] zyga: sorry, was in a meeting I see you wrote more. will get back to that in a wee bit [10:26] mvo: no worries, I'm going through the last one you requested [10:26] mvo: https://api.travis-ci.org/v3/job/520673622/log.txt 6722 failed in canonical-livepatch test [10:26] PR snapd#6728 opened: overlord/devicestate: extra measurements related to populateStateFromSeed [10:26] uhmmm.. unhappy day for travis [10:26] mvo: copied the log here: https://paste.ubuntu.com/p/QnVW7vCqJ6/ i'll restart the travis job [10:31] mvo: https://github.com/snapcore/snapd/pull/6627#pullrequestreview-227110498 [10:31] PR #6627: devicestate: deal correctly with the "required" flag on Remodel [10:31] https://github.com/snapcore/snapd/pull/6726 is red [10:32] PR #6726: tests: do not hardcode go1.10 in travis <⚠ Critical> [10:32] zyga: yes, it'll land as 6722 [10:32] ack [10:33] unless anyone has a review they want to pull me into, I will resume working on tests [10:34] hmm canonical-livepatch failed there too :/ [10:35] yeah [10:35] one of those days, as they say [10:35] yeah, looking at this now [10:35] looks like something changed there as well [10:37] pedronis: re your question to https://forum.snapcraft.io/t/how-to-disconnect-a-snap-from-internet/10901/10 from yesterday - yes, we have a spread test for this - it's main/disable-autoconnect [10:38] pstolowski: thx [10:43] PR snapd#6729 opened: tests: update canonical-livepatch now that -gcp kernels are supported [10:45] PR snapd#6730 opened: tests: fix master [10:47] mvo: looks like we need a one PR to unite all fixes [10:48] mborzecki: yeah, I added this in 6730 [10:48] mborzecki: or is it missing something? [10:49] fwiw, I'm also looking at livepatch now, maybe we can make the needle we look for better [11:08] huh, store.RefreshCandidate still exists [11:08] * Chipaca nukes it [11:15] sil2100: so re. the wifi failures you were talking to cwayne about earlier - nothing has changed in core18 but systemd? [11:38] PR snapd#6729 closed: tests: update canonical-livepatch now that -gcp kernels are supported [11:38] PR snapd#6730 closed: tests: fix master [11:39] * zyga hugs everyone who helped to fix master [11:39] PR snapd#6722 closed: tests: set selinux-clean test to manual for now [11:39] PR snapd#6726 closed: tests: do not hardcode go1.10 in travis <⚠ Critical> === chihchun is now known as chihchun_afk [11:44] mvo: so master is fixed now :) === ricab is now known as ricab|lunch [11:50] https://forum.snapcraft.io/c/release seems out of date. Where do relesae announcements go now? [11:55] rbasak: snapd releases are listed on https://github.com/snapcore/snapd/releases/ [11:56] greetings, how can I hide my ssid using snap wifi-ap? [11:57] zyga: thanks. I was hoping for a summary of changes though. For example, that page doesn't say what happened in 2.38. I thought that's what the forum topic was for/ [11:57] ? [11:57] rbasak: it used to say that [11:57] rbasak: just not all releases are listed with equal verbosity [11:57] mvo: ^ [11:57] there is also https://forum.snapcraft.io/t/the-snapd-roadmap/1973 [11:58] The bugfix releases are all detailed, but I was wanting to know more about the high level stuff happening in feature releases. [11:58] pedronis: ah that's useful - thanks! === chihchun_afk is now known as chihchun [12:04] isomari_: no idea. Are you sure you're in the right place to ask that? [12:08] anyone brave enough to do a 2nd review of https://github.com/snapcore/snapd/pull/6688 ? :) [12:08] PR #6688: gadget: add validation of cross structure overlap and offset writes [12:09] involves familiarizing oneself with gadget.yaml and some hair pulling while reading the code :) === chihchun is now known as chihchun_afk [12:12] mborzecki: I can look but only after the standup, trying to wrap up a bigger chunk here [12:12] zyga: thanks! [12:15] cwayne: mvo: so what issue are we seeing now for core18, is it in any way netplan-related? [12:18] Chipaca: Was asked to come here from ubuntu. [12:20] Chipaca: I found the answer. Thanks [12:27] PR snapd#6731 opened: overlord: make the store context composably backed by separate backends for device asserts/info etc [12:45] mborzecki: what's the status of master? needs 6727 ? [12:46] pedronis: master should be green, but selinux-clean test is disabled, 6727 enables the test and carries a fix for the policy [12:46] ok, thx [12:49] PR snapd#6704 closed: overlord/devicestate: measurements around ensure and related tasks [12:57] rbasak: sorry, I have not updated for the coming 2.39 features yet. or do you need info about a much older release on the forum? what version are you looking for? [12:58] mvo: nothing specific, thanks. === ricab|lunch is now known as ricab === chihchun_afk is now known as chihchun [13:29] mvo: hey, sorry to bother you. what is the trick with either creating a debian/fedora qemu autopkgtest image (autopkgtest-build-qemu unstable debian-sid-64.img fails for me) or accessing google (no access to google:debian-sid-64:tests/main/...) [13:29] hey [13:29] jdstrand: for debian you can get the qemu cloud image they have [13:29] AFAIK it works out of the box [13:29] hey zyga :) [13:29] (they == debian) [13:29] the fedora one I don't recall [13:29] if you wait a little I may have some links for you [13:29] hmm, googled for it. /me keeps looking [13:29] jdstrand: you can ask gustavo for a google spread key [13:30] jdstrand: in theory there is https://spread.zygoon.pl/images/ but I don't keep it up-to-date [13:30] jdstrand: I don't know how to create fedora images, I suspect its also "just" downloading their cloud image and adjusting the login creds [13:30] jdstrand: ^ mvo's advice is really best [13:30] ack. I see in spread.yaml what to change them to. let me see what I can find [13:31] I thought I had a google spread key. where is that stored? [13:31] jdstrand: it's essentially this: run any cloud image with cloud-init inside to the point where you have shell, set passwords and you _might_ be ready [13:31] jdstrand: but because google images have customization applied we don't really know [13:31] jdstrand: it's in ~/.config/gcloud [13:31] * zyga afk for 15 minutes [13:31] I have stuff in that dir [13:32] * jdstrand looks for the images [13:32] thanks! [13:32] ogra: did you happen to look into wireguard on Ubuntu Core? [13:33] * jdstrand notes wireguard upstream still says not to use it in production [13:36] * jdstrand may just use zyga's images and upgrade [13:36] oh those are just ubuntu. I can create those fine [13:38] PR snapd#6732 opened: tests: run livepatch on 18.04 as well [13:40] sergiusens, nope [13:42] sergiusens, there are two snaps in the store touch [13:42] *though [13:53] re [13:53] sorry, I was grabbed into a family lunch [13:54] jdstrand: let me know if I can help [13:54] I can regen and upload fresh images [13:54] I think I can upload some more too, there's a 10GB quota on that directory === chihchun is now known as chihchun_afk [14:15] PR snapd#6719 closed: many: move auth.AuthContext to store.DeviceAndAuthContext, the implemention to a separate storecontext package === chihchun_afk is now known as chihchun === chihchun is now known as chihchun_afk === Eickmeyer is now known as Eickmeyer___ [14:23] * dot-tobias says hi [14:30] is it possible to print output from hook scripts when they're running? I only get output if it fails, and then it's only the last ~5 lines [14:34] PR snapd#6692 closed: interfaces: cleanup internal tool lookup in system-key [14:39] dot-tobias: redirect the output to some log file under $SNAP_COMMON? [14:40] or /tmp [14:40] though tmp is harder to find [14:45] * zyga reviews 6688 [14:46] mborzecki: hey [14:46] mborzecki: remark about the gadget work [14:46] zyga: hm? [14:46] I think it would be great if you could add some docs , namely the key types Volume, VolumeStructure etc should be documented [14:46] they don't have to be long [14:47] I would put a paragraph or two in the file header, ahead of the docs of other stuff (perhaps package docs) for some quick context [14:47] then individual types can describe their role in the bigger picture [14:49] zyga: ok, i'll add something [14:50] PR # closed: core-build#11, core-build#22, core-build#26, core-build#37 [14:51] PR # opened: core-build#11, core-build#22, core-build#26, core-build#37 [14:51] mvo: Chipaca: pstolowski: store context refactoring has landed [14:52] pedronis: great, thx [15:00] pedronis: current status: https://pastebin.ubuntu.com/p/2TsZDK3wFZ/ [15:03] cool === cachio_ is now known as cachio [15:19] zyga, hey, I see this error when fedora 29 is updated https://paste.ubuntu.com/p/rv9Kzh24m7/ [15:19] zyga, any idea what could be causing this? [15:21] cachio: looking now [15:21] no idea, what does the unit do? [15:21] as in, what is systemctl cat man-db-cache-update [15:22] what does jurnald say about that unit? [15:24] zyga, I am regenerating the image to reproduce it because the image is dsicarded after that error [15:24] try to get an interactive shell [15:24] or expand the debug section to know more [15:26] zyga, sure, I am creating a new image to get an interactive shell [15:52] PR snapd#6733 opened: state: add possible error return to TaskSet.Edge() === setuid_ is now known as setuid === pstolowski is now known as pstolowski|afk [16:28] zyga, https://paste.ubuntu.com/p/x9s3GtZJgp/ [16:34] cachio: hey [16:34] cachio: what does the unit do? [16:34] can you provide the output of "systemctl cat"? [16:35]