[01:05] PR snapd#8398 opened: usersession/userd: add "slack" to the white list of URL schemes handled by xdg-open [05:45] morning [06:47] mvo: hey [06:48] mvo: the console_conf PR in subiquity has landed [06:49] mvo: i'm assembling the image to try everything out [06:54] hey mborzecki [06:54] mborzecki: niiiiiiiiiiiice [07:01] morning [07:10] hey pstolowski [07:37] pstolowski: hey [07:39] PR snapd#8397 closed: cmd/snap-confine/mount-support-nvidia.c: add libnvoptix as nvidia library [07:43] Hey [07:43] Had a very fractured night [07:43] Barely awake now [07:43] Sorry, I will be late today [07:44] I will not be at the desktop sync [07:50] PR snapd#8382 closed: packaging: detect/disable broken seed in the postinst [07:56] pedronis_, zyga, pstolowski, mborzecki I will release 2.44.2 this morning, please let me know if there is anything last-minute you want to have in it [07:57] PR snapd#8226 closed: interfaces/policy/policy_test.go: add more tests [07:57] ack === pedronis_ is now known as pedronis [07:57] mvo: I would really like to have https://github.com/snapcore/snapd/pull/8089 [07:58] PR #8089: features: enable robust mount ns updates [07:58] mvo: but it's up to you [07:59] zyga: hm, it's not on master yet, we should at least get it there [08:00] mvo: yeah but it's pending for a long time [08:00] just shinging some light on it [08:00] zyga: ta [08:00] if it helps our customers enable it via gadget [08:00] and it's well-known to improve [08:01] er, cause improvements [08:01] * zyga is trying to say awake [08:01] zyga: I'm not against it, just saying it needs to land in master first and get reviews etc [08:01] zyga: haha - did you had coffee already :) ? [08:01] +3/-2 - let's review it :) [08:01] mvo: yeah, my wife just brought me a cup [08:02] 2-3 AM, 4AM, 6AM wake-ups (for me) last night, not great [08:03] meh, when i repack cor20 snap the system no longer boots, it stops at initrd-switch-root and complains it cannot resolve /sbin/init even though /sysroot/sbin/init is there (though it points to /lib/systemd/systemd but it's there too) [08:03] mvo: thanks, i don't have anything [08:04] mborzecki: is there any write-up on how to try core20 locally? [08:04] mborzecki: can you please review the tiny branch referenced above (8089) [08:04] zyga: try or build? [08:04] mborzecki: be able to run from source [08:04] mborzecki: boot and see what happens, etc [08:04] mborzecki: (in qemu) [08:05] zyga: i don't think so, but it's the usual thing, grab the model, build image with u-i, run qemu but adding UEFI [08:05] grr during the transition to new laptop i forgot to set git author initially and managed to commit with defaults to all my PRs, and have to fight with CLA checks (need rebasing to fix as it's for past commits, not just latest) [08:05] mmm [08:06] pstolowski: ouch, there's a way to rewrite the author and committer documented in git rebase [08:06] pstolowski: you didn't migrate your HOME? [08:17] zyga: i know of git commit --amend --reset-author which works for last commit; i'm not sure rebase has anything specific like that, but simply rebasing fixes this [08:17] hm, CI fails on 'apt update' [08:18] zyga: yeah i didn't migrate home, i wanted to have clean slate (instead of removing cruft after migration) [08:18] * zyga nods [08:21] aah [08:21] w8, how's that even possible? [08:24] mborzecki: 2 st phase of debugging [08:24] 2st is a new number, between 1st and 2nd [08:24] * zyga should sleep [08:26] * mborzecki trying a differnt rsync incantation [08:28] rsync --wingardium-leviosa --dwim [08:28] it works on 2st try [08:31] yup, much better now [08:47] could somebody retrigger tests for https://github.com/snapcore/snapd/pull/8320 ? [08:47] PR #8320: interfaces: updates to login-session-observe, network-manager and modem-manager interfaces [08:48] PR snapd#8399 opened: interfaces/policy: fix comment in recent new test [08:49] btw. snap debug state is super slow when there's many tasks [08:55] PR snapd#8400 opened: osutil: make the TestWriter() test less racy [09:07] mvo: let's merge it but I left a kind of -1 comment [09:07] zyga: tell me how to do it non-racy :) [09:07] zyga: I thought aobut it but couldn't come up with an idea [09:07] that's much more towards a solution than what we usually do :) [09:07] mvo: no that I disagree it is hard [09:08] zyga: I mean, we test that during the time until the deadline is triggered we get data [09:08] zyga: *if* the system is that slow that there is just no thread scheduled during $time we lost [09:08] mvo: but just doing 0.1 seconds * number_of_times_we_had_to_change_it really feels iffy [09:09] mvo: I get that, but in general a test for more than one concurrent activity must not be based on wall clock time [09:09] zyga: well, $time is a fundamental part of the test, so I don't really know how to do it non-racy, I'm open for ideas though [09:09] mvo: here it seems that way but really what you want is to show a certain sequence of events happens in two threads [09:09] mvo: not that it took this amount of time or that amount of time [09:09] zyga: yes, in general I agree, but this test is precisely about testing wall-clock behavior [09:09] mvo: anyway, I will look [09:10] mvo: I think at some point the tipping point makes us research how to deal with sutuff [09:10] *stuff [09:10] mvo: I can only re-iterate that this is not about that wall clock time _at all_ but about the sequence [09:10] mvo: in a way if we find a way to do this we will get correct *and* faster tests [09:10] zyga: did you look at the test? it's about hitting a deadline and what happens until this deadline is hit [09:11] as arbitrary sleep durations go away [09:11] mvo: I did, perhaps my point is not expressed clearly enough though [09:11] mvo: anyway, I will look [09:11] zyga: don't get me wrong, I agree 100% in general, but I don't see how to do this for this specific case [09:12] hmmm soo, the chooser run nicely in a local terminal, but is stul reading from stdin when running in a uc20 shell, same command line but different effect [09:12] even simple echo '{}' | python3 -m console_conf.cmd.tui --recovery-chooser-mode is stuck [09:13] mborzecki: oh no! [09:13] mvo: mmh [09:13] mborzecki: can you debug log the mode of stdin/stdout [09:14] mborzecki: are you in binary or text mode [09:14] mborzecki: what's the encoding? [09:14] ah w8, maybe that [09:14] mvo: those tests are really testing context.WithTimeout, not our code [09:14] mborzecki: cheers :) [09:14] mvo: so a bit unclear if need all of them, or the cancel one is enough [09:14] or we just need our own base context [09:15] pedronis: fair enough, we could remove it entirely [09:17] pedronis: happy to update the PR to remove it, wdyt? [09:17] mvo: I think the Cancel one is enough, but maybe I'm missing something ? [09:17] we lose the happy case ? [09:18] maybe the happy case is just with a cancelable that we cancel after and not before [09:18] pedronis: I guess so, I don't know more about this particular case then you :) it was a drive-by from my bug triage [09:18] PR snapd#8401 opened: snap: improve TestWaitRecovers test [09:23] mborzecki: the chooser was merged but I will continue with my review [09:23] mborzecki: there are some things that probably ought to be followed upon [09:24] can i make stage-packages: architecture dependent? [09:25] mwhudson: iirc "- on amd64\n - "pkg"" works [09:25] mvo: just found a forum post with a syntax like that, yeah [09:26] mvo: is it documented at all? [09:26] mwhudson: [09:26] mwhudson: https://forum.snapcraft.io/t/architectures/4972 [09:26] mwhudson: I thnk that might be the docs [09:26] mwhudson: actually no [09:26] mwhudson: it's just about the architecture :/ [09:27] like can i say - on amd64, arm64\n - efibootmgr [09:27] damn, it works over ssh, but not in a console [09:27] zyga: ^^ [09:27] mborzecki: what do you get in the console [09:27] mwhudson: yeah, this should work [09:27] mborzecki: binary/text? [09:27] mborzecki: and what encoding if text [09:27] even though i added LANG, TERM and whatnot in the console [09:28] mborzecki: did you add the debugging? [09:28] mwhudson: I couldn't find docs for this, might be worthwhile to check with degville and/or the snapcraft guys if "stage-package:\n- on $arch:\n - pkg1" is documented somewhere and just hard to find [09:29] mvo: i shall whine on the forum! [09:29] mwhudson: :) +1 [09:33] mvo, could you please retrigger tests for https://github.com/snapcore/snapd/pull/8320 and https://github.com/snapcore/snapd/pull/8220 ? [09:33] PR #8320: interfaces: updates to login-session-observe, network-manager and modem-manager interfaces [09:33] PR #8220: interfaces/seccomp: allow passing an address to setgroups [09:41] mvo, degville: what's "try" about in this context? [09:42] abeato: sure, looking [09:43] aaah i found https://snapcraft.io/docs/snapcraft-advanced-grammar [09:49] PR snapd#8320 closed: interfaces: updates to login-session-observe, network-manager and modem-manager interfaces [09:50] mborzecki: mvo: what's the status of #8253 ? [09:50] PR #8253: snap-bootstrap: expand data partition on install [09:50] mvo: do we need it for beta? [09:51] pedronis: thought i +1'ed it already, let me check [09:52] pedronis: there's an open question about explicitly identifying expand/noexpand for partitions at the gadget yaml level, but i don't think we can resolve it now [09:58] mborzecki: I would say let's always expand and if we have a use-case *not* to do that we can add an extension to the gadget.yml (cc pedronis) - irrc this is what I suggested in there too, no? [10:15] mvo: it sounded you suggested the reverse, to have an explicit way to expand [10:15] pedronis: that was my first thinking but that's silly, it will be the common case I think, let me clarify [10:20] pedronis: I updated the PR with my thinking, hope it makes sense [10:29] PR snapd#8378 closed: o/configcore: introduce core config handlers (3/N) [10:38] PR snapd#8402 opened: o/configstate: add backlight option for core config [10:40] PR snapd#8160 closed: overlord/configstate: add backlight option [10:47] PR snapd#8220 closed: interfaces/seccomp: allow passing an address to setgroups [10:57] mvo: soo it works :P [10:57] mvo: need to patch subiquity though [11:00] morning folks [11:00] hey mborzecki seems I missed the deadline on your subiquity PR, but I still would like to do a review [11:01] ijohnson: I'm doing a review as well [11:01] ijohnson: hey, no problem [11:01] I think that's good [11:01] and yes, you should too [11:01] btw. so with patched subiquity, it works, i can trigger the chooser, the UI runs and all, reboot is requested [11:01] but then i get this: https://i.imgur.com/IYb0mgs.png [11:02] it'd noble that it's trying fd0 but i doubt it's gonna succeed anytime soon [11:02] fd0? [11:02] whaat :) [11:02] insert assertion in disk A and pres return to continue ;D ? [11:02] *press [11:02] haha right? :) [11:03] haha wow [11:04] mborzecki: https://github.com/CanonicalLtd/subiquity/pull/655#pullrequestreview-385648364 [11:04] PR CanonicalLtd/subiquity#655: console_conf: implement UC20 recovery chooser [11:05] pedronis: is this https://forum.snapcraft.io/t/how-to-install-and-enable-syslog/15012 the general idea wrt persistent storage config flag? [11:07] post merge review for https://github.com/snapcore/snapd/pull/8370#pullrequestreview-385530483 [11:07] PR #8370: snap-bootstrap: fix disk layout sanity check [11:07] pstolowski: yes, something like that, not sure the directory needs to be created or not [11:07] it needs to be created for auto [11:07] but not sure about persistent [11:10] ijohnson: can you review https://github.com/snapcore/snapd/pull/8389 as well, it would help me make progress [11:10] PR #8389: tests: make session tool way more robust [11:10] sure I was meaning to look at it [11:11] it's not the end of that road [11:11] but it's a step along the way [11:11] pedronis: ok. so we want to support actual systemd storage type as a config option (auto|persistent), and not simplify it/hide it behind e.g. persistent=true|false? [11:16] pstolowski: just persistent=true|false, I don't think auto makes a lot of sense for use on core [11:17] pstolowski: it might be that we implement false as auto [11:17] but not sure , we can discuss in the PR [11:17] so should the auto-import udev rule have KERNEL!="fd*" as well? [11:18] pedronis: ok, thanks [11:18] pstolowski: https://github.com/snapcore/snapd/pull/8402#pullrequestreview-386327449 [11:18] PR #8402: o/configstate: add backlight option for core config [11:19] zyga: thanks [11:24] zyga: approved with a question about busctl [11:24] ijohnson: looking [11:25] FYI, the contributor needs to go through the CLA process [11:25] I didn't find any documentation on that in our HACKING [11:25] should we have some? [11:25] https://github.com/snapcore/snapd/pull/8398 [11:25] PR #8398: usersession/userd: add "slack" to the white list of URL schemes handled by xdg-open [11:27] ijohnson: replied [11:28] thanks [11:32] ijohnson: I added one more comment at the bottom of session-tool spread test [11:32] ijohnson: I wanted to take a break from this monstrosity and look at other things for a while (reviews) but I need to address this in a follow-up [11:32] ijohnson: if you are happy with my comments I will merge this now [11:32] sure no problem [11:32] go for it [11:33] thank you! [11:33] I'll include -- and the comments you mentioned in a follow up [11:33] and now reality will verify my oh-so-alredy-regretful commit message [11:33] "make session tool way more robust" [11:33] PR snapd#8389 closed: tests: make session tool way more robust [11:34] next PR: "make session tool way way way more robust for realz this time (I promise)" [11:34] ijohnson: I will probably use "tests/session-tool: you cannot wait to see how this will break next" [11:35] zyga: "tests/session-tool: this one weird trick makes old linux distros so mad" [11:37] ijohnson: "tests/session-tool: mark as manual and move to tests/masochism" [11:41] I'll go make coffee and return to prepare the big PR with things that have recently landed [11:49] mborzecki: nice, is the patch complicated? [11:50] mvo: no, it's quite trivial, got merged already [12:14] ogra: https://github.com/snapcore/snapd/pull/8329#pullrequestreview-385876868 [12:14] PR #8329: interfaces: allow raw access to USB printers [12:26] mborzecki: is your comment on https://github.com/snapcore/snapd/pull/8377 a +1? [12:26] PR #8377: sandbox/cgroup: add ProcessPathInTrackingCgroup [12:29] PR snapd#8399 closed: interfaces/policy: fix comment in recent new test [12:39] hi, all :-) [12:42] PR snapd#8377 closed: sandbox/cgroup: add ProcessPathInTrackingCgroup [12:43] I didn't write C++ yet. I read a lot about it, around 12 years ago. Now, i help a very very litle bit with the backend of #WirVsVirus. What DB layer is scalable, and NoSQL. What DB should is a choice for this task ? [12:43] Can anybody give a hint ? [12:44] sdhd-sascha: don't use json with manual serialization [12:45] zyga: ? [12:45] sdhd-sascha: we don't have experience with nosql beyond that I think [12:46] i found mongo, with this: https://github.com/datastax/cpp-driver [12:46] But, why use in 2020 a language where i have to free memory manually ?.... Oooh [12:46] Don't know why C++ was the choice... ;-) [12:48] zyga, commented [12:48] PR snapd#8401 closed: snap: improve TestWaitRecovers test [12:49] * ogra curses ... why did nvidia *completely* change with the nvidia-4XX series .. [12:49] ogra: to punish us for not pushing ahead with nvidia-runtime snap [12:49] yeah apparently ... [12:50] mvo: master is over 50 minutes in travis now [12:50] mvo: we're going to see failures soon [12:51] sadly i have a snap where i need to force-preload the nvidia libs (it completely unsets LD_LIBRARY_PATH and is proprietary) ... works just fine with nvidia-3XX and before ... but with 440 ... i get [12:51] ERROR: ld.so: object '/var/lib/snapd/lib/gl/libOpenGL.so' from LD_PRELOAD cannot be preloaded (cannot open shared object file): ignored. [12:51] * sdhd-sascha ogra: ooh, nvidia ;-) [12:52] the lib is there with find and ls and the ldd dependencies dont seem to have changed vs 3XX either ... i dont get why i cant preload it [12:52] ogra: it's classic? [12:52] nope [12:52] strict [12:52] hmm [12:52] zoom-client [12:52] can we ldconfig to look at /var/lib/snapd/gl [12:53] and drop that silly SNAP_LIBRARY_PATH thing? [12:53] well, all these vars get unset by the zoom binary ... i cant make any use of them anyway [12:53] ogra: hence > can we ldconfig to look at /var/lib/snapd/gl [12:53] ogra: so no variables at all but it _still looks_ [12:54] ogra: we could even be more ... say, creative [12:54] and have a new mount point [12:54] /app [12:54] hmm [12:54] and mount current revision of the app there [12:54] and have a fixed ldconfig [12:54] does ldconfig actually load anything ? i thought it only resolves deps ... i need them in ram before the app starts [12:55] so we take libs from [/var/lib/snapd/lib/*, /app/lib/$triplet, regular locations] [12:55] or do you mean combine ldconfig and LD_PRELOAD ? [12:55] ogra: it finds them [12:55] e.g. copy a .so to /usr/local/lib [12:55] and run a binary that needs it [12:55] then ldconfig [12:55] then run it again [12:55] we could essentially tell the linker, here they are [12:55] yes, but zoom doesnt if it isnt loaded already [12:55] ? [12:55] this is entirely unrelated to zoom [12:56] well, i can try ... [12:56] it would be some work [12:56] but as a spike to see if it works [12:56] I think it could [12:56] repackage core snap to have different ldconfig [12:59] ijohnson, fyi I have UC20 running yesterday, the crash happened because I was greedy and was no giving enough memory to kvm - I've found I need at least 2GB. [12:59] ah yes that's another thing I forgot to tell you :-) [13:00] 2GB ... nobody thinks of the netbooks ! [13:00] ijohnson, and the problem with the eth was related to the image, the one I built locally had some problem, maybe I need edge ubuntu-image [13:00] the cdimage one worked in the end [13:19] abeato: ah that's great news [13:21] ijohnson, yeah, really happy to start testing uc20 :) - thanks for your help [13:27] https://github.com/google/thread-weaver [13:29] moving snapd to java now ? [13:29] core 22j [13:29] hah [13:31] we can finally support arm7tdmi [13:32] yay, finally going for ARMv4 :) [13:33] * ogra digs in the box for old nokia phones to run core on [13:43] PR snapd#8219 closed: interfaces: use udev backend if udev socket exists [13:46] PR snapd#7963 closed: xdgopenproxy: forward requests to the desktop portal <⛔ Blocked> [13:52] mborzecki: follow up https://github.com/snapcore/snapd/pull/8403 [13:52] PR #8403: sandbox/cgroup: avoid making arrays we don't use [13:52] PR snapd#8403 opened: sandbox/cgroup: avoid making arrays we don't use [14:18] hey I fixed the usb audio thing, yet again my usb hub seems to be causing problems :-/ plugging the device directly into my desktop makes it work fine [14:18] * ijohnson is happy with high fidelity audio output now [14:35] PR snapd#8404 opened: client: increase timeout in client tests to 100ms [14:54] zyga: you will love this test failure https://paste.ubuntu.com/p/yKy6G2x59x/ (in a PPA) [15:01] PR snapd#8089 closed: features: enable robust mount ns updates [15:06] mvo: looking [15:06] Mvo: heh [15:06] zyga: I will look later [15:06] zyga: in a meeting right now [15:07] * zyga hugs mvo [15:07] zyga: but yeah, lots of fun, the PPA is super slow right now [15:07] I’m sorry, I know we are in tough times [15:19] back now [15:19] * zyga took a walk [15:20] I guess people are renting dogs to walk now [15:20] I see more people with dogs than I ever did here [15:28] * zyga stares at tig output [15:28]