[05:37] morning [06:48] mvo: hey [06:51] good morning mborzecki [06:53] mvo: i've pushed an update to arch, and working on updates for fedora [06:55] mborzecki: nice, thank you [07:05] morning [07:07] pstolowski: hey [07:32] mvo: it looks like the edge channel for the core snap hasn't published for a while, and doesn't seem to reflect master. Is that a known problem? [07:37] jamesh: we need jdstrand to unblock publication, the review tools flaged edge incorrectly [07:37] jamesh: or some other store reviewer for that matter [07:42] PR snapd#8917 closed: osutil/disks: add mock disk and tests for happy path of mock disks <⛔ Blocked> [08:36] good morning [08:36] mvo I'm filing the paperwork now [08:37] PR snapd#9088 opened: cmd/snap-preseed: use snapd from the deb if newer than from seeds [08:39] pedronis thank you! :) [09:22] mvo: I approved that core revision [09:22] pedronis: thank you [09:31] pedronis: there might be more to approve https://dashboard.snapcraft.io/snaps/core/revisions/9810/ [09:33] there's a lot of them [09:33] let' see [09:34] pedronis: yeah, I think you will have to approve 5 or so and then it should have recovered (one approval for each arch) === Haxxa is now known as Guest846 [09:47] mvo: I think now I approved all archs once, let's see [09:50] pedronis: thank you! fwiw, I'm working on 8982 right now, it needs some tweaks for not leaking FDs in the error case now that the export is split [09:53] morning folks [09:54] good morning ijohnson [09:54] hey mvo [09:55] mvo: thanks for merging 8917, but I don't suppose you saw the squash merge label for that PR ? [09:55] it's not a big deal, just a little messy now IMHO [09:57] mvo: the subsequent revisions are getting green on their own now [09:57] * pedronis lunch [09:58] ijohnson: oh no, sorry for that [09:58] pedronis: great, thank you [10:00] mvo: no worries, just my OCD wanting to have a clean git history without all my silly "merge feature X into feature Y" and "merge master into feature Y" commits [10:04] hey ijohnson ! [10:04] hey pstolowski [10:04] did you try out the ubuntu-bartender thing at all? no worries if you didn't, happy to show you how I debugged things with that in a little bit [10:14] ijohnson: no, not yet, since you mentioned you needed manual hacks i assumed it makes sense to wait for you [10:15] pstolowski: sure, do you wanna join the SU HO in 45 minutes ? [10:15] ijohnson: yes, sure [10:15] sounds good [10:20] The access check attributes on snapFileCmd ("/v2/snaps/{name}/file") don't really make sense. From what I can see, the polkit action it specifies will never be used [10:55] ijohnson: i've a small interruption here, may need 15 or so, can i ping you when ready for HO? [10:55] sure no worries [10:57] o/ [10:57] hey zyga how are you feeling ? [10:57] I'm in the office, training my legs to stand and ... just do that [10:57] much better, stronger by the day [10:58] nice that's really great to hear [10:58] still pathetically weak but making progress :) [10:58] starting Wednesday I should be able to sit for more than half an hour, or so the doctors tell me [10:58] how have you guys been? [10:58] things are going well I think, not too many fires [10:59] that's great [10:59] how are tests? [10:59] is everything working okay? [10:59] I rebooted our instance after boot loader patches went out [10:59] not too bad actually, not that much "actions missing in action" kinda thing [11:00] still some red from uc20 prepare and store 408s randomly [11:00] but the tests seem to be syncing up properly which is good [11:01] ok [11:01] that's great to hear [11:01] my goal for today is to sent some docs to mvo [11:01] (done) [11:01] and to exercise (in progress) [11:01] nice [11:01] and to see a bit more about my branches (bonus) [11:01] I'll try to address the feedback you gave me [11:01] I'm off until Thursday [11:01] but I'll try to help a little [11:06] hey zyga, glad to hear your condition is improving [11:06] yeah, :-) [11:07] my daughter told me that I move like my father (who is old and moves slowly :) [11:08] ouch [11:27] * zyga break [11:27] tired o/ [11:41] ijohnson: sorry, it took a bit longer; can we talk now? [11:41] pstolowski: sure, give me 5 minutes I want to start flashing my SD card, it takes a bit to finish [11:41] ijohnson: sure [11:46] pstolowski: ok ready joining the HO now [11:46] ijohnson: ok [12:01] pedronis: 13 files changed, 315 insertions(+), 51 deletions(-) just the writer & bits in gadget/install, devicestate install and stubs in boot [12:01] pedronis: not great, but not horrible either [12:02] the tests inflate the diff a bit [12:02] mborzecki: sounds ok [12:09] * pstolowski lunch [12:33] PR snapd#9089 opened: many: introduce content write observer, install mode glue, intial seal stubs [13:34] pff chrome died at the end of standup === mborzeck1 is now known as mborzecki [13:47] PR core20#79 closed: Add secureboot-db package, try #2 === mborzeck1 is now known as mborzecki [14:00] mvo: I'm looking and it looks like Samuele approved core so you are set. you said it was flagged incorrectly, but it dropped a file. I think the review-tools dtrt? [14:00] jdstrand: yeah, all good, thanks! [14:04] jdstrand: (in a meeting, will get back to you) [14:06] mvo: no worries, just wanted to make sure everything was ok. thanks! [14:08] jdstrand: yeah, I think the issue was that edge go a (much older) core and then we got the new version. this lead to the review-tools flagging I think [14:09] mvo: yes, that can definitely happen as the review-tools (and the store at the point that the tools are run) have no concept of 'edge' or 'stable' [14:10] mvo: we now have a role rotation for store manual reviews, so we should in theory be more proactive, but you can always reach out :) [14:19] hmm 2020-08-04T13:37:42.0599670Z Failed for "gopkg.in/yaml.v2" (failed to clone repo): exit status 128 [14:21] 2020-08-04T13:28:36.1975163Z fatal: unable to access 'https://github.com/snapcore/go-gettext/': transfer closed with outstanding read data remaining [14:21] is github going down again? [14:28] pstolowski: I added a question to #9087, we can chat tomorrow if it needs discussion [14:28] PR #9087: o/snapstate: check available disk space before downloading a snap on install (4/N) [14:28] pedronis: ok [14:30] pedronis: hmm i didn't think of multi snap install, i think you're right [14:30] pstolowski: well also refreshes, this code is used by refreshes too [14:31] there's some questions there as well [14:31] pstolowski: we we should have a chat [14:31] tomorrow or when it makes sense for you [14:32] pstolowski: I mean, if you are busy with other things, this can probably wait a bit [14:32] Issue core20#80 opened: networking does not persist in a reboot loop on arm64 pi4 [14:33] pedronis: sure, let's see how it foes with xenial+squashfs problem on preseeding [15:06] pstolowski: about seeeding error, one option could also be to look at in Error changes, when we do the in flight check in ensureSeeded [15:07] but is not optimal [15:08] pedronis: indeed, that would work. but yes [15:12] pstolowski: though maybe it's just enough to go over the changes in the api code [15:13] pedronis: we exit on task errors when preseeding. snap-preseed could inspect state by itself if exit status != 0 [15:13] pedronis: it was commit 24d77f474dc53846a16289d19721292d16832cd8 [15:13] pstolowski: this is not about preseed, it's about finishing the seeding [15:13] pedronis: ah you're right [15:15] yea, I think the option I prefer is just going over the changes in getSeedingInfo if seeded is false [15:16] it's very self contained and less risky [15:17] pedronis: yes that's ok [15:27] pedronis: given the email from CPC, should we adjust `snap debug seeding` output to be more "yaml", I see for example right now if we can't calculate a time and output "-", that makes it invalid yaml [15:27] * cachio lunch [15:28] pedronis: because otherwise it's all valid yaml [15:29] but maybe there are other edge cases I don't see that are not valid yaml [15:33] PR snapd#8998 closed: tests/cmd/snap-bootstrap/initramfs-mounts: add test case for empty recovery_mode [15:54] ijohnson: sorry, I forgot, we have special support in the unicodeMixin to emit things that are valid yaml for out patterns [15:54] mmmm, so the current code is buggy then [15:55] ijohnson: look at .dash in cmd_info.go for example [15:55] ah so if we use "–" instead of "-" then yaml is happy [15:55] yeah I see it now [15:56] well, it's a bit more complicated than that, but yes, there are helpers to pick things [15:56] pedronis: ok, given this is it okay if I respond to CPC's email with instructions on how to use snap debug seeding with the caveat that we found a bug just now where it's not always valid yaml ? [15:57] well, we need to add error too [15:58] sure, but that won't change the formatting will it? [15:58] I was assuming if seeding error'd then we would just exit with non-zero exit code and add something else to the yaml output [15:58] yes, but I would prefer to send one email with all the bits [15:58] instead of many [15:58] ok, that's fine [15:59] I can file a PR to fix the unicode yaml issue then [15:59] basically I we should first get closer to what we want in .4 [15:59] and then mail them [16:00] yes makes sense I will not email them [16:14] PR snapd#9090 opened: cmd/snap/debug/seeding: use unicode for proper yaml <⚠ Critical> [16:14] pedronis: pstolowski: I opened ^ which fixes the yaml issue for snap debug seeding [16:15] ijohnson: thx, pstolowski: I'm looking into adding seed-error [16:15] nice [16:15] pedronis, ijohnson thanks! [16:44] PR snapcraft#3238 opened: db: introduce generalized datastore [16:57] jdstrand: regarding https://bugs.launchpad.net/snapd/+bug/1889695, do you anticipate you will have a fix ready in the next few days ? we would like to include that in the next point release of snapd 2.45.4 [16:57] Bug #1889695: [k8s-support] permission denied error when calling readlink on /proc/1/ns/pid [17:00] ijohnson: I could fix that, but they need the other things I'm doing for microk8s. I won't have that done for 2.45.4 and was planning 2.46. if it goes fast, *maybe*, but I doubt it will go fast [17:04] ijohnson: I have code, I will post a PR after dinner [17:19] jdstrand mmm I see, so the issue is that I assume a new version of various $CONTAINER_TECHNOLOGIES is what is causing the problems? [17:20] jdstrand also I don't know how critical the k8s thing is for field, I would recommend asking Tony about the priority of that [17:28] ijohnson: I proposed #9091 [17:28] PR #9091: cmd/snap: display the error in snap debug seeding if seeding is in error [17:28] pedronis ack I will take a look [17:29] PR snapd#9091 opened: cmd/snap: display the error in snap debug seeding if seeding is in error [17:59] ijohnson, hey [18:00] I see the cgroup test failing in debia [18:00] https://paste.ubuntu.com/p/9rk24R2265/ [18:00] I tried and it is easy to reproduce [18:01] do you know something about that? what could be the route cause? [18:01] Let me look [18:03] cachio: no I don't know what that issue is [18:04] it is failing in debian sid [18:04] not sure if it is because an update on a dependency [18:08] ijohnson: well there are a number of things. new workloads aren't working right and I need to investigate the best fix for that, then there are various access issues, and then there is the different behavior depending on the kernel version [18:10] jdstrand ack makes sense [18:10] I've released 2.45.3.1 to openSUSE [18:10] * zyga gets out of office [18:10] cachio: yeah I kinda suspect an upstream change in Debian here [18:11] ijohnson, yes, I don't see any change in our side which could be affecting that [19:38] * cachio afk