[07:20] morning [07:51] PR core18#186 closed: Handle PPAs being served from ppa.launchpadcontent.net [07:57] hi mborzecki, mvo [07:58] hey [08:07] PR snapd#11181 closed: usersession/userd: support Cloudflare WARP URL scheme [08:08] mardy: do the microk8s guys know whether `k` is needed there too? https://github.com/snapcore/snapd/pull/11286#discussion_r798416451 [08:08] PR #11286: i/b/kubernetes_support: open access to NVIDIA GPUs [08:08] or whether the PR in its current form solves their problem [08:09] PR core20#129 closed: Handle PPAs being served from ppa.launchpadcontent.net [08:16] mborzecki: I didn't hear any feedback [08:18] morning [08:25] pstolowski: hi! [08:38] pstolowski, mborzecki: maybe you can help me: I made these changes (https://paste.ubuntu.com/p/TPFgn3pprR/) and now the tests are failing like this: https://paste.ubuntu.com/p/cbRkzB8tTZ/ [08:38] do you know what I need to do in order for the task to advance its state? (a call to settle() is already present) [08:42] mardy: are you sure that's the only change? [08:43] mardy: also, is thare Retry returned by any task? I usually add a loop to print all tasks and their statuses in such cases [08:44] mardy: i've added a dumpTasks() helper in managers_test.go, maybe you can use that to print the task states [08:45] maybe you can also grab the logger buffer and print it out? [08:49] * mardy tries [08:54] here's with dumpTasks: https://paste.ubuntu.com/p/VYfCFYH8k4/ [08:54] mborzecki: do you have any example where we print the logs? [08:57] mardy: look for logbuf, restore := logger.MockLogger(); also task.Log() and change.Log() [08:58] mardy: or try something like this: https://paste.ubuntu.com/p/3m5D7SsxJk/ [08:59] mardy: auto-connect may retry if there is a conflict [08:59] that would be 1st suspect [09:00] Retry means task remains in "Doing" and is reconsidered some time later [09:08] pstolowski: but how can I trigger a retry? [09:09] mardy: by resolving the conflict [09:10] mardy: but hold on [09:11] indeed, after applying mborzecki's patch I see that the last log line is "task 2 err: task should be retried" [09:12] good morning [09:12] zyga[m]: hi! [09:12] hey zyga[m] ! [09:13] mardy: with your changes, is IsClean() going to return false if task is no Ready yet? [09:16] pstolowski: yes (unless I did something wrong) [09:16] pstolowski: "clean" should mean "ready and the cleanup handler ran (if present)" [09:17] mardy: got it, makes sense [09:22] mardy: i don't understand why TestAutoConnectDuringCoreTransition is failing with conflict after your changes, seems like maybe the task is not considered clean. I'd add debug in checkAutoconnectConflicts() and printy IsClean and Ready state and just run this test [09:23] pstolowski: OK, will do; but I don't know how to run a single test :-) [09:24] mardy: go test -check.f TestAutoConnectDuringCoreTransition (or another substring that matches testsuite/testnames) [09:29] pstolowski: thanks! So, the task is ready (task.Status().Ready() returns true), but task.IsClean() is false. I wonder if the logic I added in taskrunner is wrong then... (https://github.com/snapcore/snapd/pull/11321/files#diff-d38844f5dd133b4b7dedd1f94bd2e344ae8762fbb756e4e155ec7276e9a88b7a) [09:29] PR #11321: overlord: fix issue with snapshot restoring on CentOS [09:30] maybe it has something to do with task <-> change status sync [09:32] oh, the change associated to the task is not clean nor ready [09:35] hey mardy, mborzecki and pstolowski - good morning! [09:36] do I understand correctly, that a change can have many associated tasks? [09:36] hey mvo ! [09:36] mardy: yes, change has many tasks. but tasks can only belong to a single change [09:37] pstolowski: but then it seems to me that the check I'm doing here is wrong, and that I should check the status on the task, and not on the change: https://github.com/snapcore/snapd/pull/11321/files#diff-d38844f5dd133b4b7dedd1f94bd2e344ae8762fbb756e4e155ec7276e9a88b7a [09:37] PR #11321: overlord: fix issue with snapshot restoring on CentOS [09:38] though given that it was Samuele to suggest it like this, I bet he had his reasons [09:42] mardy: looks good to me, otoh maybe checking just the task would be semantically the same in this case. but by definition change is only clean if all tasks are clean (just like Ready status), so nothing wrong with it afaict [09:56] Is it still the case that *gadget* snaps (small additions on top of the official Raspberry Pi gadget like splash screen image and snap configuration/connections) require a paid brand store upload + auto-update? Just to clarify, don't want to open a discussion about business models 🙂 [10:03] tobias_: a brand store in order to publish gadget snaps yes afak, connections/configuration can be put in gadget.yaml regardless of that, i'm not sure what you mean by auto-update though [10:04] tobias_: as for brand store, if it's an open source project something can be probably arranged, iirc gnuradio has a brand store [10:20] mborzecki: Yeah sorry, I meant publishing a gadget snap so that it can be refreshed (updated) from the store on deployed devices. I currently build images with a local gadget, but e.g. when Pi kernel assets moved from the gadget to the kernel snap, existing devices could no longer refresh the official pi-kernel snap until users copied my newer gadget snap to their device and ran a manual refresh. Same with the migration from mir-kiosk to [10:20] ubuntu-frame – requires a manual gadget refresh. We're a small open source project and have mostly non-commercial home users at the moment, but do intend to sell B2B support licenses for the software to keep the lights on 🙂 [10:20] Thanks for the clarification! [10:57] PR snapd#11351 closed: tests/nested/manual/remodel-cross-store,remodel-simple: wait for serial [11:32] PR snapd#11352 opened: packaging/fedora: sync with downstream, packaging improvements [12:53] PR snapd#11353 opened: systemd: do not reload system when enabling/disabling services [13:33] PR snapd#11335 closed: gadget/install: add unit tests for install.Run() [13:38] PR snapd#11345 closed: tests/main/snap-system-key: reset-failed snapd and snapd.socket [14:12] PR snapcraft#3634 opened: dependencies: missing library resolution (CRAFT-422) [14:57] PR snapcraft#3628 closed: Autotools Plugin (v1): Fix fatal crash when running autogen.sh or bootstrap [15:13] PR snapd#10427 closed: interfaces/cups: add cups-socket-directory attr, use to specify mount rules in backend [15:43] PR snapd#11320 closed: osutil/mkfs: Expose more fakeroot flags [17:29] PR snapd#11354 opened: gadget: identify/match encryption parts, include in traits info [17:37] PR snapcraft#3629 closed: sources: make submodule fetching configurable [18:04] PR snapd#11341 closed: tests: updated the documentation to run spread tests using external backend [18:49] PR snapd#11355 opened: Changes necessary for ubuntu-image to support preseeding extra snaps in classic images [18:53] PR snapcraft#3632 closed: 20.00