[04:40] PR snapd#8374 opened: .github: register a problem matcher to detect spread failures [05:39] PR snapd#8375 opened: github: always run the "Discard spread workers" step, even if the job fails [05:43] morning [06:15] hmm core20 hangs in spread tests again? [06:16] looks like it's hitting job kill-timeout when rebooting during prepare [06:21] hmm perhaps https://github.com/snapcore/snapd/pull/8373 fixes that [06:21] PR #8373: tests/lib/prepare.sh: use only initrd from the kernel snap <⚠ Critical> [06:26] o/ [06:26] morning zyga [06:26] hey james [06:27] zyga: here's a simple PR for the github actions CI: https://github.com/snapcore/snapd/pull/8375 [06:27] PR #8375: github: always run the "Discard spread workers" step, even if the job fails [06:27] make the cleanup run even if the tests fail [06:27] super nice [06:27] thank you James [06:27] I didn't know about the implicit if [06:28] I also had a go at adding a problem matcher: it adds the error to the annotation list, but doesn't jump to the log line as I thought it did: https://github.com/snapcore/snapd/pull/8374 [06:28] (maybe I was misremembering how that worked) [06:28] PR #8374: .github: register a problem matcher to detect spread failures [06:28] I love this :) [06:29] little by little but even this is so much better than what we had before [06:29] that's fantastic [06:29] I think we could move more stuff over this week [06:29] let's see what we can get [06:30] zyga: jamesh: hey [06:30] hi mborzecki [06:32] wish the uc20 spread job actually worked [06:32] mvo: hey [06:32] hey mvo [06:33] * zyga is pre breakfast still so will go away for a while [06:33] zyga: Longer term, it might make sense to have Spread output error annotations when it thinks it is running under the Github runner [06:33] it could provide extra info like the task file name [06:34] jamesh: hopefully the runner makes it easy to detect such scenario [06:34] jamesh: yeah, I think that would be great, spread has specific travis features, it could have github features [06:34] I think the goal should be to phase out travis [06:34] so that nobody is stuck waiting [06:34] then optimize [06:34] otherwise it's ugnly job level env vars again or somesuch :/ [06:34] mborzecki: there are environment variables it is documented to set, similar to Travis [06:36] https://help.github.com/en/actions/configuring-and-managing-workflows/using-environment-variables says it will always set GITHUB_ACTIONS=true [06:38] ha nice [06:38] hey mborzecki and zyga ! [06:39] how are you doing? it's quite cold today [06:39] yesterday was both +18 and -2 at night (and snow) [06:40] zyga: it was ~13C before noon, then 2 with a snow/rain mix in the afternoon here [06:40] felt like the afternoon was a completely different day [06:51] breakfast time [06:53] PR snapd#8375 closed: github: always run the "Discard spread workers" step, even if the job fails [06:58] PR snapd#8365 closed: seed: add Info() method for seed.Snap [07:03] morning [07:04] hey pstolowski - good morning [07:04] fwiw lxd tests are broken right now [07:04] not sure why yet, but it seems pretty consistent [07:12] PR snapd#8374 closed: .github: register a problem matcher to detect spread failures [07:18] mvo: thanks for doing the merges. I think the problem matcher is about as good as it can get: further improvements would probably need to happen in Spread itself [07:19] jamesh: yeah, I was thinking that we might need to modify spread [07:19] jamesh: that's fine, thanks again for this, it's so much better than what we had before [07:19] mvo: e.g. it would be nice to tie the error to the task.yaml file, but the problem matcher can only use information present in the error message [07:24] power outage :/ moved to my parent's [07:31] travis job failed again in #8373 [07:31] PR #8373: tests/lib/prepare.sh: use only initrd from the kernel snap <⚠ Critical> [07:31] mborzecki: probably the lxd issue [07:31] mborzecki: also, does it have two reviews? [07:31] kill-timeout reached after mar300623-305765 (google:ubuntu-core-20-64:tests/regression/) reboot request [07:32] mvo: all test suites failed on uc20 [07:32] (on prepare) [07:32] mborzecki: uh, no that then :) [07:32] mborzecki: right, because of the kernel/initramfs issue? [07:32] wish the console was easily accessible [07:32] mvo: yeah [07:32] mborzecki: does it only fail in gce or also with qemu? [07:33] mvo: haven't tried qemu yet, i'll try to obtain the image from gce [07:33] re [07:33] mborzecki: yeah, qemu should give a lot more visiblity into this, so we still don't know what is going on :( ? [07:33] sorry, it's hard to organize kids still [07:58] mvo: hi, #8325 has a conflict now also as we know needs a bit more work [07:58] PR #8325: snap-bootstrap: copy auth data from real ubuntu-data in recovery mode [07:59] pedronis: thanks, look today (many meetings this morning unfortunately) [08:29] any clue why it's no longer possible to connect via ssh to a spread node? [08:30] spread can do it, but once i got the debug shell, i cannot establish any new ssh connections [08:39] mborzecki: hi, I made a comment in the stand up notes about a thing that landed Friday, also did review some of your open PRs [08:39] pedronis: thanks! [08:40] re [08:42] pedronis: also wondering whether we should support GET on /v2/systems/{label}, but there does not seem to be any potential use cases atm [08:42] mborzecki: we can add it later, but yes would leave it alone for now [08:42] ack [08:48] PR snapd#8376 opened: daemon: make POST /v2/systems/