[12:48] <AsciiWolf> ahayzen, Hi, thanks for the Steam update! :-) By the way, could you look at this comment about merging steam-installer into Debian from Simon McVittie: https://salsa.debian.org/games-team/steam/-/merge_requests/2#note_209734 ? Thanks!
[12:59] <cpaelzer> Laney: you had this neat autopkgtest command that was spawning the test on s390x canonistack
[12:59] <cpaelzer> Laney: I have adapted that for my use, but it is stumbling over the keys (yes I have used mine :-) )
[12:59] <cpaelzer> I see in the debug output that nova correctly has my usual keypair
[13:00] <cpaelzer> e.g. on "openstack server show"
[13:00] <cpaelzer> I have modified /usr/share/autopkgtest/ssh-setup/nova to keep the failing instance around for debugging
[13:00] <cpaelzer> but despite nova telling me it has used my keypair I can't use it to log in
[13:00] <cpaelzer> is there a bell ringing what I should look out for?
[13:01] <cpaelzer> or is this down to checking what the script does step by step ...
[13:01] <Laney> not really, if you look in the console-log you should see cloud-init installing the key
[13:01] <Laney> eliminate autopkgtest completely and use 'openstack server create' to make a server using that keypair, see what happens then
[13:03] <cpaelzer> $ openstack server create --key-name paelzer_canonistack-bos01 --flavor m1.small --image ubuntu/ubuntu-bionic-daily-s390x-server-20201123-disk1.img cpaelzer-s390x-systemd-manual
[13:03] <cpaelzer> works fine and ssh login with key like -i ~/.canonistack/paelzer_canonistack-bos01.key works well
[13:04] <Laney> does openstack server show list the correct key_name?
[13:04] <cpaelzer> yes in both cases autopkgtest/manual it lists the same
[13:04] <Laney> I don't know, sorry, I can't really see how this can be being broken by autopkgtest
[13:05] <cpaelzer> np, thanks for thinking with me about it
[13:05] <Laney> but we didn't see any actual output, so perhaps there would be a missed clue in there
[13:09] <cpaelzer> I've sent you the output of the failing run for comparison with your working setup
[13:30] <cpaelzer> Laney: thanks, seems to work now
[13:30] <cpaelzer> corrent workaround is odd but working, over the next uses of it I can differentiate what of that is really needed
[13:31] <cpaelzer> step 1: create an instance manually; step 2: run autopkgtest with the same name => will create a new instance with the same name (different UUID) but things seem to work (test runs)
[13:31] <Laney> O_O
[19:04] <tdaitx> Laney: I just updated bug 1903929, those triggers are caused by openjdk uploads (8, 11, 15, 16, 17) but even so it only runs the test with the default-jdk (which is in fact correct, but triggering should also only be done when the openjdk from default-jre/-jdk is uploaded)