On Sat, 2016-03-05 at 21:24 +0000, Fedora compose checker wrote: > Missing expected images: > > Cloud raw-xz i386 > Atomic raw-xz x86_64 > Cloud raw-xz x86_64 > > Images in this compose but not 24-20160304.n.0: > > Design_suite live i386 > Design_suite live x86_64 > > No images in 24-20160304.n.0 but not this. > > Failed openQA tests: 61 of 78 So I've spent the last few days swearing at fedmsg configuration, but here's roughly what's going on... The glibc issues seem to be more or less resolved. Server DVD install works, and the post-install boot that's part of the basic install test works also, but it seems like when the post-install tests run (which start from a snapshot taken right after the basic install test finishes) the system fails to boot. Not sure why, have to look into that. Workstation live boots but the installer doesn't run; this is likely https://bugzilla.redhat.com/show_bug.cgi?id=1313098 ;, assuming the openQA VMs are getting Wayland. KDE live boots, the installer starts up, then crashes: https://openqa.fedoraproject.org/tests/7675/modules/_boot_to_anaconda/steps/5 I'm not sure why, we'll have to reproduce that one locally too. x86_64 network install tests almost all fail because of https://bugzilla.redhat.com/show_bug.cgi?id=1313949 and https://bugzilla.redhat.com/show_bug.cgi?id=1313957 - Pungi 4 is currently including i686 kernels in the x86_64 Everything repo (which it should not) and anaconda is installing the i686-PAE kernel when doing an x86_64 install (which it probably shouldn't). So the install works, but the system fails to boot because it's using a 32-bit kernel and a 64-bit userland and that ain't gonna go well. Oddly enough a couple of tests seem to use a 64-bit kernel and work, dunno why that is. All 32-bit images seem to be not booting again; we'll have to look into that too (and see if it's reproducible on other systems). There are a couple of openQA screenshots that need redoing as well, I think, but it's a bit hard to keep track of among the chaos. There is also an openQA scheduler bug which causes the 'universal' tests often to run with the wrong image (they should always use the Server DVD when it's available but in fact they're currently using whichever 'eligible' image happens to show up first in the metadata, which is often a network install image), which also isn't helping; I have a fix for that which just needs sign-off. It's gonna be a fun week ahead! I think I'm mostly done wrangling the process changes (famous last words), so I'm hoping to have time to look into the actual *bugs* at last...of course, if folks can help out figure out all the bugs, that'd be great :) -- Adam WilliamsonFedora QA Community MonkeyIRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . nethttp://www.happyassassin.net -- test mailing list test@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe: http://lists.fedoraproject.org/admin/lists/test@xxxxxxxxxxxxxxxxxxxxxxx