On Sun, 2016-01-24 at 13:55 -0700, Chris Murphy wrote:
On Sun, Jan 24, 2016 at 10:02 AM, Adam Williamson
<adamwill(a)fedoraproject.org> wrote:
> On Sun, 2016-01-24 at 06:47 +0000, Fedora compose checker wrote:
> > Missing expected images:
> >
> > Kde disk raw armhfp
> > Cloud_atomic disk raw x86_64
> > Kde live i386
> > Kde live x86_64
> >
> > No images in this compose but not Rawhide 20160122
> >
> > Images in Rawhide 20160122 but not this:
> >
> > Cloud_atomic disk qcow x86_64
> > Cloud_atomic disk raw x86_64
> >
> > Failed openQA tests: 8 of 63
> >
> > ID: 4040 Test: x86_64 universal server_multi@uefi
> > URL:
https://openqa.fedoraproject.org/tests/4040
> >
> > ID: 3980 Test: x86_64 universal server_multi
> > URL:
https://openqa.fedoraproject.org/tests/3980
>
> The other fails look legit, but this is failing because the disks
> stopped being 'sda' and 'sdb' and started being 'vda' and
'vdb'. I'd
> say this was an os-autoinst 4.3 change, but the odd thing is, it didn't
> happen when we had 4.3 on staging. So it may be a kernel 4.5 thing.
> I'll take a look at it in more detail tomorrow.
Maybe a change in default bus type to VirtIO somewhere in the libvirt
multiverse? I see vdX if the disk bus is VirtIO, but it's sdX if it's
IDE, SATA, or SCSI.
No, openQA doesn't use libvirt. It calls qemu directly.
...oh, I know why this is! This is the test that wasn't running at all
on 4.3 in staging because of some stuff about how openQA handles disks
(kinda a long story, not worth going into), we fixed that by changing
the test config a bit, but I didn't think ahead enough to realize we'd
wind up with virtIO not SATA disks. All makes sense now, I'll fix it
tomorrow.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net
http://www.happyassassin.net