On Fri, 2020-04-03 at 14:58 +0200, Normand wrote:
On 4/3/20 2:52 PM, Normand wrote:
> Hello Adam,
>
> I identified two tests that randomly failed on
>
openqa.stg.fedoraproject.org (1)&(2)
>
> We do not have such problem in our IBM intranet openQA.
>
> So I am wondering if difference could be related to
> configuration/software versions differences of P9 hosts.
>
> Our P9 is installed with f31, but no updates since 03/17
> ===
> $rpm -qa |grep -iE ^'kernel-header|qemu-kvm|slof' |sort
> kernel-headers-5.5.8-200.fc31.ppc64le
> qemu-kvm-4.1.1-1.fc31.ppc64le
> SLOF-0.1.git20191022-1.fc31.noarch
> ===
Oops above if for openQA server, the P9 is:
===
kernel-headers-5.4.7-200.fc31.ppc64le
qemu-kvm-4.1.1-1.fc31.ppc64le
SLOF-0.1.git20190114-2.fc31.noarch
===
> From last autoinst-log.txt and serial0.txt it seems you have same slof
> older kernel and specific qemu.
> Do you have other specific changes ?
> ===
> Linux 5.3.16-300.fc31.ppc64le
> qemu-4.2.0-1.fc31.infra
> SLOF release 20191022
Sorry, I forgot to reply to this mail.
Currently all the workers are booted to 5.3.16-300.fc31.ppc64le and
have SLOF-0.1.git20191022-1.fc32.noarch and
qemu-4.2.0-1.fc31.infra.ppc64le . As the package name implies, that's
an infra build (there's a special repo on the Fedora infra boxes with a
few 'side builds', I can't recall why we have qemu 4.2.0 on f31, but we
do). I think I updated SLOF because the newer qemu requires a newer
SLOF.
I can update the kernels and reboot the boxes to see if that helps
anything, changing qemu and SLOF might be a bit trickier...
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net
http://www.happyassassin.net