action #125798
Visual differences in GRUB menu on different x86_64 UEFI workers
0%
Description
Here are 5 different LTP jobs booting the exact same UEFI/SecureBoot QCOW image on different workers:
https://openqa.suse.de/tests/10651590 openqaworker16:14, GRUB needle mismatch
https://openqa.suse.de/tests/10658203 openqaworker16:18, pass
https://openqa.suse.de/tests/10659306 openqaworker16:7, GRUB needle mismatch
https://openqa.suse.de/tests/10659346 openqaworker17:12, GRUB needle mismatch
https://openqa.suse.de/tests/10659359 worker9:11, pass
It appears that GRUB menu size changes depending on not just the worker but also specific worker slot.
Possibly related to poo#114523 but this time it's happening on x86_64.
Related issues
History
#1
Updated by okurz 3 months ago
Updated by - Status changed from New to Feedback
- Assignee set to okurz
- Target version set to Ready
Please fill the description according to the ticket template https://progress.opensuse.org/projects/openqav3/wiki/#Defects
I doubt this is related to the infrastructure. The worker slots on the same machine running just openQA worker instances in qemu are just different processes of the very same software. I compared autoinst-log.txt from the first fwo URLs you shared and I found:
- https://openqa.suse.de/tests/10651590/logfile?filename=autoinst-log.txt :
qemu-system-x86_64 -device virtio-vga,edid=on,xres=1024,yres=768 - https://openqa.suse.de/tests/10658203/logfile?filename=autoinst-log.txt :
qemu-system-x86_64 -device VGA,edid=on,xres=1024,yres=768
That surely makes a difference. I suggest you take a look into where the QEMUVGA=virtio setting in the first job comes from and why it's not supplied in the second case.
#2
Updated by MDoucha 3 months ago
Updated by okurz wrote:
Please fill the description according to the ticket template https://progress.opensuse.org/projects/openqav3/wiki/#Defects
Observation¶
LTP SecureBoot tests fail to boot on some x86_64 workers.
Steps to reproduce¶
- Run LTP SecureBoot tests
Impact¶
LTP SecureBoot tests fail to boot on some x86_64 workers.
Problem¶
LTP SecureBoot tests fail to boot on some x86_64 workers.
Suggestion¶
- Review UEFI configuration on x86_64 workers
Workaround¶
Run SecureBoot tests on one of the few unaffected workers
I doubt this is related to the infrastructure. The worker slots on the same machine running just openQA worker instances in qemu are just different processes of the very same software. I compared autoinst-log.txt from the first fwo URLs you shared and I found:
- https://openqa.suse.de/tests/10651590/logfile?filename=autoinst-log.txt :
qemu-system-x86_64 -device virtio-vga,edid=on,xres=1024,yres=768- https://openqa.suse.de/tests/10658203/logfile?filename=autoinst-log.txt :
qemu-system-x86_64 -device VGA,edid=on,xres=1024,yres=768That surely makes a difference. I suggest you take a look into where the
QEMUVGA=virtiosetting in the first job comes from and why it's not supplied in the second case.
QEMUVGA=virtio comes from the Incidents-Kernel-Azure medium because SLE-15SP4 kernel-azure does not support any other video device. https://openqa.suse.de/tests/10658203 is the only job on the list that tests kernel-default instead of kernel-azure.
The other successful job (https://openqa.suse.de/tests/10659359) tests kernel-azure so you can compare the failures with that one. It's a direct clone of the 3 failed jobs.
Updated by #5
Updated by osukup 3 months ago
on worker9:
il | qemu-ovmf-x86_64 | balíček | 202008-150300.10.14.1
locked package with ovmf from Leap 15.3 ( 2 years old )
on openqaworker16/17:
i+ | qemu-ovmf-x86_64 | balíček | 202202-150400.5.5.1 | noarch | Update repository with updates from SUSE Linux Enterprise 15
latest ovmf from Leap 15.4 ...
#6
Updated by okurz 3 months ago
- Related to action #113366: Add three more Prague located OSD workers size:M added