action #18444
closed[sles][functional]Make partitioning_raid test compatible with UEFI
100%
Description
Was created mainly due failing RAID tests on aarch64, see https://bugzilla.suse.com/show_bug.cgi?id=1032058
The problem is that ESP partition /boot/efi is created on RAID1 volume which is not supported by UEFI - EFI is compatible only with MBR/GPT tables with FAT partition and cannot read mdadm magic superblocks. So the machine is not able to boot after installation.
Updated by okurz over 7 years ago
- Subject changed from Make partitioning_raid test compatible with UEFI to [sles][functional]Make partitioning_raid test compatible with UEFI
Updated by thehejik over 7 years ago
- Checklist item changed from to [x] BIOS+RAID, [x] UEFI+RAID+ESP, [ ] UEFI+LVM(RAID)+ESP
the changes are visible at https://github.com/os-autoinst/os-autoinst-distri-opensuse/compare/master...thehejik:raid_uefi
Updated by thehejik over 7 years ago
It's a bit in conflict with https://progress.opensuse.org/issues/11434
Updated by thehejik over 7 years ago
- Checklist item changed from to [x] UEFI+LVM(RAID)+ESP
Updated by thehejik over 7 years ago
- Checklist item changed from [x] BIOS+RAID, [x] UEFI+RAID+ESP, [x] UEFI+LVM(RAID)+ESP to [x] BIOS+RAID, [x] UEFI+RAID+ESP, [x] UEFI+LVM(RAID)+ESP, [x] BIOS+LVM(RAID)
- % Done changed from 0 to 90
waiting for accepting PR https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/2740
and needle one PR https://gitlab.suse.de/openqa/os-autoinst-needles-sles/merge_requests/352
Updated by thehejik over 7 years ago
I found one issue on aarch64 (but only slow xgene2 and mustang2 workers, on seattle6 it works fine) in setraidlevel() from partitioning_raid test.
When selecting wanted RAID level some keystrokes (eg. Tab) are ignored and then the test is failing, see https://openqa.suse.de/tests/881471#step/partitioning_raid/130
I would like to add a wait between keystrokes so then it will look:
sub setraidlevel {
my ($level) = @_;
my %entry = (0 => 0, 1 => 1, 5 => 5, 6 => 6, 10 => 'g');
send_key "alt-$entry{$level}", 2; <---- WAIT 2s HERE
send_key "alt-i", 2; # move to RAID name input field <---- WAIT 2s HERE
send_key "tab"; # skip RAID name input field
}
Imo there is no better solution for it because wait_screen_change() will not match flashing cursor which is the only change in the screen.
Updated by thehejik over 7 years ago
thehejik wrote:
I found one issue on aarch64 (but only slow xgene2 and mustang2 workers, on seattle6 it works fine) in setraidlevel() from partitioning_raid test.
When selecting wanted RAID level some keystrokes (eg. Tab) are ignored and then the test is failing, see https://openqa.suse.de/tests/881471#step/partitioning_raid/130
I would like to add a wait between keystrokes so then it will look:
sub setraidlevel { my ($level) = @_; my %entry = (0 => 0, 1 => 1, 5 => 5, 6 => 6, 10 => 'g'); send_key "alt-$entry{$level}", 2; <---- WAIT 2s HERE send_key "alt-i", 2; # move to RAID name input field <---- WAIT 2s HERE send_key "tab"; # skip RAID name input field }
Imo there is no better solution for it because wait_screen_change() will not match flashing cursor which is the only change in the screen.
PR for the above https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/2752
Updated by thehejik over 7 years ago
I changed setraidlevel function to use wait_screen_change instead of do_wait within send_key. Original PR fixed and still valid.
Updated by thehejik over 7 years ago
thehejik wrote:
I changed setraidlevel function to use wait_screen_change instead of do_wait within send_key. Original PR fixed and still valid.
Hmm, the change didn't help so much :( partitioning_raid still failing on aarch64 workers xgene2 and mustang3. Test is passing only on aarch64 worker seattle6 (but it was passing even before without wait_screen_change). Any idea?
Updated by thehejik over 7 years ago
thehejik wrote:
thehejik wrote:
I changed setraidlevel function to use wait_screen_change instead of do_wait within send_key. Original PR fixed and still valid.
Hmm, the change didn't help so much :( partitioning_raid still failing on aarch64 workers xgene2 and mustang3. Test is passing only on aarch64 worker seattle6 (but it was passing even before without wait_screen_change). Any idea?
New attempt for the fix https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/2777
Updated by thehejik over 7 years ago
The last attempt didn't help as well :(
As a workaround I will introduce new WORKER_CLASS=qemu_aarch64_working_raid for seattle6 which is the only worker where the test is passing.
Time to time it fails even on seattle6 worker.
Updated by thehejik over 7 years ago
- Status changed from In Progress to Resolved
Closing, the test is now compatible with EFI, but I found a new issue poo#18828 where the story continues.