QA » openQA Project » openQA Tests

Also happens in boot_encrypt
https://openqa.suse.de/tests/1224378#comments

hm, ok. so that's interesting because we now have "disable_grub_timeout" so we should not miss the grub screen. So this is a different issue then.

PR: https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/3821

PR merged, triggered one job manually:

openqa_clone_job_osd 1235824 _GROUP=0 TEST=okurz_crosscheck_cryptlvm_minimal_x

Created job #1239194: sle-15-Installer-DVD-ppc64le-Build321.5-cryptlvm_minimal_x@ppc64le -> https://openqa.suse.de/t1239194

Test moved forward, but lines https://github.com/os-autoinst/os-autoinst-distri-opensuse/blob/master/tests/installation/grub_test.pm#L76-L79 make it stuck again, checking for the grub screen again. Should be removed now with grub timeout disabled?

After discussion with the team about the use of this test and also with the developer of the offending lines it seems reasonable to exclude them instead of delete them.
PR: https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/3837

PR closed accidentally: New PR open: https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/3853.

I got to a solution that is suppose to unblock staging and also fix the ppc64. PR modified (not closed, sorry for the previous confusion). It was successful with 64bit. Asked mgriessmeier for verification run in ppc. The problem that the grub appears before the passphrase need to be further discussed with developers I think.

It still failing for ppc64: https://openqa.suse.de/tests/1242988# but at least unblock failures for 64bit: https://openqa.suse.de/tests/1242989#

We can see failures in grub_test and boot_crypt tests. Sometimes the first does actions that the second should do due to the following differences in behavior for each architectures:

cryptlvm tests for SLE15:

• ppc64le: Display boot option -> Enter passphrase to decrypt vda3 -> Enter passphrase to decrypt vda3 (only typed twice for cryptlvm test) -> System booting -> User login prompt. https://openqa.suse.de/tests/1243046#step/boot_encrypt/1
• aarch64: Enter passphrase to decrypt hd2,gpt2 -> Display boot option -> Booting system. https://openqa.suse.de/tests/1242987#step/boot_encrypt/2
• 64bit: Enter passphrase to decrypt hd0,gpt2 -> Display boot option -> Enter passphrase to decrypt vda2 -> System booting -> User login prompt. https://openqa.suse.de/tests/1242989#step/boot_encrypt/3

Comparing with SLE-12-SP3:

• 64bit: Display boot option -> System booting -> Enter passphrase to decrypt cr_virtio-1-part2 -> User login prompt. https://openqa.suse.de/tests/1058314#step/boot_encrypt/1

Not sure what is the right approach to continue, meaning the correct workflow for each architecture. Set ticket on feedback.

Created needle for typed password: https://gitlab.suse.de/openqa/os-autoinst-needles-sles/blob/master/boot_encrypt-encrypted_disk-typed_password-20171106.png
Verification run: https://openqa.suse.de/tests/1244702# (still scheduled)

Verification failed because grub_test is typing passphrase for ppc. Excluding this part only for ppc: https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/3872
New verification seems successful: https://openqa.suse.de/tests/1244836#step/boot_encrypt/3

Current status of this ticket:
ppc64le:

• https://openqa.suse.de/tests/1245349#step/first_boot/2: It started to ask for the passphrase twice: so it is failing because boot_encrypt cannot handle typing more than one passphrase. Last build to this one was successful with the changes introduces by the PR because for some reason it was asking for passphrase once. (see last comment line)
• https://openqa.suse.de/tests/1246219#step/boot_encrypt/3: Passphrase is asked once, so the result is successful.

This task need to be picked up by someone else due to is required local verification for ppc to consolidate logic in tests grub_test and boot_encrypt.
That could be handled in a different ticket due to the main fix of this ticket is done, that is to pass the grub screen (due to disabled grub timeout) or to be continued in this one (asking Oliver about it).

Last re-triggered of the build: https://progress.opensuse.org/issues/25854 asked for password once, so both jobs are successful.

Regarding comments in related bug https://bugzilla.suse.com/show_bug.cgi?id=1062806#c25 seems to be reasonable that the first passphrase does not appear in ppc64le due to a dedicated /boot partition is present in this architecture outside the encrypted volume. In aarch64 and x86_64 there is no dedicated /boot partition, therefore grub prompts passphrase as it wants to load kernel and initrd from the LVM encrypted partition.
Current jobs related with encryption are successful so it might not be needed further actions from our side on ppc in this ticket:
https://openqa.suse.de/tests/1268728
https://openqa.suse.de/tests/1268722