Bugzilla – Bug 1224150
[Build 20240511] boot_encrypt: grub does not ask for encryption key
Last modified: 2024-05-16 14:14:39 UTC
## Observation Grub no longer asks for the decryption key. The snapshot contained a grub2 change: +- Fix gcc error with CFLAGS=-Og + * grub2-grubenv-in-btrfs-header.patch + Which sounds harmless there were also plymouth changes, but the failure seems too early for plymouth openQA test in scenario opensuse-Tumbleweed-NET-x86_64-lvm-encrypt-separate-boot@64bit fails in [boot_encrypt](https://openqa.opensuse.org/tests/4179275/modules/boot_encrypt/steps/2) ## Test suite description Same as lvm-full-encrypt, but with separate boot not encrypted partition. Maintainer: riafarov INSTALLONLY=1 UNENCRYPTED_BOOT=1 DESKTOP=gnome ENCRYPT=1 FULL_LVM_ENCRYPT=1 ## Reproducible Fails since (at least) Build [20240509](https://openqa.opensuse.org/tests/4172240) ## Expected result Last good: [20240508](https://openqa.opensuse.org/tests/4169788) (or more recent) ## Further details Always latest result in this scenario: [latest](https://openqa.opensuse.org/tests/latest?arch=x86_64&distri=opensuse&flavor=NET&machine=64bit&test=lvm-encrypt-separate-boot&version=Tumbleweed)
Tested with grub2 reverted: failure persists Which makes plymouth the more likely candidate. Reverted that one now to test
With reverted Plymouth, we pass over the decryption of the disk - i.e Plymouth as issue confirmed https://openqa.opensuse.org/tests/latest?arch=x86_64&distri=opensuse&flavor=NET&machine=64bit&test=lvm-encrypt-separate-boot&version=Tumbleweed (Tests fails, but unrelated to boot decrypt)
The qcow2 is generated at https://openqa.opensuse.org/tests/4182076/asset/hdd/tumbleweed-encrypt-debug.qcow2 BTW, this failure no longer happens in my recent reruns.
(In reply to Jia Zhaocong from comment #4) > The qcow2 is generated at > https://openqa.opensuse.org/tests/4182076/asset/hdd/tumbleweed-encrypt-debug. > qcow2 > > BTW, this failure no longer happens in my recent reruns. ==> With reverted Plymouth, we pass over the decryption of the disk - i.e Plymouth as issue confirmed https://build.opensuse.org/package/rdiff/openSUSE:Factory/plymouth?linkrev=base&rev=129 The old state has been resubmitted to the devel project as https://build.opensuse.org/request/show/1173570
Zhaocong also found that: The password is being asked on serial console, https://openqa.opensuse.org/tests/4179275/logfile?filename=serial0.txt I confirmed that is correct. Thank you Zhaocong. When I launch the image offline, the password screen display normally. even I'm watching the serial output, the result is the same.
(In reply to Dominique Leuenberger from comment #5) > (In reply to Jia Zhaocong from comment #4) > > The qcow2 is generated at > > https://openqa.opensuse.org/tests/4182076/asset/hdd/tumbleweed-encrypt-debug. > > qcow2 > > > > BTW, this failure no longer happens in my recent reruns. > > ==> With reverted Plymouth, we pass over the decryption of the disk - i.e > Plymouth as issue confirmed Hi Dominique: Do you mean it's always happen, or you just run it once and find fails when decrypt the disk? Thank you!
(In reply to Jia Zhaocong from comment #4) > The qcow2 is generated at > https://openqa.opensuse.org/tests/4182076/asset/hdd/tumbleweed-encrypt-debug. > qcow2 > > BTW, this failure no longer happens in my recent reruns. Hi Zhaocong: Could you show me the exact command to launch the VM the same with the openQA platform? Especially the serial console device. Since I didn't see this problem happen on my qemu VM. as comment6 said. It is a device initialization time race related bug, I can not decide if it is introduced by this update. But even it happens 1 time, it should be a problem. and it's necessary to reproduce the it offline to collect more infos. Thank you so much!
https://openqa.opensuse.org/tests/latest?arch=x86_64&distri=opensuse&flavor=NET&machine=64bit&test=lvm-encrypt-separate-boot&version=Tumbleweed#next_previous Go back one page until you see snapshot 0511 The test ran 5 times into this error - when it finally passed (with validate_encrypt failing) is when snapshot 0511 was repushed to QA with Plymouth reverted so: definitively NOT a one-time failure
As plymouth was revered in Factory: you CANNOT rely on Factory for your tests.