Hello all! Are there any known issues where Anaconda won’t read LVM volumes? My primary laptop is a UEFI booted CentOS 8 install with /home and root on LVM volumes. Since you can’t automagically transmogrify a UEFI booted CentOS into a Rocky install, I decided I would just wipe out CentOS and install Rocky on top. I’ve done this dozens of times re-installing CentOS. I would select ‘custom partitioning’ on the disk selection tool. Within that tool I would be able to select the individual LVM volumes that already existed and assign their mount points. For the root Logical Volume I would have it reformat; for the /home LV I would leave it alone, in order to preserve all of the data under /home.
With Rocky I am unable to open the physical volume that is on my disk drive, and therefore I am unable to assign the LV’s to their mountpoints. Rocky knows this is an LVM volume because it presents the disk partition as an LVM volume with the correct name. Is this type of installation not supported?
Yes, I could just copy the data onto an external drive and copy it back, if I wanted to. No, I don’t want to…
I used the migration script to migrate my home system which is UEFI boot from another RHEL Clone to Rocky Linux 8.4, the script also deals with UEFI boots. If you search the script you will see the parts that are done to handle a smooth efi migration to Rocky Linux. Here’s what me efibootmgr output looks like.
Timeout: 1 seconds
Boot0000* Rocky Linux
Boot0001* UEFI:CD/DVD Drive
Boot0002* UEFI OS
Boot0003* Red Hat Enterprise Linux
Boot0004* UEFI:Removable Device
Boot0005* UEFI OS
Boot0006* Hard Drive
Boot0008* UEFI:Network Device
Do you have secure boot enabled? That may be the difference here. In my case I have secure boot enabled, and there is a check for this in the script. If it’s enabled the script errors out with “EFI Secure Boot is enabled but Rocky Linux doesn’t provide a signed shim yet. Disable EFI Secure Boot and reboot.”
I disabled SecureBoot and the script ran, but it then completely clobbered all of my installed packages and then choked on an NPM package (which I never had installed on the first place). End result was that most of the installed packages were still CentOS but the “base” packages had been changed to Rocky. I just restored from backup and I’m going to retry here in a little bit, disabling all of my extra repositories first.
Yes that must be the difference, I don’t have secure boot enabled, I didn’t read that far into the script but I see it now. I don’t think if you had secure boot on and changed it to off that it would have any effect as in how packages are synced or and installed etc. I suggest you report an issue on the rocky-tools github project page.