9.3 Installation fails at Grub2-common

Greetz and thanks for even looking. I’ve tried installation several times and I’m a 20+ year veteran on Linux. I test lots of distros to learn and keep my hand in play, but my Main has been Slackware for 21 years.

Installation each time gets as far as “grub2-common” and fails with -

“ERROR in POSTRANS script in rpm package grub2-common”

I have an ext4 partition for root, and ext2 partition for /boot, and a vfat for /boot/efi with each one specced to be reformatted by Rocky. The vfat partition is flagged “esp,boot”.

What am I missing?

UPDATE - I did finally get a successful install by trying all 3 partitions (/, /boot, /boot/efi) on the same hard drive but I want it on an NVME drive. I deleted partitions on the NVME drive to try will all 3 on the same SSD but it fails on grub2-common just like al;l other attempts other than all on the spinner.

I’d just copy the HD install over to NVME and edit “/etc/fstab” if it wasn’t for initramfs. I’d like to boot it with rEFInd but I’ll live with Grub IF I can get a decent install on NVME.

Please, any suggestions gratefully accepted and acknowledged.

UPDATE 2 - Verbatim Error (I’m convinced it’s a legit bug)

"The following error occurred while installing the payload. This is a fatal error and installation will be aborted.

DNF Error: Error in POSTTRANS scriptlet in rpm package grub2-common"

Despite numerous warnings like you must reformat foo partition, you should select a swap partition, all of which I followed most times until I tried not following out of frustration (I’ve spent days on this mess) there is zero warning to alert me to any errors on my part in setup, especially that routinely halt at the same grub2 install/config AND then the “Exit Installation” radio button does not initiate anything. It just stays there, no drive bus flashing lights, no change in appearance, no keystroke affect, just zero activity until I hard Reset. Just FTR I do verify the install media (USB “burned” from iso via Balena Etcher).

There are some impressive features with the installer, like displaying file system and even Label during partition selection, but error this is ridiculously bad having no guidance or warning, not to mention a week with no response here. Please, some kind soul make a suggestion even if it is just to provide more info about hardware or my process. Thank you iun advance.

How is the nvme connected to the motherboard? Is it the primary drive as listed in the firmware boot order table? I’ve found that each firmware manufacturer has quirks that make it tricky to boot when multiple disks are involved and this may effect installation as well. When it comes to hardware it is difficult to have such at hand to duplicate the issue so a lot of guessing gets involved.

Thank you for taking the time to respond, jbkt23, I seriously appreciate it! My Main PC is a multiboot monster and has been ever since I was TeamOS2 way back in the day. It currently has 5 internal drives with ~27 partitions on it providing almost a dozen bootable systems.

It now has several defunct partitions that were Rocky Linux install attempts. The only one that ever got past the grub2-common error, completed, booted and ran had all of it’s partitions on a spinner, the 4th drive. However the root partition was both too small and I’m spoiled by the bandwidth of NVME so that minor success motivated me to give Rocky some upgraded importance.

Currently, my latest (attempted but failing on grub2) locations are all on Disk 0 with the single exception of “swap” which is still on the spinner. I’ve tried verifying disk as well as network install and every attempt excepting that on the old spinner fails in precisely the same manner. I vastly prefer rEFInd or elilo as a bootloader but I have had easy success with Mangeia, Arch, CentOS, and OpenSuse to name a few that ultimately get booted for me via rEFInd but began life with Grub2 (and still will work that way)…well… the ones I’ve kept more than a few months.

FWIW I normally set firmware for the highest priority boor order as rEFInd but most, including even Rocky failure’d installs, reset their own boot loaders as Priority 0 and after trying them for Pass/Fail, I reset to rEFInd keeping the Pass’es and efibootmgr delete the Fails.

I totally agree with you that variations in hardware creates so many variables it is exactly that, guesswork, especially when the installer warns about non-critical issues but hasn’t a check or a clue about bootloader Pass/Fail. In my case as an EE and someone who began building PCs with DOS 3, my hardware is pretty much top notch.

My Main (I have several active homebuilt PCs including 3 towers, 2 laptops, and an ARM-based network backup machine) is built on an Asus Hero z490 platform with 16GB top notch RAM, powered by a gold standard Corsair 800 watt PSU, cooled to under 40C average, and all the NVME drives are EVO Pros, with the one I’m trying to install Rocky on in the primary M.2 slot. Just FTR, I wiped the partitions of working other distro systems on that drive to accommodate Rocky. Those installed, booted, and ran just fine, which is why I felt desperate enough to wipe a working system in order to stop the fail. That has not worked for Rocky…yet.

Thanks again for your help but in my view this is some odd bug. My /boot partition is 3GB ext2, /boot/efi is 500MB fat32 (which Rocky install demands be reformatted to “EFI” despite that I know all about “boot,esp” flags) and the installer should be able to determine something as basic as bootloader viability and deny continuation until that is addressed, or better just do what virtually every other distro that has to comply with UEFI standards does, and succeed on a proper setup.

I don’t know why the install is forcing you to reformat the esp but I currently create separate esp partitions for all my dual boot installs or tri boot for that matter. Disk realestate is cheap especially for that tiny partition and as far as I can tell you can have as many as suits. You’ll figure it out eventually.

I don’t care that Rocky requires reformatting ESP, but do care that it continues install attempts when it should “know” it will fail with simple grub2-common script requirements since it does it repeatedly.

FWIW, via rEFInd, I can boot the Rocky install attempt to an emergency prompt complaining “switch root failed” and the log tells me “sysroot” and one other I can’t recall atm are to blame.

Admittedly I’m not very familiar with systemd system repairs at such a level and I do prefer the old skool SysVInit human readable scripts that just do exactly what I tell them to do, no more or no less, even if it’s wrong, At least I can learn from that. The biggest problem currently for me is I don’t know what is lacking or if I can provide it/them to at least get switch root satisfied.

Astalavista, Baby. I’m done with Rocky. The juice isn’t worth the squeeze for me. It was just vastly easier to get DaVinci Resolve installed and working on Open Suse… plus I dislike Gnome anyway. There were things about Redhat style I did enjoy and respect but not enough to keep tearing my eyes out over a stupid install freeze.