Hi Akemi:
In the July 19, entry 17 on this post, I mentioned that I had a test on CentOS 8.5.2111
which worked. With several subsequent test failures, I tried to repeat this one on
the HP Pavilion, and it no longer worked.
On the understanding that the CentOS version number should match the dud/kmod
version number, i.e. CentOS-8.6 and dd-sata_nv-3.5-5.el8_6.elrepo.iso have a “six”
in the appropriate places, I downloaded CentOS 8.4.2105 boot .iso from the Japanese
archive.
On the HP Pavilion, using CentOS 8.4.2105, rather than use the dud file, I
initialized a USB stick with the label, OEMDRV, and loaded the file,
kmod-sata_nv-3.5-4.el8_4.elrepo.x86_64.rpm on it. On booting up the CentOS
dvd, I added the following statement to the grub command line:
inst.dd=hd:LABEL=OEMDRV:/kmod-sata_nv-3.5-4.el8_4.elrepo.x86_64.rpm
CentOS booted up o.k. and I selected the English language, and moved to master
selection screen. The disks were recognized on the destination screen, but there was
no ethernet available, and as a consequence, the installation source was not available
because the BaseOS repo could not be reached.
I went onto the Compaq Presario with the CentOS 8.4.2105 DVD and added a file
to the USB stick, kmod-forcedeth-0.0-6.el8_4.elrepo.x86_64.rpm, and to the grub
command line, added a second statement,
inst.dd=hd:LABEL=OEMDRV:/kmod-forcedeth-0.0-6.el8_4.elrepo.x86_64.rpm
Again, CentOS 8.4.2105 booted up o.k. and after language selection, the master
selection screen came up. The disks were recognized and ethernet was now available
and could be enabled. The installation source screen was puzzling. An .iso could be
selected from OEMDRV if available, and an .iso could be downloaded from the
internet, with ftp, or http, or a few other options, but there was no apparent recognition
that the CentOS dvd was in fact the main installation source ???
I repeated this test format on both the HP Pavilion and the Compaq Presario using the
Rocky 8.6 dvd and the two files,
kmod-sata_nv-3.5-5.el8_6.elrepo.x86_64.rpm and
kmod-forcedeth-0.0-7.el8_6.elrepo.x86_64.rpm
Both tests failed after a series of messages, ending with
“started dracut pre-pivot and cleanup hook”.
After several minutes, there was a lengthy series of dracut-initqueue timeout messages ended
by an exit to the dracut emergency shell on the Compaq Presario.
On the HP Pavilion, nothing further happened.
There were some significant error messages:
-after the message “extracting sata_nv”, modprobe: ERROR:
could not insert ‘sata_nv’ invalid argument
-cat: ‘/sys/class/net/*/device/modalias’ no such file or directory
this message occurred after extracting kmod-forcedeth as well as sata_nv
I used the same USB stick and USB ports for both the CentOS and Rocky tests,
so I don’t understand why the sata_nv and kmod-forcedeth files were so
problematic for Rocky, and without incident on CentOS.
Len E.