My goal is to install the Server version of Fedora 28.1.1, on a second Intel S1200SPLR server (after the first was proved to have a hardware problem) with a Xeon E3-1275 v5 CPU, 16GB of 2133 Unbuffered ECC DDR4, the onboard Intel P530 integrated HD Graphics, and a 480GB SSD, all components approved by the server board manufacturer. There are no cards in the system.
I am using the directly connected monitor on the only VGA port (back panel) and have specified that in the BIOS/Settings; IPMI is not available to me. I am not familiar with dracut.
First, I tested with Ubuntu 18.04.1 LTS, which installed AOK from its checksummed LiveUSB. There was no lag.
Next, I tried Fedora Server 28.1.1, from a LiveUSB generated by the Fedora LiveUSB Writer. The terminal window went blank shortly after booting from its checksummed LiveUSB. I got the same result with Fedora Workstation 29.1.2. I repeated the attempts with an ISO written onto different flash drives with Rufus.
Pressing e
in GRUB2 to replace quiet
with nomodeset
did not help. Neither did adding inst.text
, vga=790
, or vga=793
(ref. http://blog.fpmurphy.com/2009/09/project-plymouth.html). Every time, Server 28 stalled at fb: switching to mgag200drmfb from EFI VGA
.
Then, to try to get some information about what's happening, I tried Fedora Spin Workstation 29.1.2 with Cinnamon, having remembered something about GNOME problems. It installs, but Cinnamon repeatedly crashes every minute, both in LiveUSB mode and after installation and reboot. Cinnamon reports it is rendering in software mode. The 28.1.1 Spin with Xfce (which uses lightdm) installs and runs OK, which does suggest something in the GNOME package is trouble.
I opened a terminal, and copied dmesg from the Fedora 29.1.2 Workstation Cinnamon install for your review. The results of lspci-k can be seen here, and it appears there's a Matrox display adapter as well as the Intel Integrated HD Graphics on the motherboard..
I have also, as per https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/deployment_guide/blacklisting_a_module, created /etc/modprobe.d/blacklist.conf which contains one line
blacklist mgag200
and rebooted, but lspci -k still shows the Matrox in use.
I also edited /etc/default/grub to remove quiet rhgb (the call to Red Hat Graphical Boot) and replaced that with inst.text then ran grub2-mkconfig -o /boot/grub2/grub.cfg
and rebooted. When I did, I opened up grub with e
and found its next to last line read
GRUB_CMDLINE_LINUX="resume=/dev/mapper/fedora-swap rd.lvm.lv-fedora/root rd.lvm.lv=fedora/swap rhgb quiet"
I changed that to
GRUB_CMDLINE_LINUX="resume=/dev/mapper/fedora-swap rd.lvm.lv-fedora/root rd.lvm.lv=fedora/swap inst.text"
but the problem persisted.
-
Again, my objective is to install Fedora 28.1.1 Server, and the Workstation installs are just an attempt to gather information for troubleshooting.
rhgb
and addinginst.text
to the kernel command line. This will prevent any switch of video adapter and it should stay (1) in text mode, and (2) on the video adapter you booted with. Though, if you actually do intend to use it as a server I would suggest switching it back to the Matrox so you can use the remote console. – Michael Hampton Jan 17 '19 at 20:17