I just checked installation of qemu-kvm & libvirt & virt-install on a Fedora Server Edition, that is a headless server. It installed 338 packages / 158 mb including poppler, mesa, wayland, libX11, gtk3, cairo and a lot of additional graphic related software.
I’m wondering, why I need Wayland or X11 etc for kernel virtualization and running virtual machines? It turns my headless server into a graphical workstation. It installs software that I don't need for running a headless server, and don't want there.
If I install the same combo on a Debian / Ubuntu headless Server it installs about 40 packages / 20 mb. I know, the packages are not the same, but Debian/Ubunt doesn’t need X11, poppler, mesa, etc. for virtualization on a headless server. Can't we do the same?
I just had a quick look on / test of Fedora Server 36 (rawhide 22-02-18).
- Installation of DVD went smooth and ssh login via administrative user worked fine
- Before I even did anything I noticed a SELinux ACE (SELinux prevents 20-chrony-dhcp from accessing file enp1s0.sources with create access) which then appeared repeatedly several times.
Didn't cause any noticeable problems, but makes an ==untrustworthy impression== of the whole distribution when we not even have our highly praised security system under control.
- Tried to install a system-nspawn mockup container without success. Couldn’t boot the container, got:
Welcome to Fedora Linux 36 (Thirty Six Prerelease)!
Failed to load BPF object: Operation not permitted
BPF LSM hook not enabled in the kernel, LSM BPF not supported
Queued start job for default target graphical.target.
Using headless server, wondering a bit about graphical target.
I got a login prompt, but after typing in the credentials the prompt was just repeated, no working login.
Starting it as system service produced the same SELinux bug that I've reported for years (since about F31 or so). Another indication that we are not in control of our security system.
Unfortunately, I’m too stupid to understand how to enter something into our QA tracking system (apart from a bug report that probably won't fare any better than the ones already there).
Has anyone else tried the rawhide pre-release version?
Here's my notes from testing the Samba portion of cockpit file sharing.
Any questions let me know.
Physical: Fedora 35 Server
Virtual: Windows 10 build 20H2
Samba installed and running. (smb & nmb)
add to smb.conf
include = registry
#added for cockpitfile sharing
reload smb if it's already running.
After the dependencies are met its close to a normal Samba share setup.
Some steps are done at the terminal and the rest is done in Cockpit
Terminal in cockpit with root privileges.
Create the directory used for sharing.
Give it the permission for the user or group.
If SELinux is enabled add the samba_share_t context to the directory.
If firewalld is running add the samba service to it.
At this point you can use File Sharing to configure the share.
(directory, name, permissions, etc.)
You can even reset or create the password for the user with access to
Mount in Windows explorer and test.
After the above is met it works great in Windows 10.
It's taken more time than I had anticipated but I have finally reviewed
the "Install a F35 VM using Cockpit" page of our user documentation.
My first issue is with the sentence "If there is no entry Virtual
Machines the admin missed to install the cockpit-machines module when
installing the virtualization support. Select Applications further down
and then Machines for installation.".
My cockpit installation did not have the "Virtual Machines" menu option
nor did I have the option to install cockpit modules. I could initially
only remove installed modules. Coming back to this days later, I now see
all cockpit modules, with the ability to install those I don't have. Is
this a known issue or have I done this wrong?
The _images/virtualization-vm-install-fedoraserver-cockpit-020.png image
does not match the form I get on F35. I have a "Connection" option which
gives me the choice between "System" and "Session". This is not
reflected in the image.
The documentation later states "After a test of the installation media
the well known installation process begins. You may select Expand at the
right top to increase readability". In my case, I HAD to click on
'Expand' for the VNC console to register my keystrokes.
In the "Optional post-installation tasks" section, the first task is
titled "Get rid of nano as default editor". I feel "Making vim the
default editor" better reflects the contents of this section. I would
also recommend replacing the command by "dnf install --allowerasing
vim-default-editor" which simplifies the task.
Excuse my ignorance, but I'm just wondering what sense the regular rawhide "compose check reports" make. Who processes this information?
E.g. in today's report "Fedora-Rawhide-20220224.n.1 compose check report" on devel(a)lists.fedoraproject.org I find under New OpenQA failures
ID: 1146980 Test: aarch64 Server-dvd-iso base_reboot_unmount@uefi
and about 17 other failure entries.
Who cares about that? Or is not necessary to take care of it?