Just finished building from spare parts my new homelab test server, and installing Fedora
34 on it. Concerning the documentation I have the following feedback, but please keep in
mind I'm not a sysadmin, just a homelab tinkerer.
1. Net Install ISO of Fedora Server worked great, sane defaults that give you a working
ssh server and cockpit out of the gate after the install is wonderful.
2. I really struggled with the documentation when it came to disk partitioning.
- I have 5 hard disks in this server build:
1 - 160 GiB
4 - 1 TB drives of different manufactures
- Planned to have the 160 GiB drive hold the following partitions and mounts (/BiosBoot,
/boot, /swap, /) with the other partitions being formatted /Biosboot as biosboot, /boot as
xfs, /swap as swap, and / as Btrfs.
- I planned to create a RAID 10 Btrfs pool with the 4 -1 TB drives for /home
- Issue #1 I thought I had UEFI booting on this system, so I followed the directions to
create a BIOSboot partition of 200 MiB, but when I attempted this both in
"Custom" and "Advanced Custom" it kept changing the partition size to
2 MiB so either this is expected behavior because Anaconda knows my system better than I
do (good possibility) or there is a bug.
- Issue #2 "Custom" just failed on me when I attempted to create a Btrfs
partition for the / partition on the 160 GiB drive that was not connected to the Btrfs
partitions on the other four drives. Perhaps this is not how I should have done it, but my
thinking was that if my 160 GiB drive failed it wouldn't affect the "data"
pool, I would just have to put a new system drive into the server, reinstall Fedora Server
and reconnect /home to the "data" Btrfs pool. Although "Custom" failed
when attempting this, it did seem to work in "Advanced Custom."
- Issue #3 I found it very difficult to determine how to plan and partition my 5 drives
for the needed paritions and Btrfs / RAID10 pool. I think this is because the
documentation has a little snippet in the Fedora Server install documentation, but then
you really need to read the two other sections that are referenced as well,
(this takes you directly to the section, and then this section is probably a good idea to
read as well
(which is not specifically referenced in the Fedora Server documentation). I'm
wondering if there would be some value, especially now that Fedora is "all-in"
on Btrfs to give a "suggested/starter/beginner" disk partitioning scheme that
covers a Btrfs RAID pool in the Fedora Serve
r documentation, or would it be possible to distill this information from the current 3
different sources in to a single resource with perhaps a flow chart for illustration
3. I'm confused as to why this is needed:
- I read the note, "According to the default installation, SSH login is only possible
using an SSH key file. However, the setup cannot be done as part of the installation. If
this step is omitted, logging in as root via SSH is not possible."
- However, I was able to SSH login in using the user that I created during the install and
gave administrative privileges to. So then I just used "ssh-copy-id" to copy my
Public Key file to the server, and then I could log in without a password.
- Did I do a non default install by adding my own administrative user? If that is the case
wouldn't it be better to make that the default so a new Fedora Server user would have
a way to ssh into their new install right away?
4. I would love to write up a documentation page about configuring your server to use
"wakeonlan" which I find is super helpful for a homelab person. I would
basically rewrite this article that I found:
- The blog writer is referring to Ubuntu server, but it worked perfectly for my system,
and this motherboard doesn't like the other methods that you can find on the internet
mostly because I think they trigger before the network interface is ready to have its
- One question that I have is, most of my rewrite would borrow heavily from this man's
work, and I feel that his steps are the best I have found on the internet searches that I
have done on this topic if your server runs systemd, which Fedora does. Do we have a way
to give this author some credit if my rewrite borrows heavily from his blog article?
I'm so thankful for the Fedora Server Working group whose work has gotten me this far,
if I can help with any of the documentation points above by submitting rewrites as pull
requests let me know. Also, because I'm so new to the server world, there may be good
reasons why my thoughts above don't make sense as well or wouldn't be helpful to
new Fedora Server users.