-rw-r--r-- 1 tom tom 1712130048 Apr 1 15:04 Fedora-Xfce-Live-x86_64-39-1.5.isoThen I do this to an 8G flash stick I have laying around:
su dd if=Fedora-Xfce-Live-x86_64-39-1.5.iso of=/dev/sde bs=64MBe very careful to get the correct device name for the flash stick! Also I find it useful to type commands into an update log (like this very one I am now composing), then but and paste them onto the command line to avoid typos.
This is a live ISO image that can either be used to run F39 or to install it. I will use it to do the latter.
It has an ASUS motherboard and the DEL key gets me to the BIOS. I select UEFI boot (which it tells me is the lexar jump drive), this takes me to a Grub screen, which is good. I tell it to test media and then boot F39. I am now running F39 from the live USB ISO.
I get a Fedora desktop with the XFCE mouse. The lowest icon on the left says "install the live CD to your hard drive". I double click it and find myself in the Fedora installer.
I want to configure the "install destination". The vital thing is on the first screen to select "Custom" rather than "Automatic" configuration. After doing this you can thrash through the non-intuitive partitioning GUI.
I would like to end up with a partition scheme about like this:
sda1 - biosboot sda2 - /boot 4G (was 600M) sda3 - / 200G (was 50G) sda4 - swap 32G (was 16G) sda5 - /u1 - the remainder (was only 11 percent full)It is important (for me anyway) to tell it to use standard partitions.
I use the "+" key to add partitions, starting with biosboot. I let that default to 2M.
When I get to /u1, I simply don't give a size and it sets it up to use the rest of the disk (about 1.5G).
This went pretty smooth, but I have done this before and residual memory helps a lot.
I enable the root account, set a password, and click the box that says to allow ssh access for root with a password.
I add a user "tom" and use the advanced section to set the uid/gid to 104:104 The installer wants to force me to use 1000 or greater, so I will have to force this issue later. I use 1004:1004 for now.
Network and hostname do not allow me to set a static IP, so this is another thing I will have to force later.
It has the Phoenix timezone set, so I am ready to begin installation. Away it goes at 4:06 PM. In less than 10 minutes (4:13 PM) it tells me that the install is done and that I should reboot. I click "Done" and it takes me back to the live CD.
I pull the flash stick and press the reset button.
On my home machine I set up an entry in /etc/hosts for "clam". It refuses ssh connections on port 22.
I type "dnf update" and after I say yes to a GPG key, it gets busy upgrading 615 packages.
When it is done there is a 6.7.10 kernel installed (and I am running 6.5.6), so ...
su sync reboot
Disable selinux Set static IP Be sure firewall allows ssh and http verify ssh access fix tom uid/gid Enable auto updates Set up web server Add /u1 (and /u2) to fstab mirroring of /u1 to /u2 reboot and check everything
su nmcli con down "Wired connection 1" nmcli con add type ethernet con-name Wired ifname eno1 ip4 192.168.0.6/24 gw4 192.168.0.1 nmcli con del "Wired connection 1"I expected to have to do a "nmcli con up" on Wired, but that seems to happen automatically when I add it (and when there is no other connection on en01).
Note that /etc/hostname contains "cholla" and I am leaving that be for now.
nmcli con modify Wired ipv4.dns 199.104.150.50 nmcli con down Wired nmcli con up Wired
firewall-cmd --list-services dhcpv6-client mdns sshIt turns out that the problem was not ssh being blocked by the firewall, but sshd was not running. Solve this via:
systemctl start sshd.service systemctl enable sshd.serviceNow that I have ssh running, I can connect from my main system via ssh and not have to roll my chair back and forth across the room. Also I don't have to fight with the screensaver. Also I can cut and paste commands from my files with notes.
I use "sestatus" to verify that it is disabled and smile.
Edit /etc/passwd and /etc/group to make it so, then:
cd /home chown -R tom:tom tomI log out and back in to ensure that I can login as user tom, and it all seems fine.
dnf -y install httpd firewall-cmd --add-service=http --permanent firewall-cmd --reload systemctl start httpd.service systemctl enable httpd.serviceI aim my browser at: http://gila/ and see the Fedora Webserver test page.
I want to use /u1/www for my web page, not /var/www/html. This involves editing /etc/httpd/conf/httpd.conf . I make the changes here to change the documentroot to /u1/www and restart the server:
systemctl restart httpd.service
dnf install dnf-automatic vi /etc/dnf/automatic.conf systemctl enable dnf-automatic.timer systemctl start dnf-automatic.timerThe only line I edit is:
apply_updates = yes
Change the IP number copy all the old content into /u1 remove the old (bad) disk set up the /u2 mount point and mount it set up the mirroring /u1 to /u2 ensure that the web pages work ensure that remote ssh works
#!/bin/sh /home/tom/Backup/cholla_daily exit 0I placed this file there, ensured it was executable and owned by root, and will keep an eye on things to be sure it is working.
The top drive in the case is the root and /u1 drive, the bottom holds /u2
The following are old well known schemes using tar and pipes to copy files:
# before moving disk cables around tar cvf /u1/root.tar --one-file-system -X /home/tom/excludes # with the new disk in place as /u1, and the old disk as /u2 # the old /u2 just "off to the side" for the time being. cd /u1 ( cd /u2; tar cvf - . ) | tar xpvf -And this info was important to jot down:
cholla IP - 199.104.150.52/24 gateway IP - 199.104.150.1 DNS IP - 199.104.150.50At this point, I rarely if ever use cholla as a desktop system in my office. It is just a server that could be headless that provides ssh and http. If I start using it in the office, I will need to install other packages and do more setup.
Adventures in Computing / tom@mmto.org