This is my guide to installing RedHat Linux onto a
blank Intel-based PC. There are, of course, variations between the
different versions of RedHat, but this guide should pretty much cover
6.2 through 7.3. (Note: pre-7.1 wasn't strong for PHP/Postgres
out-of-the-box. Actually, 7.0 might be good, but I only used it a tiny
bit, before I was into PHP/Postgres. I've been using 7.1 extensively, and
7.2 & 7.3 a bit.) There is no single perfect way to do an
installation,
and I'm sure my way is not even close. All I know is, "This is what I do
and it works." I recently had to squeeze RH7.2 onto a system with a
1GB HD and left some things off and lots of stuff wasn't working: no
PICO, no PINE, and even though I added myself (brian) during
installation just like I always do, /home/brian did not exist and I didn't
get my usual shell
when I logged in. Logging in as root looked like it always did, but
logging in as me didn't. (Some of that might have been because I made
/home from a RAID, but I'm not sure. Further investigation is needed.)
This guide is really
only here a) as a way to keep notes for myself and b) to give friends a
baseline system that matches mine so I can help them with it down the
road. If you want the official guide, it's right
here.
Why do I do what I do? When I put together a Linux system, I might want
some desktop stuff, so I put on X, KDE, and Gnome. If I think I might want
to play videos or songs, I add multimedia support. DOS/Windows
Connectivity = Samba (I think) so that goes on, and Printing Support goes
on because (I think) that's needed for the Virtual PDF/Samba printer.
HTTPD (Apache webserver) and SQL Database (Postgres) go on because I'm
playing with that stuff a lot these days.
- Obtain Linux. For me, the best method is to go to an installfest and have someone there burn
a copy for you. (Quick note for those new to Linux: Linux is distributed
under a license that specifically allows unlimited redistribution.) Second
best (if you have a fast Internet connection and a CD burner) is to
download it. If all
else fails, go buy it.
- Prepare for installation, part 1. Will you be installing it all
alone
or with another OS? Is it going
to be an easy-going box with everything on one partition, or will it
actually see some serious use (i.e., web or file server) and therefore
require things like multiple partitions? Now is the time to figure that
out.
- Prepare for installation, part 2. Once you know what you're
going to do, start getting ready. Assemble the system and gather up a
monitor, keyboard, and (optionally) mouse. Will the system boot to a CD?
If not, use rawrite (Windows) or dd
(Linux) to make a boot floppy. OK, are you all ready? Then pop in the
disc, power up the system, and keep reading.
- Installation, part 1. This part is pretty easy. I'll have to
rewrite
this someday as I'm doing an installation to make sure I get all the steps
correct, but I've done it a few times, it's pretty easy, and my memory's
decent.
- First question: High-res (default), low-res, or text. I've had
a couple
wierd experiences with the graphical installers (strange colors a couple
times, I've got one monitor that doesn't work with the default GUI
settings, and who knows what'll happen with a laptop) so I just run in
text mode. It's not that scary.
- The basics: English, US
keyboard, mouse, etc.
- Type: Your choices are Workstation, Server, or Custom (maybe
others).
I always do 'custom'.
- Disk preparation: every box is different, and pages and pages
have
been written on partitioning schemes. For a basic home system, one big
partition (/) is fine, plus swap. If you want to multiboot, that's a whole
other ball of wax, and if your system is going to get some serious use, I
recommend making separate partitions for /boot, /var, /tmp, and /home. I
think
the option "just take over my whole disk" is usually there; take it, if
you want. It's as good as anything else.
- Packages: If I think I'll ever sit down at the box locally, I
put on
KDE and Gnome. (Those also need XWindows in the first place, of course.)
If I'm going to be doing Samba serving, I think it's included if you
choose DOS/Windows connectivity. I don't print, but many people do, so
Printer Support is usually good. Also, if you're going to do a virtual PDF
printer with Samba, I think Printer Support covers that, too. HTTPD server
(Apache) is good, Anonymous FTP server is something I don't use but it's a
good way to get files onto your machine if you aren't familiar with other
methods, and SQL Server (Postgres) is something else I really like. I
don't use Emacs and I've gotten by OK (usually) without the Development
stuff, but I usually throw Utilities on thereI'm not sure what it
does, but it sounds useful and it's about 7 MB. If in doubt, throw it all
on there--disk space is cheap. What I've listed above should take up less
than 1 GB; an 'everything' install will be a good amount more.
- Network: If applicable; season to taste.
- Installation, part 2. After that, it's pretty much yes, yes,
add a user, yes, yes, yes, go.
Next time I do an installation I'll do a detailed step-by-step writeup as
I run the installer. For the most part, it is as quick and painless as a
Windows 95 installation and usually faster than a Windows 98, NT, 2000,
or XP installation.
- Installation, Part 3. Do you want a boot floppy? Only you can
answer that question. Do you want to boot into text (init 3) or GUI (init
5)? Again, it's up to you. Once at the machine, it's a simple case of
issuing init 3 as root to go from GUI mode to text, or
init 5 to go from text mode to GUI. Changing the default
is a simple matter of changing one line in one file, but the best way to
go is to choose what you want the final state of the machine to be.
Workstation, go with GUI. Server, go with text, and start the GUI manually
as needed.
- First boot. Once the machine has booted up, I run
setup as root and, under the 'system services' option,
enable HTTP (almost always), SSH (always), SMB (if desired), and Postgres
(if desired.) (What you're choosing here are the services you want to
start when the machine boots.) Then, quit setup and reboot. Yes, they can
all be turned on by hand, and you'll earn the ridicule of hardcore
"reboots are for hardware upgrades and kernel rebuilds" types, so have
this excuse ready if you encounter them: "There is no other way to be
absolutely sure that the services I want will, indeed, be present after a
reboot without testing." But, the reality is, it's the quickest and
easiest way to turn them all on if you don't know all the commands
yet, and yes, it is a good idea to make sure you have everything
set up correctly and everything you want happens at boot time. Besides, if
you do all this when the system is 3 minutes old, you won't chip into your
uptime much. :-) Note that if you plan to work in more than one runlevel,
you'll need to run setup in each.
Congratulations, you are now ready to go. For a basic Linux desktop,
you've got KDE and Gnome, each of which come with office apps. For server
use, you've got a scriptable
webserver (Apache with PHP and Perl), a file server (NFS for Linux/Unix,
Samba for Windows, and Netatalk for Macs is just one RPM download away),
and a relational database (PostgreSQL) that can be easily accessed with
PHP and Apache.
[Brian Ashe's home page]
[brianashe.com/linux]