My choice of book was Fedora Linux by Chris Tyler, published in late 2006 by O’Reilly. I choice it because it covers SELinux and Virtualisation, and more generally seems to explain each topic in a comprehensible way. It’s over 600 pages long, so pretty comprehensive, too.

I’ve never been a fan of when computers insist on having the operating system CD inserted when they ought to be able to get whatever they need from the internet. Fedora would sometimes ask for the CD when using the Add/Remove Software (package manager) feature. Fortunately, it’s easy to stop that happening: simply delete the repository called InstallMedia (which can be found in in the Edit menu –> Repositories window in the package manager.)

While cron is used to schedule regular tasks, the at and batch commands are used to schedule one-off tasks at some point in the future. The difference between at and batch is that at does a task at a given time in the future, while batch does the task when the system is not doing very much, specifically when the “load average” drops below 0.8.

To set up at instructions, type at followed by a time at the command line, as in:

at 13:00

or

at now + 1 hour

To set up batch instructions, simply type:

batch

With both of these, enter your commands, line after line until finished, and then press Ctrl-D while on a blank line. Type the following to get a list of commands set to be completed in the future:

atq

The files /etc/at.allow and /etc/at.deny are used to restrict access to both at and batch.

Cron is the method of getting Linux to repeatedly do a task at regular intervals. For example, it can automatically produce a backup in the middle of the night or save some status information to a log file. It can send an email reminding people to file their TPS reports every Friday at 2pm.

The key file related to cron is /etc/crontab, and this file along with the /etc/cron.d/ and /var/spool/cron directories, are checked each minute; the crond daemon reads them and decides if there is anything to do.

Ordinary (non-root) users have their cron files stored as /var/spool/cron/crontabs/username. They can add new cron tasks is by using the crontab command. To edit their crontab, they should type in:

crontab -e

Root can edit another user’s crontab by typing in:

crontab -u username -e

Processes can be scheduled with a line containing the instruction in the following format:

Minute   Hour   Day   Month   DayofTheWeek   Command

An asterisk instead of a number for example in the month column means that it will follow the instruction regardless of which month it is. If you put 1-4 in the day column it will follow the instruction on the first to the fourth of the month, while 1,2,4, means that it will follow the instruction of the first, second and third of the month.

To restrict the use of cron to a particular set of users, or to specifically exclude some users, alter the files /etc/cron.allow and /etc/cron.deny.

Linux installs a range of directories on the hard disk. Here are some of the most important:

/ is the root of the filesystem.

/boot is where the Linux Kernel and Bash are stored.

/home is where the each user’s files are stored.

/root is where the root user’s files are stored.

/etc contains configuration files.

/etc/skel is where you put files if you want them to appear automatically in new users’ home directories.

/dev is where the devices on the system, like CD-Rom drives and modems are represented.

/bin is where many system commands are stored. Imagine an inventor’s garage – it might look like a bin but it contains lots of useful tools.

/usr/bin is like /bin; the difference is arbitrary.

/sbin and /usr/sbin contain system commands only available to root (the superuser).

/lib contains program libraries.

/mnt contains mount points for removable media.

/opt contains optional programs.

/proc contains currently-running Kernel-related processes.

/tmp is where the system stores temporary files.

/usr contains little programs available to all users.

/var is for variable files, such as logs and temporary spool files for printers.

Every file and directory on a Linux filesystem has permissions attached to it. They can be referred two in two formats: by a six digit set of letters or by a set of three numbers. Here’s an example in letters:

rwxr-xr-x

The first three letters mean that the file can be read, written to and executed by the file’s owner; the middle three mean the file can be read and executed by the group of users the owner belongs to; and the final three letters means that everyone logged into the server can read and execute the file.

In a directory listing (using the command ls -l), you would see an extra character before the first ‘r’, normally either a dash (-) or the letter d. A dash means it’s an ordinary file and a d means it’s a directory.

Permissions can also be referred to, as I mentioned, by three numbers, as in: 755.

The 7 means that the file can be read by the owner (4), written by the owner (plus 2) and executed by the owner (plus 1) etc. The number-based approach is commonly used when changing a file’s permissions using the chmod command, e.g.

chmod 755 file.txt

There are a number of useful commands for dealing with processes. One of them is called top. This command lists, in an continually-updated way, the processes running on the system and has an inbuilt facility to a kill a particular process. However, I prefer using the command ps, in the following way:

ps -auxww | less

Then the kill command can be used, sometimes with option 9 if a process is a particularly reluctant quitter, as in:

kill -9 2099

This kills process number 2099. The -9 bit means that the operating system doesn’t bother to give the process any ability at all to close on its own accord.

The best archiving tool to use on Linux is probably Gzip. By best, it is a format that everyone else using Linux can understand, and it is better at compressing files that the older compress command. A newer format called bzip2 has better compression rates but apparently not everyone can understand the format. Admittedly, that’s according to my 1999 edition of Running Linux but it’s certainly true that I’ve seen lots of Gzip (.gz) files, but can’t remember seeing any bzip2 (.bz2) files.

The gzip command compresses a file, the gunzip command uncompresses a file and the following command tells you how compressed a file is:

gzip -l filename.txt.gz

One quirk compared with the Zip archive commonly found on Windows and the Mac is that a .gz archive merely contains one file. As such, you will commonly find gzip archives called things like filename.txt.gz (it contains a single text file). To compress a whole range of files, you have to use an extra command called tar (stands for tape archive). Tar does no compression on its own: it simply lumps a selection of files into a single file for easy transportation. Once files are part of a Tar file, you use gzip to compress the tar and end up with a file looking something like backup.tar.gz.

Or you can cheat and use the z option, as in:

tar czvf home.tar.gz /home

This puts the home directory into a tar archive, compressed by gzip.

One of the machines on which I’ve installed Fedora boots directly into the GUI, the other into the command line. On the latter, I have to type startx before I get to the GUI, which is inconvenient. The reason it has defaulted to this arrangement is that I had to use the text-based installer to get Linux up and running because it didn’t have the correct driver for the graphics card (which has subsequently been installed separately).

So how should I fix this, making Fedora boot straight into the GUI? The answer is in the /etc/inittab file, where I simply change the default runlevel to 5 (from 3).

FTPing has always been a graphical, drag-and-drop thing for me. When I was running a Red Hat web server a few years ago, I used to use the wget command sometimes, especially when the hosting company sent an email around asking for the software to be upgraded. But for FTP itself, I’ve been firmly a GUI man. Until now.

There are a number of commands that do FTP, including ftp, sftp and lftp. The RHCT book I’m following recommends using lftp as this is the one Red Hat exam prep guide refers to, and because it’s particularly powerful. I wanted to know if lftp (like sftp) sends passwords securely, so I looked at the man pages for lftp and was informed that lftp defaults to a secure transmission of the password where available. Excellent.

To log into an FTP site anonymously, you enter:

lftp ftp.redhat.com

or to connect with a username and password, type:

lftp username@ftp.redhat.com

or

lftp -u username ftp.redhat.com

From there you have a Bash-style command line with the ability to enter commands like ls. The important extra commands are: put and mput, along with get and mget (the ‘m’ lets you work with multiple files. To finish the FTP session, type quit.