A practical guide to Fedora and Red Hat Enterprise Linux, 7th Edition (2014)

---

See the transmission-cli man page for a list of options. One of the most useful options is –u (––uplimit), which limits how much bandwidth in kilobytes per second the swarm can use while downloading the torrent from the local system (upstream bandwidth). Use –U (––no-uplimit) to place no limit on the bandwidth the swarm can use. The following command prevents BitTorrent from using more than 100 kilobytes per second of upstream bandwidth:

$ transmission-cli -u 100 Fedora-xxxx.torrent

BitTorrent usually allows higher download rates for members of the swarm that upload more data, so it is to your advantage to increase this value if you have spare bandwidth. You need to leave enough free upstream bandwidth for the acknowledgment packets from your download to get through or the download will be very slow.

The name of the file or directory BitTorrent saves a file or files in is specified by the torrent. You can specify a different file or directory name by using the ––save_as option. The transmission-show utility displays the name the BitTorrent file will be saved as, the size of the file, the name of the torrent (metainfo file), and other information:

$ transmission-show Fedora-19-x86_64-DVD.torrent
Name: Fedora-19-x86_64-DVD
File: Fedora-19-x86_64-DVD.torrent

GENERAL

Name: Fedora-19-x86_64-DVD
Hash: 687077ec95f03111a0f7326e42ba518bb7161c29
Created by:
Created on: Mon May 27 20:08:14 2013
Piece Count: 17113
Piece Size: 256.0 KiB
Total Size: 4.49 GB
Privacy: Public torrent

TRACKERS

Tier #1
http://torrent.fedoraproject.org:6969/announce

FILES

Fedora-19-x86_64-DVD/Fedora-19-x86_64-CHECKSUM (1.14 kB)
Fedora-19-x86_64-DVD/Fedora-19-x86_64-DVD.iso (4.49 GB)

RPM: The RPM Package Manager

RPM (RPM Package Manager; www.rpm.org) works only with software packages built for processing by RPM; it can download, install, uninstall, upgrade, query, and verify RPM packages. The files that hold these packages have a filename extension of .rpm. RPM uses the local RPM installation database to track the locations where software packages are installed, the versions of the installed packages, and the dependencies between the packages. RPM uses the Berkeley Database (Berkeley DB or BDB) and stores the database files in /var/lib/rpm. RPM forms the basis for yum: yum finds packages, downloads them from repositories, and then calls RPM to install/upgrade/remove the packages.

Source RPM packages are frequently found in a directory named SRPMS (source RPMs), whereas binary RPM packages usually reside in Packages. When you download binary packages, make sure they are relevant to the local operating system (both distribution and release—for example, Fedora 15).¹ They should also be compiled for the appropriate processor architecture:

1. Many RPM packages run on releases and even distributions other than the ones they were compiled for. However, installing packages intended for other distributions can create problems.

Processor architectures

• i386 covers all Intel- and most AMD-based systems; the N and Z Series Atom processors are also based on this architecture.

• i586 covers Pentium-class or higher processors.

• i686 refers to Pentium II or better and includes MMX extensions.

• S390x is for IBM System/390.

• ia64 is for the Intel 64-bit Itanium (not compatible with x86_64)

• alpha is for the DEC/Compaq Alpha chip.

• ppc and ppc64 denotes the Power PC 32- and 64-bit processors.

• x86_64 is for Intel and AMD 64-bit processors (not Itanium); the 230 and 330 Series Atom processors are also based on this architecture.

• sparc covers the Sun Sparc processor.

noarch

The name of the rpm file contains almost all the necessary information. Packages marked noarch (“no architecture”) contain resources, such as images or scripts, that are run by an interpreter. You can install and run noarch packages on any architecture.

yumdownloader

See page 540 for instructions on using yumdownloader to download RPM package files.

Querying Packages and Files

The following rpm command displays a list of one-line summaries of all packages installed on the system:

$ rpm -qa
samba-common-4.0.9-1.fc19.x86_64
mythes-1.2.3-4.fc19.x86_64
gnome-themes-standard-3.8.4-1.fc19.x86_64
xorg-x11-drv-openchrome-0.3.3-1.fc19.x86_64
caribou-0.4.11-1.fc19.x86_64
gnome-shell-3.8.4-2.fc19.x86_64
gstreamer-plugins-good-0.10.31-9.fc19.x86_64
libnl-1.1-17.fc19.x86_64
gnome-shell-extension-window-list-3.8.4-1.fc19.noarch
...

The –q option queries the package database; the –a option specifies all packages. Without –a, –q takes the name of a package as an argument and displays information about that package only. For instance, rpm –q ypserv tells you whether the ypserv package is installed and, if so, which version is installed. Use the –ql options to list the files in a package.

$ rpm -q ypserv
package ypserv is not installed

$ rpm -ql logrotate
/etc/cron.daily/logrotate
/etc/logrotate.conf
/etc/logrotate.d
/usr/sbin/logrotate
/usr/share/doc/logrotate-3.8.4
/usr/share/doc/logrotate-3.8.4/CHANGES
/usr/share/doc/logrotate-3.8.4/COPYING
/usr/share/man/man5/logrotate.conf.5.gz
/usr/share/man/man8/logrotate.8.gz
/var/lib/logrotate.status

With the –qi options, rpm displays information about a package:

$ rpm -qi logrotate
Name : logrotate
Version : 3.8.4
Release : 1.fc19
Architecture: x86_64
Install Date: Tue 04 Jun 2013 07:29:46 PM PDT
Group : System Environment/Base
Size : 97523
License : GPL+
Signature : RSA/SHA256, Tue 30 Apr 2013 08:32:12 AM PDT, Key ID 07477e65fb4b18e6
Source RPM : logrotate-3.8.4-1.fc19.src.rpm
Build Date : Tue 30 Apr 2013 02:16:03 AM PDT
Build Host : buildvm-07.phx2.fedoraproject.org
Relocations : (not relocatable)
Packager : Fedora Project
Vendor : Fedora Project
URL : https://fedorahosted.org/logrotate/
Summary : Rotates, compresses, removes and mails system log files
Description :
The logrotate utility is designed to simplify the administration of
log files on a system which generates a lot of log files. Logrotate
allows for the automatic rotation compression, removal and mailing of
log files. Logrotate can be set to handle a log file daily, weekly,
monthly or when the log file gets to a certain size. Normally,
logrotate runs as a daily cron job.

Install the logrotate package if you need a utility to deal with the
log files on your system.

You can use the –qf options to determine which package a file belongs to; it works with installed packages only. Use the yum whatprovides command (page 537) to search repositories for the same information. The following command shows that more is part of the util-linux package:

$ rpm -qf /bin/more
util-linux-2.23.1-2.fc19.x86_64

Include the –p option with other options to query an uninstalled RPM package file:

$ $ ls ypser**
ypserv-2.31-2.fc19.x86_64.rpm

$ rpm -qip ypserv-2.31-2.fc19.x86_64.rpm
Name : ypserv
Version : 2.31
Release : 2.fc19
...

Installing, Upgrading, and Removing Packages

Although it is frequently easier to use yum or gpk-application, you can use rpm to download, install, upgrade, or remove a package. Run rpm with root privileges; although you can run rpm as a nonprivileged user, you will not have permission to write to the necessary directories during an install or uninstall, and the procedure will fail. During a query, you do not need to work with root privileges, so you can and should work as a nonprivileged user. Give the –U (upgrade) option, followed by the name of the file that contains the RPM version of the package you want to install. The –U option upgrades existing packages and installs new packages (as though you had used the –i option). For kernels, use –i (not –U) to leave the old kernel intact when you install a new kernel. Add the –v (verbose) option to display more information about what is happening and the –h(or ––hash) option to display hash marks as the package is unpacked and installed.

Installing a package

For example, the following command installs the samba package on the local system:

# ls samba**
samba-4.0.6-3.fc19.x86_64.rpm

# rpm -Uvh samba-4.0.6-3.fc19.x86_64.rpm
Preparing... ################################# [100%]
Updating / installing...
1:samba-2:4.0.6-3.fc19 ################################# [100%]

When you install a package, the rpm file must be in the working directory, or you must use a pathname that points to the rpm file. If you specify an FTP or HTTP URL, rpm will download the package before installing it (but it will not download or install dependencies):

# rpm -Uvh http://dl.fedoraproject.org/pub/fedora/linux/updates/18/x86_64/bind-9.9.3-4.P2.fc18.x86_64.rpm

Removing a package

The following command removes the samba package. You can give the command from any directory:

# rpm -e samba
error: Failed dependencies:
samba is needed by (installed) system-config-samba-1.2.100-1.fc19.noarch

When you run this command, rpm reports that another package, system-config-samba, is dependent on the samba package. To remove the samba package, you have two choices: You can ignore the dependency by including the rpm ––nodeps option or you can remove the dependent package and then remove the samba package.

# rpm -e system-config-samba
# rpm -e samba

If you remove the samba package without removing the package that is dependent on it, the utilities within the dependent package will not work. In the preceding example, the system-config-samba utility will not work.

When you use rpm to remove a package, rpm queries the package database to find the information it needs to uninstall the package and removes links, unloads device drivers, and stops daemons as necessary. Refer to the rpm man page for more rpm options.

Installing a Linux Kernel Binary

The following steps install a Linux kernel binary from an RPM package. If you simply want to install the latest kernel for the local system, give the command yum install kernel instead. Refer to Chapter 14 when you want to configure and rebuild a kernel from source files, rather than installing a new, prebuilt kernel binary.

1. Run rpm with the –i option to install the new kernel. Do not use the –U option: You are installing a new kernel that has a different name than the old kernel; you are not upgrading the existing kernel.

2. Make sure the new kernel works before you remove the old kernel. To verify the new kernel works, reboot the system using the new kernel. You might want to wait a while before removing the old kernel to make sure that no problems arise with the new one.

3. When you are satisfied with the new kernel, remove the old kernel using rpm with the –e option; you must specify the kernel version number. If necessary, you can manually remove an old kernel by removing files whose names contain the release number from /boot or / (root). Remove information about the old kernel from grub.conf (page 590). Removing the kernel manually is not recommended because it is too easy to miss a file and because it does not remove the kernel from the RPM database.

Installing Non-rpm Software

Most software that is not in RPM format comes with detailed instructions on how to configure, build (if necessary), and install it. Some binary distributions (those containing prebuilt executables that run on Fedora/RHEL) require you to unpack the software from the root directory (i.e., you must cd to / before unpacking it).

The /opt and /usr/local Directories

Some newer application packages include scripts that install themselves automatically into a directory hierarchy under /opt, with files in a /opt subdirectory that is named after the package and executables in /opt/bin or /opt/package/bin.

Other software packages allow you to choose where you unpack them. Because many different people develop software for Linux, there is no consistent method for installing it. As you acquire software, install it on the local system in as consistent and predictable a manner as possible. The standard Linux file structure has a directory hierarchy under /usr/local for binaries (/usr/local/bin), manual pages (/usr/local/man), and so forth. Because many GNU build tools search the /usr/local hierarchy by default and could find the wrong version of a utility if you install developer tools there, putting these tools in /opt is a good idea.

To prevent confusion later and to avoid overwriting or losing the software when you install standard software upgrades, avoid installing nonstandard software in standard system directories (such as /usr/bin). On a multiuser system, make sure users know where to find the local software and advise them whenever you install, change, or remove local tools.

GNU Configure and Build System

The GNU Configure and Build System makes it easy to build a program that is distributed as source code (see www.gnu.org/software/autoconf). This process requires a shell, make (make package), and gcc (the GNU C compiler; gcc package). You do not need to work with root privileges except to install the program.

The following example assumes you have downloaded the GNU which program (ftp.gnu.org/pub/gnu/which; page 255) to the working directory. First unpack and decompress the file and cd to the new directory:

$ tar -xvzf which-2.20.tar.gz
which-2.20/
which-2.20/EXAMPLES
which-2.20/posixstat.h
...
which-2.20/configure.ac
which-2.20/COPYING

$ cd which-2.20

After reading the README and INSTALL files and installing the gcc package, run the configure script, which gathers information about the local system and generates the Makefile file:

$ ./configure
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for a thread-safe mkdir -p... /bin/mkdir -p
checking for gawk... gawk
checking whether make sets $(MAKE)... yes
checking whether to enable maintainer-specific portions of Makefiles... no
...
config.status: creating Makefile
config.status: creating maintMakefile
config.status: creating tilde/Makefile
config.status: creating config.h
config.status: executing depfiles commands

Refer to the configure info page, specifically the ––prefix option, which causes the install phase to place the software in a directory other than /usr/local. For package-specific options enter the command configure ––help. Next, run make:

$ make
make all-recursive
make[1]: Entering directory `/home/sam/which-2.20'
Making all in tilde
make[2]: Entering directory `/home/sam/which-2.20/tilde'
source='tilde.c' object='tilde.o' libtool=no \
DEPDIR=.deps depmode=pch /bin/bash ../depcomp \
gcc -DHAVE_CONFIG_H -I. -I.. -g -O2 -c tilde.c
source='shell.c' object='shell.o' libtool=no \
DEPDIR=.deps depmode=pch /bin/bash ../depcomp \
gcc -DHAVE_CONFIG_H -I. -I.. -g -O2 -c shell.c
rm -f libtilde.a
ar cru libtilde.a tilde.o shell.o
...
source='which.c' object='which.o' libtool=no \
DEPDIR=.deps depmode=pch /bin/bash ./depcomp \
gcc -DHAVE_CONFIG_H -I. -g -O2 -c which.c
gcc -g -O2 -o which getopt.o getopt1.o bash.o which.o ./tilde/libtilde.a
make[2]: Leaving directory `/home/sam/which-2.20'
make[1]: Leaving directory `/home/sam/which-2.20'

$ ls which
which

After make finishes, the which executable is in the working directory. Some packages place the executable in a subdirectory. If you want to install it, give the following command while running with root privileges. By default this command installs which in /usr/local/bin so it is known to be the local version of which.

$ su -c "make install"
Password:
make install-recursive
make[1]: Entering directory `/home/sam/which-2.20'
Making install in tilde
make[2]: Entering directory `/home/sam/which-2.20/tilde'
make[3]: Entering directory `/home/sam/which-2.20/tilde'
...

You can complete the entire task with the following command line:

$ su -c "./configure && make && make install"

The Boolean AND operator (&&) allows the execution of a subsequent command only if the previous step returned a successful exit status.

Keeping Software Up-to-Date

See page 533 for a discussion of why it is important to keep software up-to-date.

Bugs

A bug is an unwanted and unintended program property, especially one that causes the program to malfunction (definition courtesy of foldoc.org). Bugs have been around forever, in many types of systems, machinery, thinking, and so on. All sophisticated software contains bugs. Bugs in system software or application packages can cause the system to crash or programs not to run correctly. Security holes (a type of bug) can compromise the security of the system, allowing malicious users to read and write files, send mail to your contacts in your name, or destroy all data on the system, rendering the system useless.

Even if the engineers fixed all the bugs, there would still be feature requests as long as anyone used the software. Bugs, feature requests, and security holes are here to stay. Thus they must be properly tracked if developers are to fix the most dangerous/ important bugs first, users are to research and report bugs in a logical manner, and administrators are to apply the developers’ fixes quickly and easily.

Bug tracking

Early on, Netscape used an internal bug-tracking system named BugSplat. Later, after Netscape created Mozilla (mozilla.org) as an open-source browser project, the Mozilla team decided that it needed its own bug-tracking system. Netscape’s IS department wrote a very short-lived version of Bugzilla. Terry Weissman, who had been maintaining BugSplat, then wrote a new open-source version of Bugzilla in Tcl, rewriting it in Perl a couple of months later.

Bugzilla belongs to a class of programs formally known as defect-tracking systems, of which Bugzilla is preeminent. Almost all Linux developers use this tool to track problems and enhancement requests for their software. Fedora/RHEL uses Bugzilla to track bugs and bug fixes for its Linux distributions. To use Bugzilla, visit bugzilla.redhat.com.

Errata

As the Linux community, including Red Hat and the Fedora Project, finds and fixes operating system and software package bugs, including security holes, Red Hat and the Fedora Project generate RPM files (page 546) that contain the code that fixes the problems. When you upgrade a system software package, rpm renames modified configuration files with a .rpmsave extension. You must manually merge the changes you made to the original files into the new files. The easiest way to learn about, obtain, and install updates is to use yum.

The Fedora Announce List (admin.fedoraproject.org/mailman/listinfo/announce) holds information about Fedora updates. Under Fedora/RHEL, packagkitd via gpk-update-icon notifies you when updates are available.

curl: Downloads Files Noninteractively

The curl utility is a noninteractive, command-line utility that retrieves files from the Web using HTTP, HTTPS, or FTP. In the following example, curl downloads the Fedora Project home page, named index.html, to a file in the working directory with the same name.

$ curl http://fedoraproject.org/index.html > index.html
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 23076 100 23076 0 0 71837 0 --:--:-- --:--:-- --:--:-- 72112

The curl ––continue (–c) option with an argument of – automatically continues an interrupted download. For example, if you decide to stop a download so you can run it in the background, you can continue it from where it left off with this option.

Chapter Summary

As a system administrator, you need to keep applications and system software current. Of the many reasons to keep the software on a system up-to-date, one of the most important is system security.

A PMS (package management system) is a set of utilities that allow you to install, upgrade, remove, and search software packages easily and in a consistent manner. It is based on a local database that holds information about each installed package, including its dependencies and version information. The development of the RPM PMS has made the process of adding and removing the software packages quite easy. You can use yum, a front end to RPM, to install and upgrade software packages. The yum utility is installed by default and is easy to configure and use.

In addition, you can use the rpm utility to install, uninstall, upgrade, query, and verify RPM packages. For packages distributed as source code, the GNU Configure and Build System makes it easy to build executable files.

BitTorrent is a handy tool for downloading large static data files such as the Fedora installation ISO images. BitTorrent can replace protocols such as anonymous FTP, where client authentication is not required.

Exercises

1. Which command installs the samba package? Which installs the Web Server package group?

2. Which command would you give to update all installed packages using yum?

3. Why would you build a package from its source code when a (binary) rpm file is available?

4. Suggest two advantages that rpm files have over source distributions.

Advanced Exercises

5. When you compile a package yourself, rather than from an rpm file, which directory hierarchy should you put it in?

6. What are some steps you should take before performing an upgrade on a mission-critical server? When should you use rpm –i instead of rpm –U?