Learn software management with advanced Linux administration in this tutorial by Frederik Vos, a Linux trainer and evangelist and a senior technical trainer of virtualization technologies, such as Citrix XenServer and VMware vSphere.
— post by Frederik Vos, provided by Packt —
In the old days, installing software was a matter of extracting an archive to a filesystem. There were several problems with this approach:
- It was difficult to remove the software if the files were copied into directories that were also used by another software
- It was difficult to upgrade software, maybe because the files were still in use or were renamed
- It was difficult to handle shared libraries
That’s why Linux distributions invented software managers.
The RPM software manager
In 1997, Red Hat released the first version of their package manager, RPM. Other distributions such as SUSE adopted this package manager. RPM is the name of the rpm utility, as well as the name of the format and the filename extension.
The RPM package contains the following:
- A CPIO archive
- Metadata with information about the software, such as a description and dependencies
- Scriptlets for pre and post-installation scripts
In the past, Linux administrators used the rpm utility to install/update and remove software on a Linux system. If there was a dependency, the rpm command was able to tell exactly which other packages you needed to install. However, the rpm utility couldn’t fix the dependencies or possible conflicts between packages.
Nowadays, the rpm utility isn’t used any longer to install or remove software; instead, you use more advanced software installers. After the installation of software with yum (Red Hat/CentOS) or zypper (SUSE), all the metadata goes into a database. Querying this rpm database with the rpm command can be very handy.
A list of the most common rpm query parameters are as follows:
|List all the installed packages.
|List the installed configuration files.
|List the installed documentation and examples.
|List all the installed files.
|Shows the package that installed this file
|Verifies the integrity/changes after the installation of a package; use -va to do it for all installed software.
|Use this parameter together with other parameters if the package is not already installed. It’s especially useful if you combine this parameter with –script to investigate the pre and post-installation scripts in the package.
The following screenshot is an example of getting information about the installed SSH server package:
The output of the -V parameter indicates that the modification time has changed since the installation. Now, make another change in the sshd_config file:
sudo cp /etc/ssh/sshd_config /tmp
sudo sed -i 's/#Port 22/Port 22/' /etc/ssh/sshd_config
If you verify the installed package again, there is an S added to the output, indicating that the file size is different, and a T, indicating that the modification time has changed:
Other possible characters in the output are as follows:
|Major/minor on devices
For text files, the diff command can help show the differences between the backup in the /tmp directory and the configuration in /etc/ssh:
sudo diff /etc/ssh/sshd_config /tmp/sshd_config
You can also restore the original file as follows:
sudo cp /tmp/sshd_config /etc/ssh/sshd_config
The DPKG software manager
The Debian distribution doesn’t use the RPM format; instead, it uses the DEB format invented in 1995. The format is in use on all Debian and Ubuntu-based distributions.
A DEB package contains:
- A file, debian-binary, with the version of the package
- An archive file, control.tar, with metadata (package name, version, dependencies, and maintainer)
- An archive file, data.tar, containing the actual software
Management of DEB packages can be done with the dpkg utility. Like rpm, the utility is not in use any longer to install software. Instead, the more advanced apt command is used. All the metadata goes into a database, which can be queried with dpkg or dpkg-query.
The important parameters of dpkg-query are as follows:
|Lists all the packages without parameters, but you can use wildcards, for example, dpkg -l *ssh*
|Lists files in an installed package
|Shows information about the package
|Shows the state of the package
The first column from the output of dpkg -l also shows a status as follows:
The first character in the first column is the desired action, the second is the actual state of the package, and a possible third character indicates an error flag (R). ii means that the package is installed.
The possible desired states are as follows:
- (u) unknown
- (h) hold
- (r) remove
- (p) urge
The important package states are as follows:
- n(ot) installed
- H(a)lf installed
- Hal(F) configured
Software management with YUM
Your Update Manager or Yellowdog Updater Modified (YUM) is a modern software management tool that was introduced by Red Hat in Enterprise Linux version 5, replacing the up2date utility. It is currently in use in all Red Hat-based distributions but will be replaced with dnf, which is used by Fedora. The good news is that dnf is syntax-compatible with yum.
Yum is responsible for:
- Installing software, including dependencies
- Updating software
- Removing software
- Listing and searching for software
The important basic parameters are as follows:
|Search for software based on package name/summary
|Search for software based on a filename in a package
|Information and status
|Update all software
You can also install patterns of software, for instance, the pattern or group File and Print Server is a convenient way to install the NFS and Samba file servers together with the Cups print server:
|yum groups list
|List the available groups.
|yum groups install
|Install a group.
|yum groups info
|Information about a group, including the group names that are in use by the Anaconda installer. This information is important for unattended installations.
|yum groups update
|Update software within a group.
|yum groups remove
|Remove the installed group.
Another nice feature of yum is working with history:
|yum history list
|List the tasks executed by yum
|yum history info <number>
|List the content of a specific task
|yum history undo <number>
|Undo the task; a redo is also available
The yum command uses repositories to be able to do all the software management. To list the currently configured repositories, use:
To add another repository, you’ll need the yum-config-manager tool, which creates and modifies the configuration files in /etc/yum.repos.d. For instance, if you want to add a repository to install Microsoft SQL Server, use the following:
yum-config-manager --add-repo \
The yum functionality can be extended with plugins, for instance, to select the fastest mirror, enabling the filesystem / LVM snapshots and running yum as a scheduled task (cron).
Software management with Zypp
SUSE, like Red Hat, uses RPM for package management. But instead of using yum, they use another toolset with Zypp (also known as libZypp) as backend. Software management can be done with the graphical configuration software YaST or the command-line interface tool Zypper. The important basic parameters are as follows:
|Search for software
|Perform a distribution upgrade
There is a search option to search for a command, what-provides, but it’s very limited. If you don’t know the package name, there is a utility called cnf instead. Before you can use cnf, you’ll need to install scout; this way, the package properties can be searched:
sudo zypper install scout
After this, you can use cnf:
If you want to update your system to a new distribution version, you have to modify the repositories first. For instance, if you want to update from SUSE LEAP 42.3 to version 15.0, execute the following procedure:
- First, install the available updates for your current version:
- Update to the latest version in the 42.3.x releases:
- Modify the repository configuration:
sudo sed -i 's/42.3/15.0/g' /etc/zypp/repos.d/repo*.repo
- Initialize the new repositories:
- Install the new distribution:
- Now, reboot after the distribution upgrade.
Besides installing packages, you can also install the following:
- patterns: Groups of packages, for instance, to install a complete web server including PHP and MySQL (also known as a lamp)
- patches: Incremental updates for a package
- products: Installation of an additional product
To list the available patterns, use:
To install them, use:
sudo zypper install --type pattern <pattern>
The same procedure applies to patches and products. Zypper uses online repositories to view the currently configured repositories:
You can add repositories with the addrepo parameter, for instance, to add a community repository for the latest PowerShell version on LEAP 15.0:
sudo zypper addrepo \
If you add a repository, you must always refresh the repositories:
Software management with apt
In Debian/Ubuntu-based distributions, software management is done via the apt utility, which is a recent replacement for the utilities, apt-get and apt-cache.
The most-used commands include:
|Search in descriptions
|Install a package
|Show package details
|Remove a package
|Update catalog of available packages
|Upgrade the installed software
|Edit the repository configuration
Repositories are configured in /etc/apt/sources.list and files in the /etc/apt/sources.list.d/ directory. Alternatively, there is a command, apt-add-repository, available:
'deb http://myserver/path/to/repo stable'
The apt repositories have the concept of release classes:
- Old stable, tested in the previous version of a distribution
They also have the concept of components:
- Main: Tested and provided with support and updates
- Contrib: Tested and provided with support and updates, but there are dependencies that are not in main, but for instance, in non-free
- Non-free: Software that isn’t compliant with the Debian Social Contract Guidelines (https://www.debian.org/social_contract#guidelines)
Ubuntu adds some extra components:
- Universe: Community provided, no support, updates possible
- Restricted: Proprietary device drivers
- Multiverse: Software restricted by copyright or legal issues
If you found this article interesting, you can explore Frederik Vos’ Hands-On Linux Administration on Azure to administer Linux on Azure. Hands-On Linux Administration on Azure will help you efficiently run Linux-based workloads in Azure and make the most of the important tools required for deployment.