Linux Package Management (Distribute Your App)

From Foss2Serve
Revision as of 22:53, 13 February 2016 by Nyeates (Talk | contribs)
Jump to: navigation, search
Title Linux package management (Packing the 'htop' command)
Overview Students will learn about rpm and dnf/yum package tooling and then actually create their own package from code, upload it to a public package repository and finally have a classmate install their compiled package. This can be advertised to students as being able to distribute their own linux-based application to the world.
Prerequisite Knowledge Students should be familiar with:
  • Beginner to Intermediate Shell knowledge (linux command line)
    • Getting around the linux file system, sudo'ing, editing files, etc.
  • Making and compiling Linux software code
    • see extended note in Comments section below
Learning Objectives Upon completion, students should be able to:
  • Install RPM's and use the dnf/yum command to inspect and install packages
  • Make and Compile existing code into a binary RPM package
  • Upload a package to an online repository
  • Install a package from an online repository

Background:

Is there background reading material?


Are there other related activities?


What is the rationale for this activity?

  • Students may wonder how they can distribute linux-based code that they develop out to other everyday users. Not many users enjoy compiling code, so the RPM package structure allows easy installation of binary pre-compiled code packages. Basically, a student could have coded a small sample command-line application and now they can send it to others easily. The other side of it is that students, as users of Linux, will be interfacing with RPM packages not of their making and it will be good to have an understanding of the backends of how this works. Students will interface with the dnf and yum commands even if simply using linux for fun. Now, they can have knowledge into how it works and how they can employ it for their careers and personal uses.


Directions:

Keep a working log

Log your shell commands, answers to questions, and commentary in a text file, wiki, or blog. You will be constructing and troubleshooting numerous linux shell command's and their outputs. Your assignment is to document these commands and the process you went through in an organized fashion. You might use bullet points or a new set of commands on each line. Make sure it is easily consumable by a human (your instructor), as well as yourself 10 years from now.

The data that is the output of the commands is not as interesting as a summary or comment of their output or on what the command has done - ex: failed, succeeded, why, what it did, what it changed, etc. Write these in complete sentences. Commentary is especially important if you run into problems. When this occurs, state the problem and how you intend to solve it. At the end, you should have a text document with all of the commands, right and wrong, that you went through to get this activity completed. It should read as a timeline of what you did and what your thoughts were, to get the assignment complete. After this is complete, you will summarize the most useful commands into a sort of cheatsheet - this can come in useful for years to come.

The history command is your friend. Use it to see what commands you have typed in the past. It is advised that you utilize multiple terminal (shell) windows at the same time. When you get sidetracked on installing a dependency, use a separate window for that. This keeps your shell history organized and parsable afterwards. You might also consider upping your terminal windows buffer for the number of lines it holds, to somewhere in the tens of thousands - just in case you decide you want to come back to something you learned a while back, or in case a command spits out thousands of lines of results - which can happen. Some terminal applications can record commands and output to text files. Either way, make your written log human readable, make it a story if you wish. When you come back to this years from now, needing to remind yourself how to package some newly minted code, you want a summary/cheatsheet and you want to be able to understand the context and follow your thoughts of why you tried certain things. You don't want to be parsing through lines and lines of shell input and output.


The activity will follow these general steps

  1. Use dnf/yum and rpm files.
  2. Learn about creating rpm's and the rpmbuild command.
  3. Package the htop project into an RPM.


Step 1: dnf / yum and .rpm's

  • First, lets do some learning before we jump into shell commands. Answer the following in your log:
    • What is the difference between dnf and yum? Why was the change made?
    • What are .rpm files? Where might you find them and how are they used?
  • Use the dnf command to view what packages have been recently installed on your machine.
    • Hint: There is a specific command for this. man dnf is your friend
    • Attempting to list all packages will often be quite a long listing.
  • Use the dnf command to install a .rpm package not currently installed on your machine.
    • Pick something simple, as you will need to operate it in the next steps.
    • Hint: If you dont find your own (more credit given for this), you can install screen
  • Use dnf to show that the new package is now installed.
  • Run or operate this newly installed utility. Where did dnf place the executable file?


Step 2: RPM's and rpmbuild

Required reading / watching:

  1. RPM Build Example Video - Watch, at least, the first 8 mins of this video from a recording of a local Linux User Group meeting
  2. RPM process Diagram from video - Find the diagram that the video references on the page labeled 12-17
    • Document authored by Guru Labs, L.C. released under the CC-BY-NC-ND 2.0 license
  3. Packaging the `wget` command - Read this example in its entirety
    • This will take some time to read through.
    • If you were to try to follow along in the shell, there may be some hickups with wget.
    • Instead, save it for Part 3; You will use this readings structure to RPM'ize the htop project.


Questions:

  • What is the rpmbuild command?
  • Quickly summarize the directory structure when making your own RPM - you will stick this into your cheat sheet later, so make it short.


Step 3: Create your own 'htop' RPM

In this section, we will take code from the htop project and turn it into an RPM package.

Htop is a tool that shows you how busy your CPUs and memory are, as well as detailed information about the various processes being run on your operating system. It is semi visual, in that it runs in the terminal, but uses colors and blocks to make "graphics" (called ncurses). It is a successor to the famed top command that comes in all linux flavors. System administrators use both of these to diagnose and troubleshoot problems.

Htop was selected in this activity because it does not come standard in most operating systems, meaning it is not already installed. It is also a relatively small/simple build and the functionality of htop is rather useful.


Instructions:

  • Packaging a command - With all of the below instructions, you will be following this example, except replace it with htop


  • Build htop manually
    • Compile the htop project to test that the code can be turned into binary form in your environment.
    • Before you can create an RPM, code must compile manually.
    • Dependency issues:
      • There is always some odd ./configure setting or dependency that needs installing (this means, other rpm packages that this project needs installed in order to compile)
      • Unfortunately, these dependencies are not installed automatically - you must install them yourself and then try to ./configure or make again
      • Some dependency issues you might run into with htop:
        • gcc, try each of these commands one at a time:
          • sudo dnf install glibc-devel glibc-headers kernel-headers kernel-devel gnutls-devel
          • sudo dnf groupinstall "Development Tools"
          • sudo dnf install gnutls
        • libncurses or unicode, "configure: error: You may want to use --disable-unicode or install libncursesw."
          • ./configure --disable-unicode # first try disabling unicode like they suggest
          • which ncurses # ncurses is usually already installed, but if not, you will need to dnf install it also
          • sudo dnf install ncurses-devel # try installing the ncurses-developer library package
    • Once you have ./configure, make, and make install compiling htop, you can move on to building an RPM for it


  • Create RPM dir structure
  • Run rpmbuild command
  • FIXME

Deliverables:

Students will hand in:

  • The working log
    • See the above lengthy note about the working log.
  • Cheat sheet / Summary
    • A quick-reference guide, truly meant for your future use.
    • List key commands and what they do
    • Succinctly explain the RPM directory structure


Assessment:

Criteria Level 1 (fail) Level 2 (pass) Level 3 (good) Level 4 (exceptional)
Installs and inspects RPMs with dnf/yum
Compiles code into binary RPM
Uploads RPM to COPR or other online repo
Installs colleagues RPM from COPR

Comments:

What should the instructor know before using this activity?

Students should know how to compile Linux software code before doing this activity.

  • There exists an activity which covers this: Introduction_to_building_open_source_software
  • Particularly, students should have experience using the configure, make, and make install commands
  • If students have not compiled in the past, it is common to not have all of the required libraries and modules already installed for the compilation process to succeed. This is a major part of creating RPM's. This setup takes time to troubleshoot and setup, and it could be different on each system if students are not using identical operating systems.


If you wanted to do this activity in Ubuntu, Debian, etc:

What are some likely difficulties that an instructor may encounter using this activity?


Additional Information:

ACM Knowledge Area/Knowledge Unit PL - Programming Languages, SE - Software Engineering from ACM_Body_of_Knowledge
ACM Topic PL/Compiler Semantic Analysis, PL/Code Generation, SE/Tools and Environments from https://www.acm.org/education/CS2013-final-report.pdf
Level of Difficulty Easy-Medium, for students that meet the pre-req's
Estimated Time to Completion 2-3 hours
Materials/Environment
  • Access to the shell of a Linux operating system that uses RPM (virtualized or on hardware can work)
  • root access is likely needed - another good vote for virtualized systems (might use openshift or virtualbox)
Author Nick Yeates
Source
License Creative Commons CC-BY

Suggestions for Open Source Community:

Suggestions for an open source community member who is working in conjunction with the instructor:

  • Why did your community decide to standardize on the autotools-based config/make building? Or, why not?
  • Explain to students why packaging skills can be used on many open source linux-based projects, across their careers.
  • Help students through the often-confusing build process
  • Help students to find alternate examples to step through - aka, finding project code other than the ones listed above and making it into an RPM.



This work is licensed under a Creative Commons Attribution 4.0 International License


Appendix

Consider breaking these additional steps into a new activity

  1. Have students host the package on a web source (ftp? Some free internet service that makes the .rpm accessible by URL)
  2. Have each student in the class pair up and try to install the other person's RPM and run the new command


Ideas on what to create into an RPM:


Other readings:

Personal tools
Namespaces
Variants
Actions
Events
Learning Resources
HFOSS Projects
Evaluation
Navigation
Toolbox