Make your own free website on Tripod.com
BACK

Setting up a Concurrent Versions System

This page will cover setting up a Concurrent Versions System (CVS) on a Linux server.  CVS comes standard with practically every modern Linux/Unix distribution. The Linux distribution being used in this example is SuSE 7.2, but any distribution with CVS will work. 

To begin with, a explanation of what CVS would probably be helpful to those unfamiliar with it.  CVS is a document management system that allows simultaneous editing of copies of the same documents and merging all the changes into one master copy.  This system is very useful to a group of software developers that are working on the same project.  With CVS, these programmers can all submit changes to the software and have them merged into a master copy in the repository.  Furthermore changes can be tracked and releases can be made based on a specified version.

In learning how to setup and use CVS, the book Open Source Development with CVS by Fogel and Bar (2001) was very useful.   I mention this book because I am using it as a guide for writing this very basic HowTo and it or a source similar to it should be consulted for a more in depth discussion of CVS.

    The CVS commands available on CVS equipped system can be used to gain client access to remote repositories.  Hosting a CVS repository, on the other hand, requires a few steps to set it and the server up.

  1. The first step is to set up the repository; this step can be accomplished with the following command:
      # cvs -d /path/to/repository init
    

    The /path/to/repository is going to be the source directory for the repository and will be a directory that you have write access to.  If your user doesn't have write access, root user access will be needed.  It will be needed not just in this step, but in future steps as well (for restarting processes, modifying passwd file, and so on).

  2. Next will be to add a new cvs group and make it the owner group and change the permissions of the directory specified in the /path/to/repository statement above.  The commands to perform these tasks are as follows:

      # groupadd cvs
      # cd /path/to/repository
      # chgrp -R cvs .
      # chmod ug+rwx . CVSROOT
    
  3. The next step will be to set up the cvs server daemon.  The daemon is monitored by the inetd daemon.  To ensure that the cvs daemon is enabled, three things need to occur: first,the /etc/services file needs to be edited to verify a line for cvspserver exists; second, the inetd.conf file needs to be edited to ensure that the cvspserver is enabled; and third, inetd needs to be restarted.  These steps also ensure that a password authentication server is running to allow authenticated log ins to the CVS server.

    1. Edit /etc/services and make sure the following line exists, you may have to scroll down a lot as the port numbers are normally in sequential order.

        cvspserver   2401/tcp
      
    2. Now open the /etc/inetd.conf file and verify that the line for cvspserver is uncommented or added if necessary.  The line should look like the following (but will be on one line) if your system uses tcpwrappers (if you aren't sure, look for other entries with "/usr/sbin/tcpd" in it else use the second option):

        cvspserver stream tcp nowait root /usr/sbin/tcpd /usr/bin/cvs \
        -f --allow-root=/path/to/repository  pserver
      

      The line should look like the following (again, it will be on one line) if tcpwrappers aren't enabled on your system:

        cvspserver stream tcp  nowait root /usr/bin/cvs -f \
        --allow-root=/path/to/repository  pserver
      
    3. If the files have been inspected and have the correct entries, the inetd daemon needs to be restarted.  If you are unsure of how to do this, you can always reboot your system, else try the following commands to get the process ID and restart the daemon:

        # ps x | grep inetd
          574 ?        S      0:00 /usr/sbin/inetd
        # kill -HUP 574
      
  4. The last step in setting up the CVS server is to set up the passwd file to use for authenticating users.  To set up the passwd file, navigate to the repository you specified earlier (/path/to/repository) and create a new blank file called passwd in the CVSROOT directory.  This file will contain user names and passwords in a format similar to the /etc/passwd file (or /etc/shadow if shadowed passwords are used).  The format of the user name/password pairs will be as follows:

     USERNAME:ENCRYPTED_PASSWORD:OPTIONAL_SYSTEM_USERNAME
    

    The USERNAME and ENCRYPTED_PASSWORD will be used to log in to the CVS server, while the OPTIONAL_SYSTEM_USERNAME will be have to be an actual user with appropriate rights to work in the repository.  This optional user name is only needed if the USERNAME is not an actual user on the system.  The encrypted password is obtained in the following manner (you will need root user access):

    1. Change to the root user (will need root's password)

        #su root
      
    2. Set the password for the CVS user.  I previously added a temporary user to /etc/passwd -this user can't log in, the account is just used for getting encrypted passwords.  Add the user with the useradd command and set the user's shell to /bin/false so the account can't be used to log in.  The following shows how to create a temporary user and set the password for this temporary user and get the password (encrypted password is in red on last line, or in between the first and second colon ":") for use in the CVS passwd file:

        # useradd -s /bin/false tmpuser
        # passwd tmpuser
        New password: (Enter password for CVS user)
        Re-enter new password: (Re-enter password for CVS user)
        Password changed
        # cat /etc/shadow | grep tmpuser
        tmpuser:9/rDrLBReEkN.:11953:0:99999:7:0::
      

      The encrypted password was found in /etc/shadow as my system uses shadowed passwords; if the line with the cat and grep commands doesn't work, try replacing shadow with passwd.  Skipping the useradd command, this process can be repeated for getting encrypted passwords to fill the user list in the CVSROOT/passwd file.

    3. Now the user name and corresponding password can be entered into the passwd file in the CVSROOT/ directory of the repository.  The following shows a new user's entry in the passwd file (The admin user is an actual account on the system, but will have a different password for CVS than for system log in.):

       admin:9/rDrLBReEkN.
      

    Now the repository and server have been set up and a user has been added so he or she can log in to the system and use all the capabilities of CVS to manage their projects.  To see how to interact with CVS, see this page on CVS use.


  1. Fogel, Karl and Moshe Bar.  2001.  Open Source Development with CVS.  2nd ed.   Scottsdale, AZ: The Coriolis Group, LLC.

BACK