Suppose you have two projects, named "one" and "two", in a single
CVS repository named "common".  Your directory structure looks something
like this:

    /home
                              |- one
                              `- two

In other words, the CVS repository is rooted at /home/cvs/common.
All the source files are in /home/cvs/common/one and /home/cvs/common/two
and the administrative files are in /home/cvs/common/CVSROOT.
Our objective is to split this one repository into two, as follows:

    /home
                   |       `- one
                   |
                   `- two
                           `- two

In the new setup, the first CVS repository is at /home/cvs/one and
the second CVS repository is at /home/cvs/two.  Each has its own
subdirectory for content and for administrative files.

The first step is to create the new directories and copy in the
appropriate files.

    mkdir /home/cvs/one
    mv /home/cvs/common/one /home/cvs/one
    mkdir /home/cvs/one/CVSROOT
    cp /home/cvs/common/CVSROOT/* /home/cvs/one/CVSROOT
    mkdir /home/cvs/two
    mv /home/cvs/common/two /home/cvs/two
    mv /home/cvs/common/CVSROOT /home/cvs/two
    rmdir /home/cvs/common

Notice that the content files for each project were moved into
their appropriate CVS repository but that the administrative
files were copied into both of the new repositories.  The next
step is to modify the CVSROOT/history files in each of the new
repositories so that they only contain information about the
files in their respective repositories.

    cd /home/cvs/one/CVSROOT
    mv history history.orig
    grep 'one/' history.orig >history
    cd /home/cvs/two/CVSROOT
    mv history history.orig
    grep 'two/' history.orig >history

We're done.  Now each project is in its own CVS repository.  Now
you can create a separate CVSTrac database for each project.