Software Installation

The software installation installs Pike and Roxen on your system. It is either done through a binary or source package. Binary packages are precompiled for each operating system and are the easiest to install. Source packages are compiled by the user and require a complete development environment. They are for people who want full control over what they install, or a last resort if a binary package fails.

Some functions, such as the Java, PHP4 and database support, require external systems or libraries to work. These systems or libraries should be installed before Roxen. If a library is installed after Roxen it might become necessary to reinstall Roxen or Pike. Fortunately the reinstallation is simple and does not affect the web sites you have created.

Unix Binary Installation

The Unix binary release is distributed as a self-installable archive, actually a combination of a sh-script and a tar archive. The result is an executable with its own installation script that guides you through the installation.

To install the binary release, simply run the archive:

It will take a while for the script to extract the Pike binary and start the real installation script. When it has finished it will prompt for the location of your installation:

If you do not enter anything Roxen will be installed in /usr/local/roxen/. Roxen is always installed in a directory called roxen.

Next it will ask you if you want to install Pike separately, as opposed to only as a part of Roxen. Installing Pike separately is recommended since it will make it possible to run other Pike programs as well. If you answer yes you will be prompted for the location of the Pike installation.

Now the actual installation of the software will take place. This takes a while and produces some more or less interesting messages. Next the installation script will create an administration interface server. If you were updating or reinstalling rather than installing a new 2.0 server you can abort the script here.

The script will prompt for the name and URL of the administration interface server, as well as the user name and password for a administraion interface user. Note that it is no longer necessary for the administration interface to run under its own port, you could assign it a URL on your normal server, for example http://my-domain/admin/. See the Ports chapter for more information.

You should probably use HTTPS for the administration interface, so that it is not possible to eavesdrop on it. Don't forget to change the certificate of that port, since it will use the unsecure demo certificate if you don't. See the Ports chapter for further information.

The last stage of the installation process asks whether you want your Roxen to connect to update servers automatically or not. If you allow it the builtin update client will fetch information about new software and updates as well as news about Roxen. The information will be displayed in the Admin tab of the administration interface.

If you don't allow this it is still possible to use the update client, it will connect to the update server and fetch information when you tell it to. The disadvantage is that you won't get automatic alerts about new security fixes and other important news.

If you allow the update client to do automatic connections the script will also help you create a Roxen Community user. The Roxen Community, http://community.roxen.com/, is a community for Roxen users. The update server is part of Roxen Community.

Now the installation is complete and the Roxen server will be started. You can connect to the administration interface by connecting to the port you configured. If you cannot connect you should look at the debug log file, roxen/logs/debug/default.1.

Unix Source Installation

The Unix source release is distributed as a tar.gz archive. It contains the source code for the Pike interpreter, necessary libraries and the Roxen server itself. The installation will work on any system, but libraries for all supported functionality is not included. Thus libraries for native database support, for image file format support (TIFF, JPEG), TrueType rendering, Java, PHP4 or GTK/GNOME support is not included.

If you are installing on a Linux or FreeBSD system most libraries will be shipped with the distribution. The commercial Unix system does however usually not ship these libraries, so you will have to install them yourself. See Appendix foo for a complete list of libraries used by Pike/Roxen.

It is necessary to install the external libraries before you install Roxen. If you later find out that some function is lacking because of a missing library you will have to reinstall Roxen, or rather the Pike interpreter. A reinstallation is done in the same way as an installation and does not affect the web sites you have created. The installation script will find the previous installation of Roxen and upgrade rather than overwrite it.

The installation process consists of four steps; configuration, compilation, software installation and Roxen installation.

The configuration is done automatically by a script created with GNU autoconf. It finds out about your system and how to compile on it. It will find which libraries are installed and if they can be used. Usually this will be fully automatic but in some cases it might be necessary to manually force the configuration system to find some libraries. See appendix bar for a list of configurations options, or use ./configure --help.

The configure script produces a log about what happens. A longer version of the log is also stored as config.log. If the configure script failed to find a certain library that does exist the config.log file will give you valuable clues to why, and what you can do to fix it. Once you have fixed the problem you can run configure again.

After the software is configured its time to compile it. This is done by invoking make which will in turn use the necessary compilers, linkers and whatnot. The compilation takes a while and produces another long log of information. Usually all messages can be ignored, but in case of errors they will provide important clues to what went wrong.

When the software has been compiled it is time to install it. This is usually done by make install. The make command, that is used to handle the compilation process, is used to handle the installation process as well. The software will be installed in the directory that you specified with the --prefix= argument to configure, by default /usr/local/.

Roxen and Pike will both be installed separately, in their own directories (pike/ and roxen/). A link to the pike binary will also be created in bin/. The roxen directory can later be moved anywhere on your system. However the pike directory cannot be moved. In case you want Pike and Roxen in different places you should set the --prefix= argument to the configure script for Pike, not Roxen.

Upgrading to a newer 2.0 release, or reinstalling the same release, is also done with make install. Only the roxen/server/ directory, that contains the actual software, will be affected. No logs, configurations or user data will be touched. After the installation the old version of the software will be moved from roxen/server/ to roxen/server.old/.

The installation also produces a progress log that is mainly interesting in case of errors.

The next step is to create an administration interface server. If you are upgrading or reinstalling you do of course not need to, your old configuration will do. This is done by invoking the create_configinterface script located in the roxen/server/ directory. If is the same script that is used by the binary installation, so we won't describe it in detail again.

The last step of the software installation is to start the server and connect to the administration interface. This is done by invokning the start script in the roxen/server/ directory. The progress of the server can be seen in the debug log, roxen/logs/debug/default.1