We make several assumptions about your network:

1. Your jobs should be able to run on any machine
   I.e., don't try to mix Linux and Solaris object code
2. You should have the same username on each machine
3. The machines should have a shared file system such as NFS
   Your project files and home directory should be shared
4. You should be able to SSH to any machine without entering a password
   How do I set this up?

Supporting software

You'll need this installed on all the machines you want to use. I use the versions listed below, but later versions may also work. It's easiest to install these to some shared location on your NFS, and add the appropriate directories to your PATH, etc.

1. Ruby 1.9.1-p129 (with pthreads enabled)
   
2. Open4 library for Ruby
   This is a available as a "Gem" for Ruby
   Typically, you can install it by running gem install open4 after installing Ruby.
   See rubygems.org for more info.

Addjob setup

1. Just edit your login scripts and add the addjob-x.xx directory to your PATH.

You'll finally need to edit addjob-x.xx/Config.rb and enter the names of the machines you want to use. Some options are available for controlling how heavily each machine is used -- see the comments in the configuration file for more details.

You can monitor your machines by pointing your web browser to

http://your-server:33333

Where your-server is the machine you ran the addjob-server program on. Here's a screenshot of what it looks like.

If your program needs special environment variables like $PATH, you'll need to put them into your .bashrc instead of your .bash_profile. Your .bash_profile will not be run by addjob jobs.

It's easiest (and most efficient) to just quote the whole job to do IO redirection. For example:

addjob "myprogram --foo --bar < infile > outfile"

This way, all the redirection is handled on the remote machine instead of across the network.