You can also serve on the next-highest available port doing something like this in Python:
import SimpleHTTPServer
import SocketServer
Handler = SimpleHTTPServer.SimpleHTTPRequestHandler
port = 8000
while True:
    try:
        httpd = SocketServer.TCPServer(('', port), Handler)
        print 'Serving on port', port
        httpd.serve_forever()
    except SocketServer.socket.error as exc:
        if exc.args[0] != 48:
            raise
        print 'Port', port, 'already in use'
        port += 1
    else:
        break
If you need to do the same thing for other utilities, it may be more convenient as a bash script:
#!/usr/bin/env bash
MIN_PORT=${1:-1025}
MAX_PORT=${2:-65535}
(netstat -atn | awk '{printf "%s\n%s\n", $4, $4}' | grep -oE '[0-9]*$'; seq "$MIN_PORT" "$MAX_PORT") | sort -R | head -n 1
Set that up as a executable with the name get-free-port and you can do something like this:
someprogram --port=$(get-free-port)
That's not as reliable as the native Python approach because the bash script doesn't capture the port -- another process could grab the port before your process does (race condition) -- but still may be useful enough when using a utility that doesn't have a try-try-again approach of its own.