Someone reported a problem with this some time ago and it turned out there was a bug in the documentation, which I've since fix well over a month ago. If you actually look at the documentation for the ganglia debug setting, you're specifying a 9 and the meaning of that flag are:
1 - print Var, Units and Values
2 - only print sent 'changed' Var/Units/Vales
4 - dump packet
8 - do not open/use socket (typically used with other flags)
16 - print socket open/close info
so that means you're going to have the variables printed on the terminal but they won't be sent over the socket to gmond. Change the 9 to a 1 or even a 17 (to comfirm the socket is being opened) and I think you'll be ok after that.
I also noticed you haven't specified a monitoring interval with -i so that means you'll be sampling once a second interactively or 10 seconds as a daemon. These intervals work great and I think you need to monitor at that frequency to get accurate data (but that's a whole different discussion) but it may be a lot for ganglia to digest if you plan to collect a lot of different data.
One thing you might want to think of is to generate data in parallel in a local file in plottable format by adding the collectl switches -P -f/dirname -ocz. You could then load this data into a spreadsheet, plot it with gnuplot or use colplot, which is a web-based plotting tool that call gnuplot for you. You can get it as part of 'collectl-utils' which can be found at collectl-utils.sourceforge.net.
Be sure to let me know if this doesn't work. The best way would be in the collectl forum on sourceforge or one of the mailing lists as I only stumbled on your posting by accident.
-mark