Nice work on this patch!
I've just looked over your work re: ganglia since it's what I'm familiar with.
The GangliaHandler in check_zookeeper.py makes the assumption that all ZK variables are uint32. Looking at the list of exported variables it appears that assumption is safe but it's something to remember. In the future, you may want to export heterogeneous metrics with strings and floats for example.
I think it might be possible to pass just the executable string 'gmetric' to subprocess.call instead of the full path '/usr/bin/gmetric'. This would allow for alternative locations of gmetric to be found using the PATH environment variable.
Your python module looks correct to me but I didn't test it.
You might want to drop a short sentence into the README that the python module only works with the 3.1.x versions of ganglia whereas the gmetric approach works for all versions.