why is my master node idle when running gnu parallel remotely?

Question

I'm using gnu-parallel to run a number of python scripts in parallel across a number of computers in my gigabit network.

The files being processed are located on a tmpfs (RAM) folder on my master node. When I start a parallel process I can see that all the cores in the remote node are working but my master node stays idle.

Here is what my parallel line looks like:

parallel --workdir ./netshare --filter-hosts -S :,192.168.88.222\
 'myscript.py {.}.shp output_{.}.shp input.shp' ::: tile_{1..445}.shp

I've tried reversing the order of the ip adresses 192.168.88.222,: with the same result.

I've tried running only on the local master node and other than the warning that no remote nodes are available all 8 local threads start working.

I thought it might be network latency but no python scripts start at all! Whereas on the remote host I see four running without problem.

As per Ole's comment I tested the issue with bzip2, here's the command:

 parallel -S $SERVER1,: 'cd /path/to/mounted/network/share && cp largefile.avi file{1}.avi && bzip2 -v file{1}.avi && rm file{1}.avi.bz2' ::: {1..100}

The test worked exactly as expected. All cores of all hosts fired up with cp then bzip. So guess the problem lies in my python script somehow.