Lots of defunct processes

Hi folks,

I’ve a student utilizing fmriprep, and the single fmriprep has generated close to 400 processes that have become defunct.

fmriprep was invoked with:

fmriprep /data /out participant --participant_label PM226OX2 --nthreads 1 --template-resampling-grid 2mm

What is up with all of these processes? They are things like (the trailing number is the coutn of defunct with that name):

apas2aseg 6
bbregister 15
cat 16
grep 172
IsLTA 64
main’ 1
mri_info 5
mri_motion_corr 6
mri_nu_correct. 3
mris_info 6
pctsurfcon 4
python 2
recon-all 37
reg2subject 32
talairach_avi 7
tcsh 7
vertexvol 14

Also, the student reports anecdotally that subsequent processing seems to take longer and longer.

It is fmriprep version v1.3.0.post2 running inside docker.

Clues on how to make this run cleaner?


This looks like the collaboration of hitting the memory limits and a Python bug (although I’m surprised given that --nthreads 1).

Could you check that the docker memory limits have been revised and processes can access more than 8GB?

It would also be really helpful to see the output log from fmriprep.