Hi all,
I am getting the shared libraries error when I run fmriprep (v1.3.2) on the cluster (Sherlock). Not sure how to fix it. Please help.
I built the singularity image using:
singularity build /my_images/fmriprep-v1.3.2.simg docker://poldracklab/fmriprep:latest
Interestingly, the same image ran perfectly last week on another dataset. However, it keeps crashing with the following error with the current dataset.
Standard error:
fslmaths: error while loading shared libraries: libnewimage.so: cannot open shared object file: No such file or directory
Return code: 127
One simple thing to check is whether you have the "/my_images/" directory/folder. "/my_images/fmriprep-v1.3.2.simg" should be where you want to store your image and differ between computers/users.
Thanks, yes I do. "/my_images" is only a placeholder in the command I pasted above.
Sorry, I misread what your issue is. What command do you use to run fMRIprep? I encountered similar problems and the cause is that I was not using srun before the singularity command.
Did you try something like
srun singularity run --cleanenv /my_images/fmriprep-1.3.2.simg \
path/to/data/dir path/to/output/dir \
participant \
--participant-label lab
-w path/to/temprory_working_dir
Thanks. Not sure about srun. But using --cleanenv flag helped.
Yes, this is likely some LD_* environment variable leaking into the container.