Hello,
I am getting the following warning messages every time I run fmriPrep (I’ve already tried this on several subjects, both using Docker on my computer and using Singularity on my institute’s cluster, and I get exactly the same errors):
201213-09:35:54,628 nipype.workflow WARNING:
Could not retrieve profiling information for node “fmriprep_wf.single_subject_07_wf.bidssrc” (mapflow 1/1).
201213-09:35:54,629 nipype.workflow WARNING:
Could not retrieve profiling information for node “fmriprep_wf.single_subject_07_wf.bids_info” (mapflow 1/1).
These warnings are followed by thousands of similar warnings, all starting with “Could not retrieve profiling information for node”, about nodes in other workflows within the single subject workflow.
These warnings are followed by a message that fmriPrep finished successfully. Indeed, even though I get all these warning messages, I still get an output of fmriPrep that looks good (at least everything in the html report looks reasonable).
I’d really appreciate help, just to let me know if I should be careful when using these outputs - or even better, any help explaining why I get these warning messages.
Hi @alonbaram, please post the command line you are executing to be sure of the accuracy of our answers.
Let’s read the warning message:
Could not retrieve profiling information for node
I assume you are executing with the --resource-monitor flag enabled. This is telling you that resource monitoring (i.e., profiling of the node) did not work. I would advice not to use this flag unless you are really getting into the weeds of resource management. Also, we have had no bandwidth to properly support such a feature, so, again, if you don’t know what you’re doing, --resource-monitor might not be a good idea.
That said, this profiler should not have any influence on the results - so, answering your question, this warning would NOT be a good reason not to trust the outputs.
EDIT: sorry, in my first post I missed that crucial “NOT” in the last line.
Hi,
I was indeed using the --resource monitoring flag. Good to know I shouldn’t use it, and even better to know I can trust the results 
Thanks a lot!
The command I was using is:
singularity run --cleanenv -B /vols/Scratch/abaram:/vols/Scratch/abaram
/vols/Scratch/abaram/my_images/fmriprep-20.2.1.simg /vols/Scratch/abaram/twoStep/BIDS
/vols/Scratch/abaram/twoStep/BIDS/derivatives participant --participant-label subjectTag \
--fs-license-file /vols/Scratch/abaram/fslicense.txt -w /vols/Scratch/abaram/twoStep/fmriPrep_workDirs/sub-{subjectTag}
–use-aroma --error-on-aroma-warnings --skip_bids_validation --resource-monitor --medial-surface-nan --mem 10 --nprocs 2
I didn’t say this, I said that this is not a good reason not to trust the results…
I wouldn’t dare to say whether you must trust the results. Indeed, fMRIPrep aims to be transparent so that you (as in, the researcher,) trust the results because you can check every step and fully understand the processing, not because “that guy who is a developer of fMRIPrep just told you so on NeuroStars” :D.