I'm trying to run network-level MSHBM to generate individual 20-network parcellations.
I've had no issues for a majority of my participants, but in 20-30% of them, the job will just hang on "M-step...", ie will print out:
Inter-region iteration 2 ... It is EM iteration.. 1.. M-step..
and then nothing will happen for 20+ hours and i have to just kill the job.
This is a group of participants that we've QC'ed etc, and successfully run MSHBM to get 17-network parcellations on in the past. I haven't yet identified anything that the failed participants have in common, but it seems like if I re-run them they consistently fail, so i suspect it's an issue specific to the data (i've had random i/o issues in the past on this cluster , but a re-run will usually work then). I've tried increasing the memory available for the job but that hasn't helped yet and i suspect isn't the issue if others are making it through fine.
Any suggestions on how i can troubleshoot? happy to provide more info as needed.
I'm trying to run network-level MSHBM to generate individual 20-network parcellations.
I've had no issues for a majority of my participants, but in 20-30% of them, the job will just hang on "M-step...", ie will print out:
Inter-region iteration 2 ... It is EM iteration.. 1.. M-step..and then nothing will happen for 20+ hours and i have to just kill the job.
This is a group of participants that we've QC'ed etc, and successfully run MSHBM to get 17-network parcellations on in the past. I haven't yet identified anything that the failed participants have in common, but it seems like if I re-run them they consistently fail, so i suspect it's an issue specific to the data (i've had random i/o issues in the past on this cluster , but a re-run will usually work then). I've tried increasing the memory available for the job but that hasn't helped yet and i suspect isn't the issue if others are making it through fine.
Any suggestions on how i can troubleshoot? happy to provide more info as needed.