Dear Conn Experts,
Thank you for the continued, great support on the forum.
I am having a problem with parallelizing a second-level analysis-
Essentially I am trying to run a second level identity matrix analysis accross a large number seeds in a seed-to-voxel analysis (the example I attach 'secondlevel_eye_batch_tocluster.m' has only four seeds but a larger analysis is planned). There are 40 subjects in the analysis.
I run the attached script from matlab on my University computing environment -
the command window on matlab says the node.**.sh file is submitted and finished
The output log (found in the qlog folder) states the job finishes succesfully.
However matlab remains busy and the folder 'second_level_eye' is not generated- at first I thought perhaps it was just taking a long time to merge but even when I run the script with the four seeds the script has never succesfully completed and the folder and files are not generated- even after being left for several hours (much longer then I would expect a four seed analysis to take)
The command window output looks like this....
'[i]secondlevel_eye_batch_tocluster %.m file run[/i]
[i]saved /.mat[/i]
[i]/.qlog/200817112158797/node.0001200817112158797.sh submitted[/i]
[i]Waiting for grid/cluster jobs to finish...[/i]
[i]conn_0001200817112158797 submitted[/i]
[i]conn_0001200817112158797 running[/i]
[i]conn_0001200817112158797 finishing[/i]
[i]conn_0001200817112158797 finishing[/i]
[i]conn_0001200817112158797 finishing[/i]
[i]conn_0001200817112158797 finishing[/i]
[i]conn_0001200817112158797 finishing[/i]
[i]conn_0001200817112158797 finished[/i]'
Some more information - when I run setup, denoise and 1st level on a grid engine from the GUI everything runs perfectly well but I can't see a way to run a second level analysis from the GUI and use the default cluster computing option.
The default HPC setting is for a grid engine computer cluster
If you could point me in the direction of potential source of the problem or any steps I can take to troubleshoot it would be greatly appreciated.
Best wishes,
Mike
Thank you for the continued, great support on the forum.
I am having a problem with parallelizing a second-level analysis-
Essentially I am trying to run a second level identity matrix analysis accross a large number seeds in a seed-to-voxel analysis (the example I attach 'secondlevel_eye_batch_tocluster.m' has only four seeds but a larger analysis is planned). There are 40 subjects in the analysis.
I run the attached script from matlab on my University computing environment -
the command window on matlab says the node.**.sh file is submitted and finished
The output log (found in the qlog folder) states the job finishes succesfully.
However matlab remains busy and the folder 'second_level_eye' is not generated- at first I thought perhaps it was just taking a long time to merge but even when I run the script with the four seeds the script has never succesfully completed and the folder and files are not generated- even after being left for several hours (much longer then I would expect a four seed analysis to take)
The command window output looks like this....
'[i]secondlevel_eye_batch_tocluster %.m file run[/i]
[i]saved /.mat[/i]
[i]/.qlog/200817112158797/node.0001200817112158797.sh submitted[/i]
[i]Waiting for grid/cluster jobs to finish...[/i]
[i]conn_0001200817112158797 submitted[/i]
[i]conn_0001200817112158797 running[/i]
[i]conn_0001200817112158797 finishing[/i]
[i]conn_0001200817112158797 finishing[/i]
[i]conn_0001200817112158797 finishing[/i]
[i]conn_0001200817112158797 finishing[/i]
[i]conn_0001200817112158797 finishing[/i]
[i]conn_0001200817112158797 finished[/i]'
Some more information - when I run setup, denoise and 1st level on a grid engine from the GUI everything runs perfectly well but I can't see a way to run a second level analysis from the GUI and use the default cluster computing option.
The default HPC setting is for a grid engine computer cluster
If you could point me in the direction of potential source of the problem or any steps I can take to troubleshoot it would be greatly appreciated.
Best wishes,
Mike