Hi,
This isn't something that I've specifically seen before, but I guess this is something that isn't often done.
I suspect it is due to jobs interrogating the file system at the same time, or very similar times, and it is probably enough to put in some delays to your jobs (either random or incremental by ROI) to try and avoid these clashes. Alternatively, if the outputs are all correct, but just have unwanted + or ++ prefixes, then you could write a little script to rename these afterwards.
All the best,
Mark
On 9 Jun 2014, at 18:19, Lutz Jacqueline <[log in to unmask]> wrote:
> Dear Experts
> I am encountering a problem when trying to run several feat_queries on a computer cluster.
> Each job touches on the same first level feat directories to get ROI data, but the featqueries have different ROI as the argument. For the output I create independent folders named after the specific ROI. So the jobs don't touch each other on the output level. Still, if i do this, i often get +output or even ++folders for ROIs. (Not all the time, no apparent pattern for what subject or what roi).
> So I assume it could be because different instances of feat_query happen to look into the same subject first level folder at the same time?
> Are you aware of such a behavior for FSL/featquery?
> And if so, is there a workaround or option for featquery to solve this? It would be very handy to run several featqueries's for different ROI simultaneously.
> Thank you
> Jacqueline Lutz
|