Dear all,
We're having difficulties getting FSL to work with the SGE on our cluster and was hoping someone could offer some insight.
In brief, fsl_sub appears to be working, with no error messages. For example, we get the following text in our output log files when we run feat:
usr/share/fsl/5.0/bin/fsl_sub -T 10 -l logs -N feat0_init /usr/share/fsl/5.0/bin/feat /psyhome/u2/lee/fsl-feeds/results/fmri.feat/design.fsf -D /psyhome/u2/lee/fsl-feeds/results/fmri.feat -I 1 -init
18260
/usr/share/fsl/5.0/bin/fsl_sub -T 48 -l logs -N feat2_pre -j 18260 /usr/share/fsl/5.0/bin/feat /psyhome/u2/lee/fsl-feeds/results/fmri.feat/design.fsf -D /psyhome/u2/lee/fsl-feeds/results/fmri.feat -I 1 -prestats
18371
/usr/share/fsl/5.0/bin/fsl_sub -T 1 -l logs -N feat3_film -j 18371 /usr/share/fsl/5.0/bin/feat /psyhome/u2/lee/fsl-feeds/results/fmri.feat/design.fsf -D /psyhome/u2/lee/fsl-feeds/results/fmri.feat -I 1 -stats
20203
/usr/share/fsl/5.0/bin/fsl_sub -T 119 -l logs -N feat4_post -j 20203 /usr/share/fsl/5.0/bin/feat /psyhome/u2/lee/fsl-feeds/results/fmri.feat/design.fsf -D /psyhome/u2/lee/fsl-feeds/results/fmri.feat -poststats 0
20451
However, in reality, no jobs are actually submitted to SGE and the jobid that is logged is merely the PID from top (i.e. jobs end up running serially directly on server).
We know our SGE is working as using the command qsub -q submits jobs to the named queue.
It just seems fsl_sub is not submitting to the queue, despite no error messages to the contrary.
Even if we type a super simple command like:
fsl_sub -T 1 -v mkdir temp
The screen output will be:
Estimated time was 1 mins: queue name is veryshort.q
executing: mkdir temp
77992
The job will have run, but again not on veryshort.q as indicated but directly on the server with PID 77992.
Any insight much appreciated!
Best,
Andy
|