In your previous post you mentioned core files being created. You should
also have SGE job failure messages in the form of emailed digests which
usually indicate what went wrong, esp. if SGE decided to kill the tasks
because they exceeded some limit associated with the queue, e.g, maximum
run time, or if they ran out of RAM, etc.
Also each SGE job will have a <job_name>.o<job_id> and
<job_name>.e<job_id> files. These may have some useful info.
On 03/17/2009 12:50 PM, Mahinda Yogarajah wrote:
> Hi Steve and Other Expert Users,
>
> We (I and our local IT experts) have been tinkering for about a week without
> success. We are still having trouble getting our SGE to run the tbss 2
> script (FNIRT) without eventually crashing and failing to complete the job.
> Our images our 128x128, and the machines should have enough RAM (4gb with
> 10 swap) to cope. We have also tried upgrading the SGE software, but
> without success. I wonder whether anyone had any other suggestions ?
>
> Thanks in advance.
>
> Mahinda
>
>
>
--
Dave Flitney, IT Manager
Oxford Centre for Functional MRI of the Brain
E:[log in to unmask] W:+44-1865-222713 F:+44-1865-222717
URL: http://www.fmrib.ox.ac.uk/~flitney
|