Hi,
Thanks for your response.
Actually, we do use SGE software for our clustering. However we do not
submit on the default delay of 15 sec, but on a delay of only 1 sec. Could the
problem be caused by the fact that the file is not filled up with actual
commands before SGE starts running it?
We would like to try to delay the submission just before the fslsub command.
Could you help us out where this command is actually located? Or is it located
in a binary code?
Many thanks
Judith
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of
Steve Smith
Sent: zaterdag 8 augustus 2009 8:32
To: [log in to unmask]
Subject: Re: [FSL] high level FEAT
Hi - yes, the initial script fills this file with dummy arguments (so that it has
the right length for SGE to see) and a later script fills it up with the actual
commands before SGE gets as far as running it.
I'm guessing that you are not using SGE software for your clustering, so this
doesn't work on your cluster? I guess your clustering software maybe reads
the file immediately it is submitted rather than once the script is ready to run?
Cheers.
On 7 Aug 2009, at 09:11, judith wrote:
> Hi,
>
> Thank you for your repsonse and advise.
> I spoke with our systems manager about your questions of OS, hardware
> and FSL version. We also tried again together to locate and fix the
> error. We ran into the following problems:
>
>
> The program feat starts a lot of featX_YY programs. One of them is
> feat3b_flame using -t ./.flame as a command file option. However at
> the time of submitting the .flame file only contains dummy as a
> program to execute. So the jobs submitted to the cluster all fail.
> Afterwards the .flame files contains flameo commands. We have no idea
> where the latter come from. We found line in feat where dummy is added
> to the .flame file but we don't understand why this is done.
>
>
>
> We think this is the reason why the error /apps/fsl/current/bin/
> flameo -- cope=filtered_func_data --
>> vc=var_filtered_func_data --dvc=tdof_filtered_func_data --mask=mask
>> --
>> ld=stats --dm=design.mat --cs=design.grp --tc=design.con --
>> runmode=fe
> occurs in the GUI.
> Indeed when we run one of these flameo commands directly from the
> command line it works. We tested our hypothesis by changing the
> fsl_sub script to execute all jobs without submitting to the grid and
> indeed this too worked fine. For some reason no dummy commands were
> written to the .flame file at that time...
>
>
> Do you have a solution for this problem?
>
>
>
> OS: Clustervision OS 5.0 (based on Redhat Enterprise 5 and CentOS 5)
>
> Arch: 64bit
>
> FSL: 4.1.2
>
> kernel: 2.6.18
>
> Many thanks
>
> Judith & Bart (systems manager)
>
>
>
> -----Original Message-----
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
> Behalf Of Steve Smith
> Sent: woensdag 5 augustus 2009 8:02
> To: [log in to unmask]
> Subject: Re: [FSL] high level analysis FEAT
>
> Hi,
>
> This is a very strange looking error - I suspect either that there is
> something wrong with your computer or file system mounting, or it is a
> weird operating system that we've not seen before. Exactly what
> hardware and OS are you running, and which version of FSL? Have you
> tried rebooting your computer?
>
> Cheers.
>
>
> On 4 Aug 2009, at 08:10, judith wrote:
>
>> Hi,
>>
>> I am having troubles performing a high level analysis using low level
>> feat directories.
>> For low level processing I was not experiencing any trouble.
>> However, using
>> these low level feat directories in a high level analysis, for
>> example a simple average of one group using mixed effects or fixed
>> effects, the high level stats crashes at the following command:
>> /apps/fsl/current/bin/flameo --cope=filtered_func_data --
>> vc=var_filtered_func_data --dvc=tdof_filtered_func_data --mask=mask
>> --
>> ld=stats --dm=design.mat --cs=design.grp --tc=design.con --
>> runmode=fe
>> Warning: no access to tty (Bad file descriptor).
>> Using this command by entering it in the terminal, it runs without
>> any problems the stats, so the data needed are available.
>> However since poststats cannot be ran seperatly, I cannot perform the
>> analysis needed.
>> Does anyone have an idea wat I am doing wrong?
>>
>> I will add the whole command line below:
>> Higher-level stats
>>
>> cat ../design.lcon | awk '{ print }' > design.lcon
>> /apps/fsl/current/bin/flameo --cope=filtered_func_data --
>> vc=var_filtered_func_data --dvc=tdof_filtered_func_data --mask=mask
>> --
>> ld=stats --dm=design.mat --cs=design.grp --tc=design.con --
>> runmode=fe
>> Warning: no access to tty (Bad file descriptor).
>> Thus no job control in this shell.
>> dummy: Command not found.
>> Warning: no access to tty (Bad file descriptor).
>> Thus no job control in this shell.
>>
>> /bin/rm -f stats/zem* stats/zols* stats/mask* ; /bin/mv dof stats
>> Warning: no access to tty (Bad file descriptor).
>> Thus no job control in this shell.
>>
>> Post-stats
>>
>> couldn't change working directory
>> to "/mic/jvhoev2/data_taal/data/CO/FE_single_group_average.gfeat/
>> cope1.feat
>> /stats": no such file or directory
>> while executing
>> "cd ${FD}/stats"
>> (procedure "feat5:proc_poststats" line 50)
>>
>
>
> ----------------------------------------------------------------------
> ----- Stephen M. Smith, Professor of Biomedical Engineering Associate
> Director, Oxford University FMRIB Centre
>
> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> +44 (0) 1865 222726 (fax 222717)
> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
> ----------------------------------------------------------------------
> -----
>
---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering Associate Director,
Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------
|