External Email - Use Caution
Dear Douglas,
thank you for confirmation.
My initial suspicion that fs_time caused the error was because of this part of the script:
if($#outfile == 0) set outfile = /dev/stdout
where it explicitly uses /dev/stdout when no output file is provided. fs_time from Freesurfer 6 doesnt do that.
And it was pretty much confirmed when I rewrote that part of fs_time:
if($#outfile == 0) set outfile = /dev/stdout; # time default is stderr, which is annoying set cmd = ($cmd -o $outfile)
changed to:
if($#outfile == 0) then set cmd = ($cmd 2>&1) # time default is stderr, which is annoying else set cmd = ($cmd -o $outfile) endif
and it worked.
From what I was able to find, it looks like '/dev/stdout` might not work on all file systems, and the one our HPC cluster is using might be one of them.
Best regards,
Renat.
________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu freesurfer-bounces@nmr.mgh.harvard.edu on behalf of Douglas N. Greve dgreve@MGH.HARVARD.EDU Sent: Friday, March 19, 2021 5:13 PM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] Freesurfer 7.1 recon-all -long exits with erros
fs_time will not affect any processing, so you could replace it with a stub. But why do you think it is fs_time?
On 3/19/2021 10:54 AM, Yakupov, Renat /DZNE wrote:
External Email - Use Caution
Dear Freesurfer developers,
I am trying to run the longitudinal processing pipeline of our data on our HPC cluster (SGE based) and it crashes at the 3rd step - generating the longitudinal runs. The exact command is 'recon-all -long tp1dir basedir -all'. There is no error message on the screen, it just says that it exited with errors.
The recon-all.log file has no error information either, other than the statement that it exited with errors. It seems to have stopped after mri_nu_correct.mni step and looking further it looks like it crashed at the beginning of orig_nu step, since orig_nu.log has this error message:
/usr/bin/time: /dev/stdout: Not a directory
Please see attached log files.
Additional information:
Freesurfer version: freesurfer-linux-centos6_x86_64-7.1.1-20200723-8b40551
uname -a:
Linux compute-0-2.local 2.6.32-504.16.2.el6.x86_64 #1 SMP Wed Apr 22 06:48:29 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux
lsb_release -a: LSB Version: :base-4.0-amd64:base-4.0-noarch:core-4.0-amd64:core-4.0-noarch:graphics-4.0-amd64:graphics-4.0-noarch:printing-4.0-amd64:printing-4.0-noarch Distributor ID: CentOS Description: CentOS release 6.6 (Final) Release: 6.6 Codename: Final
What is interesting is that running it on the headnode directly without qsub works fine. The headnode has the same OS and tcsh versions.
In my rather poorly informed opinion, it looks like fs_time is the culprit. For a quick fix, is it safe to assume that using fs_time from Freesurfer 6 wont affect the processing pipeline in any way?
Thank you!
Best regards, Renat.
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edumailto:Freesurfer@nmr.mgh.harvard.edu https://secure-web.cisco.com/1fBaAVs4m-N9pEZRkGjuMLPTYrzczlfqONTjqSBIg5ryMEI...