Hi all,
I use a few SPM scripts wrapped as in example 1 here:
http://nipy.sourceforge.net/nipype/devel/matlab_interface_devel.htmlWhen running them as part of a workflow (either serially or in parallel), they sometimes "hang". CPU drops to baseline, no further jobs are executed or finished. When I ctrl-c the script, it exists at sleep(2) on line 260, base.py. Further investigation revealed that this is due to "jobids" on line 326 always being empty.
Strangely, as soon as I cancel the script, CPU use increases. The top command says that MATLAB is working very hard on something as it should be. It is as if MATLAB is being held back.
If I keep re-running and canceling, I make progress through the workflow i.e. MATLAB nodes that previously "hanged" are completed. I'd appreciate any thoughts anyone might have regarding this issue.
Best,
Dan
Here is the output of nipype.get_info()
{'commit_hash': '<not found>',
'commit_source': '(none found)',
'networkx_version': '1.7',
'nibabel_version': '1.3.0',
'numpy_version': '1.7.1',
'pkg_path': '/usr/local/lib/python2.7/dist-packages/nipype-0.9.2-py2.7.egg/nipype',
'scipy_version': '0.12.1',
'sys_executable': '/usr/bin/python',
'sys_platform': 'linux2',
'sys_version': '2.7.3 (default, Feb 27 2014, 19:58:35) \n[GCC 4.6.3]',
'traits_version': '4.3.0'}