Doug:
There's over 200 gig space on that drive, and I have since test copied over 100 Meg to the mri directory (as the same user, so permissions seem OK).
Questions:
1. Is mghWrite failure salient, or is this just mri_remove_neck checking for an existing file in order to delete it (ie: it's not really an error).
2. In general, does recon-all happily restart at wherever stage you tell it, and overwrite existing files, or is some special action necessary to deal with existing files?
Thanks,
Graham
At 5/30/2006 10:10 AM, Doug Greve wrote:
Can you verify that the disk was not full?
doug
Graham Wideman wrote:
Folks:
In our first attempt at a complete run of recon-all, we hit this odd failure: recon-all stopped after the mri_remove_neck stage, claiming that:
mri_remove_neck couldn't mghWrite to nu_noneck.mgz
But then appeared to perform the neck removal anyway and write the output
But then the subsequent step: mri_em_register failed saying that it
can't read nu_noneck.mgz
... and sure enough, there is no such file there.
The recon-all.log for these steps is copied below.
This doesn't appear to be a permissions problem, as the script happily wrote other files to the same mri directory.
I do wonder whether mri_remove_neck was actually using the correct directory (we are not using the default SUBJECTS_DIR), but the logic for that is in an executable that I don't think we have source for?
But the primary troubleshooting question is why the mghWrite failed, but processing continued anyway, with the subsequent "writing output" message suggesting success of this step (but no file to show for it).
Clues?
Thanks,
Graham