Files fail to save directly to network from MATLAB

[j-tseng]

Affected files:

fcp_4_beamforming_tmplGrid.m

• atlas_beamforming_results.mat containing catmatrix, var_explained, srate, coords (0.364GB)

fcp_5_taskconnectivity_localSave.m

• fcp_5_conn_mat_[method].mat containing conn_mat variable for each participant (0.000235GB)
• fcp_5_allParticipants_conn_mats_[method].mat containing all_conn_mat variable for all participants (0.012GB)

make_NBS_ready.m

• [nickname]_datamatrix_[frequencies].mat containing organized participant data for specific frequency band (0.0037GB)
• [nickname]_designmatrix.mat containing design matrix (0s and 1s) for NBS statistical testing (429 bytes?)

After NBS

• [nickname]_NBSResults_[frequency].mat - saved manually from workspace (0.203GB)

So there's a mixture of size actually, maybe the error isn't directly related to size? Could be the nature of what's being written to file perhaps?

[j-tseng]

Weird. I tried every variant and somehow ended up breaking MATLAB's interpretation of the filepaths. isfolder(...) works on the symbolic link, but attempting to save to a location on there returns a "file/folder does not exist" error. Other errors that I've encountered include "unknown error" on movefile, saved but file may be corrupted and could not close...

It worked once by saving locally then using movefile, so really hard to find any logic behind why it would suddenly break. It just doesn't like the symbolic links and network locations.

[j-tseng]

we started using network fileshares as data and outputs storage and using built-in nautilus mount tools would cause issues when the /run/user/gvfs locations were adopted (it would include symbols like : and \). instead, use a /etc/fstab approach for mounting - much more stable when saving things.