/autorec

Automated reconstruction job submission tool.

Primary LanguagePythonMIT LicenseMIT

autorec

Automated reconstruction job submission tool.

Installation:

  • Copy this directory and its contents to the acquiring MRI system. We'll refer to the location as AR_PATH.
  • Create AR_PATH/ar_lib/site-local.py (can be site-ANY.py). Set (at a minimum) DEFAULT_RECON_MACHINES (tuple of ip addresses.) See AR_PATH/ar_lib/arsite.py for other variables that can be set.
  • Create a pre-shared (passwordless) SSH key:
    cd AR_PATH
    ssh-keygen -f id_recon
    
    Hit return without entering a password (twice).
  • Copy the created id_recon.pub into your reconstruction user's ~/.ssh/authorized_keys (~/.ssh and ~/.ssh/authorized_keys must not be world read/writeable) on DEFAULT_RECON_MACHINES[0]. (Assuming it connects, it will attempt DEFAULT_RECON_MACHINES[1] ... if provided and [0] fails.)
  • Execute AR_PATH/ar_lib/ssh.py to test SSH connectivity.
  • Enter AR_PATH/connectionTest and run ./connectionTest (Will create files in ~USERNAME/incoming/testing on reconstruction system.)

To run automatically after an acquisition:

  • Create reconNNNN/conf.py (copy template/conf.py)
  • Edit the new conf.py to reflect your needs (files to fransfter, etc.)
  • Place a symlink to autorec.py in /usr/g/bin matching the son-of-recon (rhrecon) value; for example: ln -s AR_PATH/autorec.py /usr/g/bin/recon1234
  • Note the symlink name is used to determine what directory (and associated conf.py file) to use.
  • Logs are created in AR_PATH/reconNNNN/logs/JOB_DESCRIPTOR/

DOIs for individual releases available at: https://doi.org/10.5281/zenodo.2739230