Data Management Protocol

Radboud University GDPR principles;

General security

  1. Ensure that you use the correct permission on your files/directories. Home directory: ensure that only you have write access:
    chmod o-w,g-w /home/$USER
    
    Take some time to study how these permission work and decide what you would like to have open.
  2. If you want to close your $HOME to everyone, including the group (thchem):
    chmod 700 $HOME
    
    you can still give access to the group (thchem) to some directories. This requires the executable flag for the group in $HOME:
    chmod g+x $HOME
    
    To make it easier for other to find the "thchem" directories, create a directory "projects":
    mkdir $HOME/projects
    
    make it accessible for the group:
    chmod 750 $HOME/projects
    
    and switch on the "sticky bit":
    chmod g+s $HOME/projects
    
    You can keep you thchem projects there, or you put soft links, e.g.:
    cd ~/projects
    ln -s $HOME/thchem_project1 .
    
    Of course, the "thchem_project1" directory must be open for the group (and preferably have "chmod g+s HOME/thchem_project1")
  3. The directories on scratch should have reading access to the group where the group should be "thchem". You can ensure this by running the for_theory.sh script on your directory.
  4. Do not use any third party services for email or dropbox purposes. Use your science.ru.nl or ru.nl account for email and SURFdrive and FILEsender for file transfer.

Data storage

We treat computer generated data differently from "human generated" data. The latter contains presentations, papers, reports, educational content etc., but also software and data reduction scripts etc. Computer generated data can in principle be regenerated and it typically takes up much more storage space.

Text storage

  1. Keep your text at /home. This will be regularly backed up.
  2. Each accepted paper or submitted report/thesis should also be stored at /vol/tcbackup, which is for this type of data with backup (we pay for this), and it is as safe as /home.
    1. Create a directory with our username (not your first name since this is not unique) and create per paper/report a subdirectory.
    2. Store the FINAL version (not old versions) of text and figures.
    3. Use
      cp -p file-in-your-homedir .
      
      This will preserve the timestamp.
    4. Per figure in a paper/report, it should be clear on which data this is based and with which data processing protocol this is treated (script). A README-file with a path could be enough. Make sure that it is detailed enough that a next student could reproduce it if necessary.

Software, script, input files

  1. Keep scripts for personal use at /home. This will be regularly backed up.
  2. Make sure that all data generated by you, like input files is also backup somewhere.
  3. Large software packages like MOLPRO or LAMMPS are stored at /vol/thchem. If you need additional programs to be installed, come and ask.

Computer generated data storage

  1. We have a 9 TByte filesystem for backup at cn86:/scratch-archive
  2. This is a so-called "raid1" filesystem made of two disks - this means that files are only lost when two disks fail. Note however: there is NO COPY ON TAPE (like for /home of /vol/tcbackup), so this backup still has a, relatively small, risk of failing. The largest risk is probably the risk of deleting files that should not be deleted.
  3. ONLY use this filesystem for backup, not during computations: intensive use of the disks increases the chance of losing both disks at the same time.
  4. The intended use is to keep files that normally would be on some /scratch filesystem for a well defined time: for example, if losing the files before publication of an article could be a disaster, but keeping them on a normal /scratch filesystem would take up too much space.
  5. To use cn86:/scratch-archive, create a directory with your username, make it readable by the thchem group, and switch on the sticky bit:
    cd /scratch-archive
    mkdir $USER
    chmod 750 $USER
    chmod g+s $USER
    (or run /vol/tchem/bin/for_theory.sh /scratch/$USER
    after putting the files there).
    
  6. In each "top-level" directory, put a README.txt that explains in a few words what the files are about, and how long they need to be kept. This may be a date, but also "until one year after publication" etc., but give sufficient detail: at some point in the future this disk will be full, and someone will have to decide what to keep and what to throw away.