1. How much disk space I can use?
Each HPC user is given 30 GB home directory. The home directory is backed up. A user can ask for extra home space, usually only for a limited amount of time (write to email@example.com). If you instead need space for temporary files produced during a computation, please consider to ask for space on the scratch filesystem.
2. How can I know how much disk space I’m currently using?
Log in, and in a terminal type
getquota_zhome.shYou will get an output like:
You are using 12.34 GB of 30.00 GB.
If you never asked for extra space, your quota is the default 30 GB. Please take care of not exceeding your disk quota. If this happens, your jobs will be killed for lack of disk space, and you’d likely not be able to run any other program and even log-in.
3. I need scratch space.
Write to firstname.lastname@example.org asking for space on the scratch filesystem, specifying the reason why you need it. A directory named after your username will be created under /work1 or /work3. Please read the readme.txt file, in the relevant scratch directory, before starting using it. There is no quota on either scratch filesystem, but the performance of the programs that use scratch degrades when there is not enough free space. So delete old or unnecessary files, and remember that scratch areas are NOT backed up.
4. How can I know how much disk space I’m currently using on scratch?
Open a terminal, and run the relevant command:
This will give you an output like:
user/group || size || chunk files name | id || used | soft | hard || used | soft | hard --------------|------||------------|----------|----------||---------|---------|--------- s123456 | 54321|| 35.11 GB | 0 Byte| 0 Byte|| 50905| 0
5. I cannot access my /SCRATCH /work1 /work2 /work3 directory from thinLinc.
It is correct, the scratch directories are not accessible directly from the File Manager running on thinlinc. To be able to access them, first open a terminal:
Applications Menu-> DTU -> xterm (application node)
and then, in the terminal type:
gnome-commander -l /work1/$USER
to access your directory on /work1 (and the same for /work2, of course).
This command opens a file manager on the cluster nodes that have all access to the scratch filesystems.
6. I accidentally deleted a file/directory in my SCRATCH folder!
Sorry, there’s nothing we can do. Scratch is meant for temporary files, and there is no back-up of any scratch filesystem. Always remember to make a copy of important files.
7. I need to copy a large amount of files/data to and from the cluster.
Please do not use the login node, but use instead the dedicated server
This will give better performance (because it’s connected to 10Gbit-Ethernet _and_ QDR-Infiniband) and does not interfere with the login of the other users. From this you can access the home, SCRATCH and work1 directory exactly as from the normal login node. If you are copying a large amount of files/directories, please use the