BEAR Portal: Portal Job Output

The BEAR Portal provides web-based access to a range of BlueBEAR services. This page will help you move your default portal storage area to give you more space.

Job Information

A directory is created for each interactive job, which contains the job’s scripts as well as the Slurm information for the underlying batch job. You can navigate to this directory from directly within BEAR Portal by following the link in the running job titled ‘Session ID’ (see the image below).

The key files in this directory are as follows:

  • output.log: Slurm job output
  • output.log.stats: Slurm job stats information
  • launch_wrapper.log: Information on the running session (JupyterLab job only)

Please note that in the case of the BlueBEAR GUI application the above files will only contain information relating to the underlying Slurm job (e.g. reasons for unexpected termination of the session) – for specific output from the applications you're running within the GUI you will need to refer to the graphical Terminal client, also within the GUI.

Moving your Default Portal Output Folder

By default, jobs run on the BEAR portal store their output in a folder called:

 ~/ondemand/

in your home directory. Because quotas on home directories are limited, you may wish to move this directory to an RDS or CaStLeS project directory.

1. Log on to the portal at portal.bear.bham.ac.uk.

You can see which project paths are available to on the Files menu:

portal-files

Decide where you want to move the ondemand folder too, then open shell by connecting to BlueBEAR directly or through the portal at: Cluster -> BlueBEAR HPC Shell Access.

2. Make sure you don’t have any existing ondemand jobs running (you can do this by clicking on My Interactive Sessions in the portal). Type these 3 commands:

 SRC=~/ondemand
DST=<YOUR PROJECT DESTINATION FROM STEP 1>
rsync --stats --archive $SRC/ $DST/ondemand-$USER

This will make a copy of your ondemand folder in the project directory. If you have a lot of small files in your ondemand folder the last command may take a minute or two. 

3. Now you can move the old folder out the way and link the new folder in its place, with these 2 commands:

 mv $SRC $SRC.old
ln -s -T $DST/ondemand-$USER $SRC

4. Run a test job to ensure your data is now available in your project folder.

Colleges

Professional Services