New usage model of JURECA
From December 2018 onwards, JURECA has a new usage model.
Details are explained here.
For the projects CLaMS and IEK-7 this means the following changes:
- There are no multiple accounts for the same user anymore
The username for access to JURECA has been changed to <lastname><number> for both projects JICG11 and the ICG1
User and group access management is handled through the site JUDOOR. You need to press the "agree" to the Usage agreement here.
- the old ssh key should still work
- the old HOME filesystems are now located under /p/project/cjicg11/jicg11xx and /p/project/ciek-7/icg1yy
The ERA-Interim data are now located at /p/fastdata/slmet/slmet111/met_data/ecmwf/era_interim/nc/hybrid/. Note that the access rights still need to be worked out by Lars Hoffmann.
When running a job script, you have to specify the project name to which the CPU hours are allocated ba adding the line #SBATCH --account=jicg11 or #SBATCH --account=iek-7 in the top lines
- CLaMS programs that use e.g. the dynamic NetCDF library need to be re-compield with updated Makefiles / mkincl
potentially .profile and .bashrc commandss need to be transferred to $HOME/.bash_profile, especially for module load commands that have been preformed automatially at the login.
the old filesystem $WORK has been replaced by $SCRATCH (/p/scratch/cjicg11/jicg11xx or /p/scratch/ciek-7/icg1yy) that still contain the old data. The project specific environment variables can be set by e.g. jutil env activate -p cjicg11. Suggested new directory names would be /p/scratch/cjicg11/<username> (more info on filesystems)