docker container for VR-CESM docker container
VR-CESM historic with CAM6 and CLM5 (no ocean) using cesm2_2_beta01.
CESM docker container for HIST_CAM60_CLM50%BGC_CICE%PRES_DOCN%DOM_MOSART_CISM2%NOEVOLVE_SWAV compset and resolution ne0uoslone30x8_ne0uoslone30x8_mt12.
- Input dataset is stored and available in zenodo (38.7 GB)
- The CESM version used is cesm2_2_beta01 that can only be accessible if you authenticate yourself properly. We pass a tarball containing .subversion folder which we used to authenticate ourselves.
First you need to instal subversion:
sudo yum install subversion -y
svn co https://svn-ccsm-models.cgd.ucar.edu/cam1/trunk_tags/cam6_1_014/components/cam
Remark: you may have to install subversion on your virtual machine:
sudo yum install subversion -y
Enter the correct username and password. Then create the corresponding tarball:
cd $HOME
tar cvf subversion.tar .subversion
We then save it in a folder that we pass when running the docker container:
mkdir -p /opt/uio/svn_config
mv subversion.tar /opt/uio/svn_config/
Make sure inputdata is available (it won't download it as we suppose it is already on disk).
- The location of the inputdata is
/opt/uio/inputdata
- The folder
vr_cesm_config/
contains the changes required for running our Variable resolution grid (made by Colin M. Zarzycki) - Model outputs are stored in
/opt/uio/archive
along with thecase
folder (it can be interesting to check timing).
Important: the folder /opt/uio/archive needs to be writable by unix group users
(see Dockerfile) otherwise you will get a permission denied when running.
sudo chgrp -R users /opt/uio/archive
sudo chmod -R g+w /opt/uio/archive
You can check it:
ls -lrt /opt/uio | grep archive
You should have:
drwxrwxr-x. 8 centos users 4096 Nov 9 15:21 archive
docker pull nordicesmhub/cesm_vr:latest
docker run -i -v /opt/uio/inputdata:/home/cesm/inputdata -v /opt/uio/archive:/home/cesm/archive \
-v /opt/uio/svn_config:/home/cesm/svn_config -t nordicesmhub/cesm_vr:latest
- We are running 1 day (144 timestep of 30mn) using 8 processors. It takes about 13GB per processors.