Users have access to multiple clusters. Each cluster has its own storage systems mounted on
/share directories. In order to provide a unified file system so that you can manage all your data from a central place we created links inside each user's
home directory. When you ssh to login node, your home directory contains the following:
[email@example.com ~]$ ls -lhta total 0 drwxrwxr-x. 2 demo demo 29 Sep 5 02:15 cluster-001 drwxrwxr-x. 2 demo demo 29 Sep 5 02:15 cluster-003 drwxrwxr-x. 2 demo demo 29 Sep 5 02:15 cluster-005 lrwxrwxrwx. 1 demo demo 13 Sep 5 02:15 dropbox -> /dropbox/demo [firstname.lastname@example.org ~]$ ls -lhta cluster-001 total 4.0K drwx------. 9 demo demo 4.0K Sep 5 02:15 .. drwxrwxr-x. 2 demo demo 29 Sep 5 02:15 . lrwxrwxrwx. 1 demo demo 18 Sep 5 02:15 share -> /cluster-001-share lrwxrwxrwx. 1 demo demo 22 Sep 5 02:15 home -> /cluster-001-home/demo
In each cluster directory there are 2 subdirectores,
share. If you want to run a job on cluster-001, you should navigate to either
cluster-001/share directory. Please note that files and directories on
cluster-001 are not accessible on other clusters.
If you have files that you would like to access from all clusters, for example - pseudopotentials, basis sets, input scripts, you can put them inside
dropbox directory is accessible on all clusters from the same path:
dropbox directory is limited to 1GB by default and it is not designed to handle large files or, moreover - to run calculation in.