RFE: please document best practice for local CERN use with AFS/EOS

[jmuf]

(CERN AFS support here): we've recently had two users wondering how to run "flashgg" at CERN, and in particular, how to best handle the split between local HTCondor submit file (must for now be on AFS) and input/output data files (should be on CERNBox/EOS, without intermediate copy on AFS). Perhaps you could - in collaboration with CMS computing - add some "best practice for local CERN users" FAQ?
Thanks in advance.