Dos and dont’s

Users connect to the Bluecrab cluster via “login.marcc.jhu.edu”, which is a gateway (Virtual machine). It checks authentication and authorization and then connects the user to the cluster head or login nodes.

Users may run short, non-memory intensive processes interactively on login nodes. For example, editing files, looking at output files, checking jobs and running some small scripts. Login nodes should not be used to run processes that consume excessive resources. Login nodes are used by hundreds of users to interact with the queuing system and compute nodes. These processes that utilize excessive resources create frustration as the system (login nodes) may not respond quickly enough.