Linux Basics
Working on SCDF interactive nodes means working on a shared, multi-user Linux system. If you are new to that, the reference material below covers the fundamentals: process control, keeping work alive across a dropped connection with screen or tmux, shell configuration, and process monitoring.
For connecting to the facility in the first place, see Accessing Clusters Through the Gateway.
Run long work in the batch system
Interactive and login nodes at the SCDF are shared and are intended for debugging, testing and short jobs.
screen and tmux keep a session alive across a dropped connection, but they are not a substitute for the batch system. If your work runs for more than a few minutes, submit it rather than running it interactively:
Batch jobs get dedicated resources, survive your logout without a multiplexer, and do not compete with other users on the interactive nodes.
Your experiment may add its own requirements — FCC users, for example, must set an accounting group on every job. Check your experiment's documentation under Projects & Experiments.
Linux reference material
Our colleagues at US ATLAS maintain a clear, well-written guide to working on a shared Linux system as part of their Analysis Facilities documentation. Rather than duplicate their work, we link to it directly:
| Topic | What it covers |
|---|---|
| Linux Process Control | Control+C vs Control+Z, suspended processes, jobs, fg, bg, kill |
| Session Management with screen and tmux | Keeping work running after you disconnect; starting, detaching from and reattaching to sessions |
| Environment Variables and Shell Configuration | PATH, HOME, USER; .bashrc, .zshrc and .bash_profile; modifying your PATH |
| Process Monitoring | top, htop, btop, ps, and finding or terminating your own processes with pgrep and pkill |
Start at the Linux Basics section for all of the above in one page.
About this material
The linked pages are written and maintained by the US ATLAS Analysis Facilities documentation team, a separate organization from the SCDF. They are not SCDF documentation. Corrections and improvements to that material should be raised with them at usatlas/af-docs.
Note that the guidance there is written for the US ATLAS facilities. Where it differs from SCDF specifics — hostnames, batch configuration, storage paths — the SCDF documentation on this site takes precedence.