Particle Physics Linux Systems Background

The general purpose Linux login servers are pplxint10 and 11. Each gives access to a condor based batch system running several worker nodes. They run the CentOS 7. All machines remain switched on except for during minimal maintenance periods. The Grid and CERN have made CentOS7 the default now so you should try to run your code on these systems.

In case of problems software or hardware the system may have to be taken down at short notice. Any problems should be directed to

Access to all Unix systems (except pplxintn) from outside of the domain is disabled. When you are away and wish to connect, login to pplxintn, then on to the others if needed. You can only use ssh. This avoids hackers "sniffing" your passwords.

Long jobs (>30 mins) should not be run on pplxint machines directly, but should be submitted to the Condor Batch System. If it is absolutely necessary, long, CPU intensive, single threaded jobs may be run "niced", eg,

nice ./

Jobs that are left running using large amounts of CPU on the interactive machines may have the priority lowered or be killed without notice.

Please do not run jobs which use high network bandwidth for long periods or that heavily hit the disk system. This helps keep interactive response acceptable. It is unlikely that groups other than SNO, T2K, Atlas and LHCb need to worry about internal network or disk access impacting interactive performance.

We keep a pool of third-party software on the systems. You can obtain a list of (recently) installed 3rd party software using the command "module avail". The software environment must be "loaded" before the software is "executed". To load the environment use "module load". The software can then be executed, e.g. to load and execute Matlab version 7.10:

module load MATLAB/7.10 matlab

You may ask for additional software to be installed by contacting us at the address above. This software will also be made available on the University cluster, ARC. All Members of the university may access this supercomputer which is running a very similar system to us.

(Some older machines remain running Scientific Linux 6 with access to a torque based batch system. The majority of the workernodes here will be migrated to CentOS7 as requirements switch)

Categories: Linux | PP | Particle | Unix | ppunix