User Guide

This documentation describes the obsolete "legacy nef" configuration which will definitively stop on 17 april 2016 : please use the new nef configuration

What will become my "legacy nef" data when the cluster stops ?

• /home and /epi/<teamname>/<username>  : still exist on the "new nef" under the same path
• /dfs/workspace/<username> will be copied to temporary storage on "new nef" to /data/<teamname>/user/<username>/TEMPO-OLD-DFS
• /dfs/<teamname> will be copied to temporary storage on "new nef" to /data/<teamname>/share/TEMPO-OLD-DFS
• other data (node local files, etc.) will be deleted

Front-end

The cluster is based on several front-end servers and a lot of compute nodes.

3 servers are available:

• nef-frontal.inria.fr : main front-end, ssh and submission front-end
• nef-devel.inria.fr : compilation, ssh and submission front-end
• A storage server (no direct access for users)

Currently, the only way to use the cluster is to connect to one of the front-end with ssh. Then you can have access to the computing resources available by using the Torque job manager.

The nef cluster is not part of the internal (production) network of Inria; therefore Inria (iLDAP) accounts/passwords are not used.

First steps

1. First you need to apply for an account on the nef cluster. You must give your ssh public key to have an account.
2. Then connect to one of the main front-end nef.inria.fr or nef-devel.inria.fr using ssh
3. Then you can import your data / sources files using scp on the nef front-end and compile your programs on nef-devel.inria.fr
4. Use the job manager to start your tasks
5. You can view the current running jobs using the Monika web interface. You can also view the system activity on nodes using ganglia

 Disk space management

Each user has a dedicated home directory on the storage server. The total available disk space for users is 15TB. All nodes have access to this storage using NFS.

Data stored on the cluster IS NOT backed up !

A quota system is activated on the shared storage server:

• The soft limit is 150GB
• The hard limit 600GB
• The delay is 4 weeks

You can use 150GB of data without restrictions; as soon as the soft limit is reached, you have 4 weeks in which to delete files and go back under the soft limit. You can never use more than the hard limit.

A warning message will be sent by mail every Sunday when a limit is reached.

You can check you current disk occupation with the quota -s command

More storage is available for the ABS, ASCLEPIOS, MORPHEME, NEUROMATHCOMP and TROPICS team members in /epi/<teamname>/<username>. Teams needing more storage should contact the cluster administrators.

A 19TB experimental distributed scratch space is available under /dfs for all users. Its first target is short term storage, quotas or time limits may be applied in the future if it becomes permanently saturated. It has low performance for activities intensive on metadata (eg : compilation, reading/writing lots of small files).

There is also temporary disk space available on each node, in the /tmp directory

• 1.1TB on Dell R815 nodes
• 100 GB on Dell PE1950 nodes
• 420GB on HP nodes
• 110GB on Carri nodes

 Softwares

All nodes are installed using a Linux Fedora 16 64bit distribution

Main softwares available on the cluster:

• 3.4.7 linux x86_64 kernel
• PGI 13.5 compilers
• GCC 4.6.3 (C,C++, Fortran77, Fortran95 compilers)
• OpenMPI 1.6.3
• Paraview 3.14.1
• MvaMpich2 1.9a2
• Java 1.6.0_24 + java3D 1.5.2
• Petsc 3.4.2
• Matlab 2015a (/opt/matlab2015a/bin/matlab), Matlab 2011b (/usr/local/matlab2011b/bin/matlab)
• CUDA 5.0 (/usr/local/cuda) + CUDA SDK (/usr/local/cuda/samples)
• DDT debugger (/opt/allinea/ddt, see Documentation)

Other tools/libraries available:

• blas, atlas
• blacs and scalapack (in /usr/local/lib64)
• openblas 0.2.8 (in /opt/openblas/)
• gmsh 2.6.1 (in /opt/gmsh)
• Python (including scipy, numpy, pycuda)
• Trilinos (/opt/Trilinos)
• mesa 7.7.1(/opt/mesa)
• Erlang
• GDB & DDD
• Valgrind
• CGAL
• GSL & GLPK
• boost 1.47