Computing and IT resources at TPI Jena FSU

Using your Jena FSU account/login you can access several resources (Note some of them do require further registration/identification):

• Webmail Jena FSU
• You should subscribe to the following mailing lists
  • TPI institute tpi_all@listserv.uni-jena.de
  • NR Jena groups numrel@listserv.uni-jena.de
  • astrogravity in Jena astrogravity@listserv.uni-jena.de
  • CoRe core@listserv.uni-jena.de
  • BAM developers bamdev@listserv.uni-jena.de
• Cloud service for calendar and share office documents
  • Nextcloud TPI
  • Nextcloud FSU
• Indico TPI server for events
• git TPI server for yours, BAM, CoRe, and various teaching repos
  
• Workstations at TPI for python, Mathematica, Maple, Matlab, etc.
• Computing clusters (see also here)
  • Abacus cluster with high-performance nodes at TPI, mostly for single-node jobs
  • ARA cluster for HPC jobs, managed by Universitätsrechenzentrum (URZ)
  • DRACO cluster for small HPC jobs
• Data storages:
  • login1.ara.uni-jena.de:/nfs04/data/gravtheo large permanent data
  • nfs2.tpi.uni-jena.de:/tpidata1/coredata/ medium-large permanent data
  • tullio.to.infn.it:/data*/numrel/ temporary data in use
• Any internal server and university resource can be access from outside using the FSU credentials and setting up a VPN connection. See instructions here.
• FSU Jena offers basic courses on various topics, check out the qualification portal

For any help you can contact the IT team at tpi_it@listserv.uni-jena.de

Current member of IT team for our group: alejandra.gonzalez@uni-jena.de and fabio.magistrelli@uni-jena.de

IT team webpage.

Computer resources managed by the group:

• tullio server and storage at INFN - Virgo Torino.
• CoRe virtual machine at TPI
• CoRe storage at TPI: /tpidata1/coredata/

Additionally, the group has acces to various HPC clusters in EU.

TPI Workstations

System

• c-serv: 12 cores, 256 GB RAM, XeonPhi 3120A
• c-serv2: 8 cores, 256 GB RAM
• m-serv: 12 cores, 384 GB RAM
• m-serv2: 8 cores, 256 GB RAM, XeonPhi 7120P
• m-serv3: ? cores, ??? GB RAM
• gpuserv: 8 cores, 256 GB RAM, 4 Nvidia-GPUs

Access

From TPI PCs ssh connect to each machine.

Use

Type

$ module avail

for installed software, use

$ module load <name>

to load a module.

Workstation can be also used remotely to run Mathematica, python etc kernels.

Abacus

Abacus is currently down (March 2023). Some nodes will be moved to Draco to form a TPI partition.

System

The system is composed of three partitions each with few Intel nodes equipped with high performances cores. Some pairs of nodes are connected with inifiniband. It is hosted and managed by TPI crew and shares the same filesystem of the TPI PCs. See this description.

Access

From TPI PCs ssh connect to head node abacus

$ ssh abacus

Use

Installed software can be listed with

$ module avail

and accessed loading the modules.

To access the nodes you need to use the slurm batch system.

ARA

System

HPC cluster with about 200 Broadwell and SkyLake nodes and Omnipath connection. GPUs are also available on some nodes, TPI members have exclusive access to an Nvidia GPU Ampere 80GB.

Support

Links to Jena URZ (University computing center):

• URZ
• Wiki and here
• Ticket system
• hpc-support@uni-jena.de

Access & login

Access to ARA cluster is currently only for users at Jena FSU. If you are at Jena, ask SB to oepn a ticket for you to get an account.

Info on the system can be found at ARAWiki (password protected link).

There're two head nodes, to login do (with your username)

$ ssh ho23kag@ara-login01.rz.uni-jena.de 

or

$ ssh ho23kag@ara-login02.rz.uni-jena.de

Proxy for git clone

To use git you need to setup a http proxy:

$ export http_proxy="http://internet4nzm.rz.uni-jena.de:3128"
$ export https_proxy="http://internet4nzm.rz.uni-jena.de:3128"
$ git clone https://bernuzzi@bitbucket.org/dradice/athena_z4c.git

Use

To list the installed software use

$ module avail

Modules can be loaded with

$ module load <name>

To access the nodes you need to use the slurm batch system.

DRACO

System

HPC cluster with tens of general-purpose Intel nodes. The cluster serves FSU and other institutions in Thurigen with cloud computing and some HPC access.

Support

Links to Jena URZ (University computing center):

• URZ
• Wiki and here
• Ticket system
• hpc-support@uni-jena.de

Access & login

Access to DRACO cluster is currently only for users at Jena FSU. If you are at Jena, ask SB to oepn a ticket for you to get an account.

To login do (with your username)

$ ssh ho23kag@@login1.draco.uni-jena.de

Use

To list the installed software use

$ module avail

Modules can be loaded with

$ module load <name>

To access the nodes you need to use the slurm batch system.