MedeA JobServer & TaskServer - Indispensable for High Performance Computing

✓ Part of the standard MedeA Environment

At-a-Glance

The MedeA®[1] JobServer and TaskServer is the central unit to manage, monitor, and archive your calculations efficiently and consistently. Thanks to the MedeA JobServer and TaskServer, a highly flexible workflow maximizes productivity, even on systems with heterogeneous hardware, and different operating systems.

Key Benefits

  • Automated processing of compute protocols and workflows

  • Machine driven post-processing and results analysis of calculations with the following compute engines: VASP, LAMMPS, GIBBS, MOPAC, and Gaussian

  • Robust leveraging of compute resources from small-sized clusters to high performance computing systems

  • Reliable long-term archiving and accounting of computed data

MedeA has a unique three-tiered architecture: The MedeA GUI as the top tier, the JobServer as the middle tier, and the TaskServer as the bottom tier.

../_images/JSTS-config1.png

For smaller compute clusters, easily install a single central JobServer on a dedicated machine, and install TaskServers on the compute nodes. With this setup, submit calculations from one or more computers, which have the MedeA GUI installed. The MedeA JobServer and TaskServer have web interfaces to manage and monitor jobs, and restrict access if necessary.

../_images/JSTS-config2.png

Use the JobServer & TaskServer as a gateway for high performance computing in computer centers or the cloud. The JobServer & TaskServer support the most common load balancing systems such as SLURM, PBS/Torque, and GridEngine.

../_images/JSTS-config3.png

The MedeA JobServer and TaskServer automatically and reliably process your jobs and calculations while you are thinking about your next groundbreaking innovation in materials science.

Specifications

  • The MedeA JobServer & TaskServer require a 64-bit Intel or AMD architectures, and can be driven by various Windows or Linux operating systems. Oldest supported versions listed; newer versions also supported:

    • Windows: 10; Windows Server 2016

    • Linux: RHEL 8, CentOS/Rocky/Alma 8, Fedora 28, SUSE 15.4, Oracle 8.8, Debian 10, Ubuntu 20.04, Mint 20

  • Memory requirements: 1-4 GB RAM per compute core

  • Hard Drive space:

    • 25 GB minimum for a full MedeA installation

    • 60-180 GB for storing user generated data

  • Supported queuing (load balancing) systems are:

    • PBS/Torque

    • SLURM

    • GridEngine

    • LSF

    • Windows HPC

Key Features

  • Compute Management

    • Automated processing of compute protocols and workflows

    • Web interface designed for on-the-fly job monitoring: hold/resume, restart, etc.

    • Flexible administration of computer resources, even in heterogeneous systems

    • Developed for multi-user operation

  • Results & Data Management

    • Quick job search filters for users, status, title, and date

    • Efficient and consistent accounting of jobs and relevant files

    • Job index in an SQLite database while data is stored on disk

    • Full access to all input and output files for authorized users

    • Share results and workflows with a team

  • Security & Integrity

    • JobServer & TaskServer run as system services (daemons) for continuous access

    • Secure communication and access via HTTPS

    • User authentication restricts access and prohibits data vulnerability

  • Maintenance

    • JobServer updates and maintenance are possible without interrupting running calculations

    • Easy exchange of jobs and data between JobServers

Required Modules

  • MedeA Environment

Tightly Integrated Modules

  • MedeA Interface Builder

  • MedeA VASP

  • MedeA MOPAC

  • MedeA Gaussian GUI

  • MedeA GIBBS

  • MedeA Phonon

  • MedeA Transition State Search

  • MedeA Mechanical Properties (MT)

  • MedeA UNiversal CLuster Expansion (UNCLE)

  • MedeA HT-Launchpad

  • MedeA HT-Descriptors

  • MedeA Forcefield Optimizer

  • MedeA Diffusion

  • MedeA Thermal Conductivity

  • MedeA Viscosity

  • MedeA CED

  • MedeA Surface Tension

  • MedeA Electronics

Supported Modules

  • All MedeA modules

Find Out More

Learn more about submitting and monitoring computations in the MedeA Environment from our Webinar:

download:

pdf