OAR's documentation follows the versions of OAR. One is advised to look at the documentation version corresponding to one's installed version of OAR.

Please find the documentation pages for OAR 2.5 at https://oar.imag.fr/docs/2.5

Please find the documentation pages for the latest version of OAR (development branch) at https://oar.imag.fr/docs/latest

If one is still running an old version of OAR (e.g. OAR 2.4), one can look at the old website.

See also the release notes of the OAR versions

See also the man pages of the OAR commands, readable from any terminal.

See also the search function of the website

This section aims at gathering additional contributions to the documentation. Please feel free to create your account on this website and request the wiki contributor access to the OAR team, if you would like to help with the content.

• Installation of OAR
  • Using OAR on Debian based systems
  • Using OAR on Redhat based systems
  • Building a virtual cluster using Docker
  • Testing OAR's installation using Vagrant
  • Example of setup with OAR energy saving mechanism
• Configuration
  • Some customization tips
  • Green computing
• Database management
  • Migration from a Mysql to a PostgreSQL database
  • How to handle a bigger and bigger OAR database
• Resources setup
  • About the job resources manager
  • Using Hyperthreading on nodes
  • Managing resources (CPU, GPU, ...)
  • Multi-cluster with priority
  • Coupling disk and compute resources
• Job admission rules
  • Custom admission rules
  • Some examples of admission rules
  • Developing admission rules
• Job types
  • More info on job types
• User interface and APIs
  • Use OAR Restful API with FastCGI and Ident on CentOS6
  • OAR DRMAA interface
• Administration tasks
  • Useful commands and administration tasks
• Job execution
  • Job event notifications
  • An equivalent to Slurm's srun for OAR
  • Passing environment variables to OpenMPI nodes
  • Using CRIU for the job checkpointing
  • Coupling oarsh with GNU Parallel
  • gvirt - make OAR jobs execute code in VMs
• Misc
  • Testing
  • Use cases and user tips
  • About the General Data Protection Regulation

This section lists documentation of platforms that use OAR as their resources and tasks manager and provides documentation pages for usage of OAR (to their users). Warning: These documentations have platform-specific information, which may not be relevant for other OAR installations (OAR is very customizable !).

• Igrida's tutorial: http://igrida.gforge.inria.fr/tutorial.html
• Grid'5000's documentation:
  • https://www.grid5000.fr/w/Advanced_OAR
  • https://www.grid5000.fr/w/Unmaintained:OAR2 (Grid'5000 account required, page is deprecated)
  • https://www.grid5000.fr/mediawiki/index.php/OAR2_use_cases (Grid'5000 account required, page is deprecated)
• ULHPC's documentation:
  • https://hpc.uni.lu/users/docs/oar.html
  • https://hpc.uni.lu/users/docs/scheduler.html
• Ciment's tutorial: https://ciment.ujf-grenoble.fr/wiki/index.php/OAR_tutorial

Please let us know (oar-users@inria.fr) if you want to contribute to OAR by having the documentation of your platform referenced here as well! It would be very welcome.

This section lists some known companion tools for OAR, please feel free to let us know if you wish us to add a tool you develop to that list.

• Kadeploy3 is an efficient and scalable and reliable cluster deployment tool that can be coupled with OAR (deploy job type)
• TakTuk is a powerful parallel remote execution tool, which can be used in OAR.
• Cigri is a lightweight grid middleware designed to run on top of a set of OAR clusters to manage more efficiently large sets of multi-parametric tasks (also called bag-of-tasks).
• GNU Parallel can run a batch of parallel tasks in an OAR job (more documentation here).
• Dask-jobqueue (more documentation here).

• Nypipe (more documentation here)
• Submitit (more documentation here)