lofar-grid-hpccloud

This repository hold resources for deploying the LOFAR software (genericpipeline) and related tools through Singularity containers. These containers are general, but at the same time somewhat tailored for SKSP use.

The master branch is empty. Currently the images on this branch (fedora-py3) are based on the Fedora 34 Linux distribution, which is available from DockerHub.

As this branch no longer includes Python 2, the genericpipeline framework is not included in these recipes anymore (see the fedora branch for that). Pipelines like prefactor (now LINC) are or have moved to CWL.

Singularity

To build a full LOFAR Singularity image, do the following:

1. Turn on MKL and/or CUDA in singularity/Singularity, if desired, by setting HAS_MKL=true and/or HAS_CUDA=true. Set them to false if you do not require those.

2. Optimise your container for a desired architecture by updating the MARCH and MTUNE variables to the appropriate values. If you want to build for a generic machine, set these to MARCH='x86-64' and MTUNE='generic', respectively.

3. Build singulariy/Singularity by running

    sudo SINGULARITY_CACHEDIR=$PWD SINGULARITY_TMPDIR=$PWD singularity build lofar_sksp.sif Singularity
   

Pre-built containers are public hosted at SURFSara. Sort by date to find the latest container there.