Due to the large computational requirements for LOFAR datasets, processing bulk data on the grid is required. This manual will detail the Dutch grid infrastructure, the submission process and the types of users anticipated to use the LOFAR reduction tools.

Overview

SurfSARA is the Dutch locations of the CERN Computational Grid and its facilities are available for general scientific computing. Because the LOFAR telescope requires significant computational resources, the reduction pipelines have been fitted to run on the Dutch Grid nodes with minimal user interaction. The GRID_LRT software package automates LOFAR data staging, job description, Pre-Factor parallelization, job submission and management of intermediate data.

Requirements:

• User account to the lofar ui at grid.surfsara.nl
• Login to the PiCaS client at picas-lofar.grid.sara.nl
• Active Grid certificate for launching jobs/accessing storage
• Membership of the LOFAR VO.
• Astron LTA credentials for staging LOFAR data

Installing:

The up to date installation instructions are here.

Attribution

If you actively use GRID_LRT, please cite this software as such below:

@misc{apmechev:2018,
      author       = {Alexandar P. Mechev} 
      title        = {apmechev/GRID_LRT: v0.5.0},
      month        = sep,
      year         = 2018,
      doi          = {10.5281/zenodo.1438833},
      url          = {https://doi.org/10.5281/zenodo.1438833}
    }

If you're using GRID processed data, also consider citing the paper below, outlining the procedure of running LOFAR data through a High Throughput Cluster:

@INPROCEEDINGS{mechev2017,
   author = { {Mechev}, A. and {Oonk}, J.~B.~R. and {Danezi}, A. and {Shimwell}, T.~W. and                             
{Schrijvers}, C. and {Intema}, H. and {Plaat}, A. and {Rottgering}, H.~J.~A.},
    title = "{An {A}utomated {S}calable {F}ramework for {D}istributing {R}adio {A}stronomy {P}rocessing {A}cross {C}lusters and {C}louds}",
booktitle = {Proceedings of the International Symposium on Grids and Clouds (ISGC) 2017, held 5-10 March, 2017 at Academia Sinica, Taipei, Taiwan (ISGC2017). Online at \url{https://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=293}, id.2},
     year = 2017,
archivePrefix = "arXiv",
   eprint = {1712.00312},
 primaryClass = "astro-ph.IM",
    month = mar,
      eid = {2},
      doi = {10.22323/1.293.0002},
    pages = {2},
   adsurl = {\url{http://adsabs.harvard.edu/abs/2017isgc.confE...2M}},
  adsnote = {Provided by the SAO/NASA Astrophysics Data System}
}

Tutorial Notebook

Best way to get acquainted with the software is with the tutorial notebook available at GRID_LRT/tutorials/LRT_demo.ipynb

Setting up Jupyter on loui

$> ssh loui.grid.sara.nl
[10:42 me@loui ~] > mkdir ~/.jupyter
[10:42 me@loui ~] > export PATH=/cvmfs/softdrive.nl/anatolid/anaconda-2-2.4.0/bin:$PATH
[10:42 me@loui ~] > export LD_LIBRARY_PATH=/cvmfs/softdrive.nl/anatolid/anaconda-2-2.4.0/lib:$LD_LIBRARY_PATH
[10:42 me@loui ~] > jupyter notebook password

Running a Jupyter notebook on loui

Assuming you have ssh login to loui, you can run this notebook on your own machine by using ssh port forwarding :

$> ssh -L 8888:localhost:8888 loui.grid.sara.nl
[10:42 me@loui ~] > source /home/apmechev/.init_jupyter

With that shell running, you can open the browser on your local machine and go to localhost:8888, and browse to the tutorials folder.

Grid job submission and queuing

Data Staging

In order to stage the data using the ASTRON LTA api, you need credentials to the ASTRON LTA service. These credentials need to be saved in a file on the lofar ui at ~/.stagingrc in the form

user=uname
password=pswd