Skip to content

Overview of the NAF pool setup

It is essential to understand some of the basic configurations in place in order to use the NAF efficiantly here are some bullet points about the basic design

  • NAF is a storage centric HTCondor pool

  • We use shared filesystems as a matter of data access, you should not need to transfer huge amounts of data through means like staging !

  • Classic access to the NAF is through remote submit nodes, also called WorkGroupServers

  • The setup on these WGS is identical to the setup of ALL NAF workers, this makes it very easy to check your code - if it runs locally on the WGS it should also run in the pool

  • We use hierarchical group quotas including features like 'fairshare' and 'surplus'

  • Standard/lite jobs (jobs with no explicit ressource requests) will always be the best and quickest way to flood the pool with your work

alt text