Fermilab Computing Division

CS Document 5684-v1

User View of Fermilab Batch Computing

Document #:
CS-doc-5684-v1
Document type:
Technical Note
Submitted by:
Michael H Kirby
Updated by:
Michael H Kirby
Document Created:
11 Feb 2016, 12:28
Contents Revised:
11 Feb 2016, 12:28
Metadata Revised:
11 Feb 2016, 12:28
Viewable by:
  • Public document
Modifiable by:

Quick Links:
Latest Version

Abstract:
Fermilab Batch Computing is implemented in several different workflows. CMS workflows, FIFE workflows, and direct submission . The simplest of these models is the direct submission wherein an experiment such as DES submits jobs to compute elements by issuing HTCondor commands directly. This method consists of user commands connecting directly with a single compute element and is therefore limited to that single cluster (in this case the Fermi
GP Grid cluster). It requires that experimenters learn HTCondor commands and actively avoid potential pitfalls. HTCondor establishes the order in which jobs will start and matches job requirements (memory, CPUs, disk) to available computing resources. Within this implementation there is no ability to dynamically go beyond the confines of the dedicated compute cluster.
DocDB Home ]  [ Search ] [ Authors ] [ Events ] [ Topics ]

DocDB Version 8.8.9, contact Document Database Administrators