Difference between revisions of "Cloud Work at Imperial"

From GridPP Wiki
Jump to: navigation, search
 
Line 1: Line 1:
This page is kept for reference only. The work was summarized in a CHEP talk:
+
'''This page is kept for reference only.''' The work was summarized in a CHEP talk:
 
[https://indico.cern.ch/event/505613/contributions/2227436/ Stealth Cloud: How not to waste CPU during grid to cloud transitions]
 
[https://indico.cern.ch/event/505613/contributions/2227436/ Stealth Cloud: How not to waste CPU during grid to cloud transitions]
  

Latest revision as of 11:11, 23 August 2019

This page is kept for reference only. The work was summarized in a CHEP talk: Stealth Cloud: How not to waste CPU during grid to cloud transitions

The goal is to run an ARC-CE pretending to be a run-of-the-mill grid node while using it to submit jobs to the GridPP cloud. (We use ARC rather than CREAM as it has built in support for Condor.)

We aim to achieve this by using the (Condor) Submit module of a glideinWMS as the batch system and then channeling the jobs via the glideinWMS to the gridpp cloud.

Setting up a standalone glideinWMS

Building glideinWMS images

Setting up an ARC-CE

Setting up a glideinWMS to work with the ARC-CE (work in progress)