Atlas-connect-l mailing list

[Rob Gardner <rwg AT hep.uchicago.edu>]

All,

We’ll have an ATLAS Connect meeting tomorrow:

Monday, 11:30 Central

Vidyo: https://indico.cern.ch/event/306944/

A number of topics we should discuss:

1)  Testing and work on the Faxbox object store (Lincoln and David)

* Unix access, xrootd access, globus access, http browser, http access (curl, wget)

* Building proxy clusters for both xrootd and http

2) Work on integrating Stampede, APF, the CONNECT panda queue  (Dave, Lincoln)

3) Proof on Demand (Lincoln)

4) SLAC meeting, https://indico.cern.ch/event/303799/.   Agenda is tight, its only one day.  What can/should we contribute to this meeting.   ATLAS Connect “meet-up” outside the meeting?

5) We need to document more usability examples.  Here is our current set of docs:  https://ci-connect.atlassian.net/wiki/display/AC/ATLAS+Connect 

ATLAS Connect User Examples

• The FAX for end-users tutorial (CERN Twiki)
• Run a simple ROOT job
• PROOF on Demand

ATLAS Connect Cluster Usage Examples

• Submitting a job from a local cluster into the RCCF  

ATLAS Connect Storage and I/O 

• FAXbox usage guide
• FAXbox storage policy

We need to build this up, based on real Tier3 use cases.   One approach is to build documentation on three fronts:

I.  ATLAS Connect System Fundamentals

Basics of simple job submission: 

    Modes

- from login.usatlas.org  (login and work directly)

- from a Tier3 (submit to usatlas.org from home, using it as a launch pad) 

- from a Tier3 cluster (submit to the local Tier3, overflow to usatlas.org)

Basics the Faxbox object store, moving files with dq2, xrd, globus, http (curl, get)

II.  Create ATLAS Connect Computing Recipes

Based on use cases profiled by the Tier 3 Implementation Committee

Setups with code, data sets, and task submission scripts, post-task output management, analysis scripts  

These should be:

- self-contained, cradle-to-grave

- regularly tested

III.  Supporting the ATLAS Offline Computing Tutorial on ATLAS Connect

The latest ATLAS Offline Software Tutorial is here:  https://indico.cern.ch/event/295572/

The idea here of course is not to duplicate, but support by 

a) Making sure they work on AC

b) Take a couple of examples (making and analyzing D3PDs) and compute on the AC platform, perhaps pre-placing datasets on faxbox:

Hands-on: Stand-alone D3PD Analysis 3h15’ 

You will learn how to use the available tools to ease your D3PD-based analysis. You will learn how to read D3PDs in a super-efficient and very convenient way. You will produce your histograms and calculate a cross section.

https://twiki.cern.ch/twiki/bin/viewauth/AtlasComputing/SoftwareTutorialAnalyzingD3PDsInROOT

Hands-on: Making your own D3PDs 3h0'

https://twiki.cern.ch/twiki/bin/viewauth/AtlasComputing/SoftwareTutorialD3PDMaking

c) Keeping the material up to date (after the next tutorial as the materials will change with new tools and the xAOD format)

---

Rob Gardner • Twitter: @rwg • Skype: rwg773 • g+: rob.rwg • +1 312-804-0859 • University of Chicago