MinutesReleaseDeployJul8

Introduction

• Meeting of the sub-committee to address release installations in the US cloud.
• Coordinates:
  • Tuesday, July 8, 9am Central/5pm CERN
  • Phone: (605) 475-6000, Access code: 735188; Dial *6 to mute/un-mute.
• Attending: Rob, Saul, Xin,

Background

We need to resolve how ATLAS distributes its releases and close this issue asap. I am not certain whose ball this was, but since I have an interest in seeing it resolved within the US Facility I'm writing this email.

Saul has a nice proposal that I think makes sense - factorizing ATLAS pacball environment as the main tool for managing releases from the specific delivery mechanism (via web-lcg job, via DQ2-Panda etc)

On the other hand Alessandro has built of very useful infrastructure (eg the web portal) for tracking releases and validations across the grid.

We need to find a solution which is the best for both ATLAS and the two major Facility infrastructures. Can we use components from both? What really is the requirement?

From Alessandro:

• Release installation pages
• pacball

Pacball + DQ2 delivery (Xin)

Action items

• We need to do this without creating undue confusion in the collaboration
• Get an official set of pacballs releaesed
  • Need to decide where to put these. Files are > 2GB in size, big web server
  • Stan and Alessandro may have already agreed to this.
• Putting them into DQ2
  • We think this has been resolved - a naming convention, a new DQ2 object
  • Atlas releases are now data like any other data
  • Cryptographically reproducible
• Now the Panda team can start installations using special Panda jobs
• Putting into the Panda monitor
• None of this is inconsistent or at odds w/ Alessandro's new Firefox plugin, or the LCG install mechanisms.

AOB

• None

-- RobertGardner - 07 Jul 2008