Atlas-connect-l mailing list

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

From today’s meeting, please feel free to correct, add.

The issues with Stampede have multiplied and involve limitations with libcvmfs (the API for Parrot).  

As a simplification measure, we will pursue the option of using a local virtual machine to host a Chirp server that exports the needed /cvmfs mount points.  The same host will be used host a local squid cache. 

And the compatibility libraries will be provided with a local tarball, since using the cernvm3 image evidently brought more overhead.  

We also discussed experimenting with docker, creating a container that mounts cvmfs.  Sierra, part of Future Grid at UCSD, has a docker testbed running that can be used.

In terms of testing, Peter is surprised Sherpa would cause hangs because it is much simpler than Reco_trf.py, which does not.  Agreed not to worry about Release 16.   First thing to do is try collection of HC jobs again, see what works.  If we want to target specific tasks, then perhaps Jedi could be used (but this will take some learning on our part).

On May 9, 2014, at 6:10 PM, Rob Gardner <rwg AT hep.uchicago.edu> wrote:

Colleagues,

We’ll have a meeting on Monday, 11:30 am Central.  

Vidyo connection https://indico.cern.ch/event/319106/. 

---

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

---

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