Minutes for FTB-MPI conference call - 2011 May 4th

• Argonne National Lab: Rinku
• Oak Ridge National Lab: Thomas, David, Hoony
• Indiana University: DongInn, Abhishek
• Ohio State University: Ouyang, Raghu, Xavier
• Lawrence Berkeley National Lab: Paul
• University of Tennessee, Knoxville: Absent

Items discussed

We discussed the following items

1. Policy manager: Rinku provided an update on the features of the policy manager. Feedback needs to be provided for the prototype that has been developed. David and Abhishek volunteered to look at it and provide feedback.
2. We discussed how to proceed with the resource and job scheduler development. We finally narrowed down on FTB-enabling slurm since most CIFTS folks with clusters have experience or usability of slurm. Rinku and Abhishek will independently look into Slurm and come up with a version that supports publishing of the events that the team has narrowed down on. Development will be with FTB 0.6 but we will start working on a sister standard for the generic resource managers/job schedulers as well as Slurm resource manager for FTB 1.0
3. There was some discussion on FTB-enabled IPMI software that OSU is developing. Plan is to find out how most supercomputing centers use IPMI and whether they integrate IPMI with other monitoring tools? Is this integration custom to each site or is it provided by vendors? And would it make more sense to focus effort in trying to get the monitoring tools (which have IPMI integrated) FTB-enabled?