Skip to content

2018 04 24

Andre Merzky edited this page Apr 24, 2018 · 2 revisions

Agenda / Notes

- use cases -> requirements -> architecture -> implementation -> testing -> release
- Milestones: 
  - Use cases, Requirements (early April)
  - Feasibility, Prototype  (early June )
  - Implementation          (end   June )

  • updates:
    • Team 1: Ioannis, Will, Jumana
      • ticket
      • use case document
      • probed free storage on local nodes, -3 sigma == size of available free space
        • only probed default queues / nodes
        • 3 sigma: small (but non-zero) failure mode, waiting for data
        • this will be baseline in resource config file
        • still investigating live observation
      • node data structure (DS)
        • inverted DS is still being investigated
    • Team 2: Vivek, Srinivas, George
      • literature study
      • added LSF to the scheduler (currently old node DS, DS is still being discussed)
      • only affect non-mpi for now
      • test cases added
      • plan to expand to MPI units
        • discussion if disk reqs apply to every node or to total node
        • use cases: per node


Clone this wiki locally