Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Table of Contents

...

May 2017

...

May 12th, 2016

  • auto scaling of stage out and stage in jobs
    • 4.8 transfer refiner will be Cluster by default.
    • auto-computation of number of stage in, stage out and cleanup jobs
      • defaults should be computed based on number of jobs at a level.
      • use a ratio or step function . 
      • come up ratio ranges for auto determination
        • 1:5 for numbers of jobs < 10K ( 20%)
        • 1:20 for number of jobs > 20k ( 5%)
      • will create a JIRA item for this

  • container stuff
    • close to having one example running
    • have not figured clustering jobs out yet.
    • mats agrees with the approach now. pegasus lite invokes the docker run commands.

  • integrity stuff
    • will make slides
    • be specific about we have done . 
    • we give them an option of running synthetic stuff
    • For 
    • also define best effort part. 
      • strict, off, minimal , best effort
    • how do we handle case where SHA exists.

March 2017

March 17th, 2016

  • monitord stdout and stderr missing 
  • the VARS one. just expose the variable. 
  • SCEC issue
    • job managers per resource
    • got fixed by one job manager per job
    • BOSCO works partly. 
  • containers call from yesterday
    • dsa
  • metadata 
    • metadata population in postscripts
    • move metadata population to the postscripts.

...