IPP Progress Report for the week 2009.11.23 - 2009.11.27

(Up to IPP Progress Reports)

Eugene Magnier

Heather Flewelling

• processed md03, md10, md07
  • md10 and md03 are distributed
  • md07 needs to be debugged and stacked
• investigated/reverted md09 processing faults from the weekend
• built ipp on c18, ran simtest on c18-19.
• worked on mask statistics (waiting for scripts to run)
• vacation wed/thur/fri!

Bill Giebink

• Worked with Tom and Steph to make accelerometer data available outside of PS1 via VNC.
• Set up and tested Mark Chun's DIMM with Jeff.
• Looked into configuration changes that are causing psLib TAP tests to fail.
• 1/2 day on Wednesday and Friday.

Roy Henderson

Paul Price

Bill Sweeney

Chris Waters

• Nebulous: Implemented a number of bug fixes and modifications to ensure no loss of data.
  • Each volume can now be associated into a cabinet, defining their physical arrangement.
  • File creation: new files without a specified volume are created on a random volume from a subset of the N least full.
  • Replication: file replication uses the same volume selection method, with the constraints that a copy not be automatically placed on the same volume as another, and that the first copy not be housed within the same volume.
  • Cull: instead of removing the first available instance, first attempt to remove an instance on the same volume as another, then attempt to remove an instance in the same cabinet as another, and then remove the first only if those fail.
  • Corrected bug that prevented the deletion of an object if it had more than one unavailable instance.
  • Fixed nebdiskd to synchronize between volume->mountedvol tables and to properly remove volumes that were no longer accessible.