Minutes from 2 July 2008

2008/07/02 CSEP meeting

Attendees: Yan, John, Masha

  1. Software development:
    • Version 8.7.0 release has been installed on csep-op on Monday, June 30, 2008:
      1. SVN: copied latest release candidate to csep/tags/Version_8_7_0
      2. Checked out tagged version on csep-op, ran build and acceptance tests successfully

      3. Installed configuration and script files required by new Western Pacific testing regions
      4. John updated web-server code with tagged release version
      5. John rsync-ed existing production data products
      6. John tried to upgrade the memory, but discovered a problem with one of the new memory dimms. We still have 8Gb of memory as for Version 8.4.1 release which should be sufficient.
      7. Removed existing forecasts files that were based on master XML template (due to Trac ticket https://intensity.usc.edu/trac/csep/ticket/91)

      8. Updated crontab with new table that includes new testing regions
      9. Issued processing for RELM Mainshock and three-month forecasts groups for current processing date (2008/05/30), which resulted in new Matlab forecasts files based on XML master template. All succeeded.
      10. Nightly Dispatchers for California, NW Pacific and SW Pacific from July 1, 2008, exercised V8.7.0 release
    • Masha started re-processing of one-day and RELM Mainshock forecasts groups in parallel:
      1. Re-processing rate is about 24 days of one-day models per day and 1 day of RELM Mainshock models per 13 hours
      2. John asked if we need to bring up batch server up
      3. Need to reprocess:
        • 10 months of one-day models - should take about 14 days
        • 10 days of RELM Mainshock models - should take less than a week
        • 10 days of RELM Mainshock/Aftershock models - don't know yet (less than RELM Mainshock group - 1 week)
        • 11 days of three-month models - don't know yet (less than RELM Mainshock group - 1 week)
      4. If everything goes smoothly, should be done with all re-processing within one month. There is no real need for batch server at the moment, unless we want to re-process faster
    • Yan reviewed forecasts maps last week when login account to intensity was created
    • Code distribution:
      1. John created a directory on cseptesting.org to put the tarball files and other downloads
      2. Masha updated Trac pages with code distribution external links to cseptesting.org directories
    • Masha opened new Trac ticket https://intensity.usc.edu/trac/csep/ticket/93:

      1. Side effect is extra downloaded files from CMT web-site which are not used by CSEP - only consuming extra disk storage
  2. Masha is out for vacation July 9 - 18, 2008. Will be back on Monday, July 21, 2008.