• Actions

    • Alastair

    • Get schema updated on LCGIC01 to fix the connection pooling
      Will do it when the registry hardware is deployed
      • No new hardware yet - but it can be done quickly.
    • Track patch 1712 for GIN
      In pre-production. Some sites have updated - one has had a problem.
      • Now marked as in production. There is still a problem on the Greek PPS site.
    • System tests for the registry - including registry replication and perhaps interworking with the old registry
      The basic test works fine in isolation. Trying to get it to work within the framework - not finding common in the PYTHONPATH
      • DONE
    • Move registry replication to use multiple resultsets - and document in the spec
      • Code done but not documentation

    Antony

    • Add metadata columns to all  tables in consultation with users
      This is waiting for patch 1712 (GIN) and schema update on LCGIC01 to fix the connection pooling
    • Test migration
      • Can do migration on one machine - it all works. DONE
    • Write script to generate VDB rpm (which should be able to contain more than one VDB)
      Consider whether it should be a lot of small cron jobs or one big one.
      • It is almost done - have now built an RPM with ETICS

    Arun

    • Inverstigate case senitivity in getPlans and refreshPlans
      • DONE
    • Deal with Savannah Bugs
      • Only bug 33626 left

    Steve

    • Track bug 21578 for adding the UI to the L&B rgma feed.
      Zdenek Salvet has said they will try to fix it. Steve has contacted Zdenek who now say he is working on it.
      • Necessary changes have been committed to 3.1 LB branch, they will  be available in the next patch of glite-lb-server.
    • Make system tests work.
      Now all three work - on one machine. Will work on basic performance tests and move onto 2 machines!
      • Problem with producer buffer test - once buffer full exception is thrown it never recovers. Asked Arun to provide the # of  tuples in store for a table and make it available to the inspector
  • Current issues - round table
    • Steve - A memory based secondary producer will destroy itself when it detects conditions where a primary producer would return either RGMABufferFullException or RGMABusyExceptions.
    • Steve - will work on user doc
    • Alastair - schema replication testing
    • Alastair - also need to check that new registry forwards correclty to the old one
  • AOB