EventStore at MN site.

We plan to use EventStore MC injection directly at MN:

• reduce I/O at Cornell during injection step
• fast appearance of MC in EventStore
• minimal overhead during MC production

In order to fullfil this goal the following steps would be required:

• keep in sync Cornell/MN releases
  • other_sources contains all third-party packages required for EventStore injection
  • no additional software is required, since injection is done in python using SQLite and MySQL plugins.
• modify existing MN MC production farm scripts and add EventStore injection ones
  • use GroupEventStoreToolkit, it is not tighted to release and can be retrived directly from CVS. So far I recommend to use v05_00_08 version.

        Example:
        export ESTOOLKIT=< location of GroupEventStoreToolkit >
        $ESTOOLKIT/ESBuilder -add < dir/file/pattern > -sqlite sqlite.db
                             -grade mc-cont-unchecked 
    			 -output < location of EventStore output files >
    			 -esdb EventStore -newDB
    

  • Documentation can be found at EventStore Wiki (slightly out of date since no data version name would be required)
  • You need to provide input/output location and assign grade to your sample. "grade" can be formilized by software coordinator and/or we will agree on them
• Add a sample into separate ES database

Need to explain: "data version name" or "specific version name" (svName) has been used in EventStore to describe data. Currently we need to provide it. But recent changes to release make it obsolete. All new files will contain it in beginrun record (provenance information). That will allow to trace down any changes if data production. The full schema of mergin MN ESDB into Cornell master ESDB is shown here: .