#persist

Notes on LTER MN Transfer to New Software Stack and New Content Approach
========================================================================


Present: Chris, Dave, Ben, Mark, Jing, Robert, Roger


Summary
-------
 * All older content at the LTER MN will be set to archived
 * The content has been replicated to the KNB MN
 * New content in the GMN-PASTA instance will be synchronized
 * Content in PASTA will have non-related identifiers (non-related to previous Metacat objects)


Issues to resolve
-----------------
 * Which node will be authoritative?
   * ~49K objects replicated to KNB, ~700 failed
   * Need to set LTER replica status to INVALIDATED for all objects
   * DECISION: LTER will remain the authoritative node
 * Will there be an obsolescence chain?
   * Issue: DataONE only supports linear obsolescence, but there's not a one-to-one correspondence from a given old science metadata object and a new science metadata object
   * Building obsolescense after the fact may be feasible
   * DECISION: Initially, there won't be a chain, but may be revisited
 * How to set objects from Metacat MN to archive?
   * Use Metacat as soruce of object PIDs
   * DECISION: Use BASH script or similar to call CN to set as archive

1. Turn off sync
What to do if you need to change the baseURL and reharvest all content.
1. Update baseURL and disable Sync on node via MemberNode call
    a. Set new baseUrl in node
    b. Set synchronization off in node
    c. call updateNodeCapabilities on the CN from the MN

2. Confirm baseURL has changed (administrative procedure on CN)
    a. Notify CN Admin
    b. confirm that d1_processing daemon's synchronization reflects the change (baseUrl and synchronization boolean)
        1) check the synchronization log file. 
        2) Hazelcast notifications will cause Synch Schedule Manager to reschedule node's harvest (deleting it from quartz scheduler)
    c. CN Admin will update last_harvest date on the node

3. Turn on sync via MemberNode call
    a. set syncrhonization on in node
    b. call updateNodeCapabilities
4. Check Synchronization process
    a. Notify CN Admin
    b. confirm that d1_processing daemon's synchronization has picked up the change
        1) check the synchronization log file. 
        2) Hazelcast notifications will cause Synch Schedule Manager to reschedule node's harvest (adding it to quartz scheduler)