Standup 2012-01-03
-----------------------------

Will plan on reviewing sprints tomorrow when Dave is back

Matt
~~~~
-- will focus on Metacat release
-- then come back to R client work after Metacat is done

Robert
~~~~~
-- Log Aggregation last of the functional pieces to finish
-- needs to work with Dexter to work out UTK firewall issues with the CNs
-- document the CN set up process, or verify the CN setup is accurate
-- discuss the DataONE certificate setup issues
-- make certain d1_common_java and d1_libclient_java hitting against the 1.0.0 tags
-- Document Node registration process.
-- merge logging into /var/log/dataone/<subcomponent>

Chris
~~~~
-- CN Replication; still problems with replication hanging when you scale up to larger numbers of objects being replicated; locking may be problematic in the HZ client at all
-- sending a request into the cluster, no response
-- possibly develop an isolated test case that replicates the problem outside of our code
    -- will help determine if this is hazelcast, or our code

Rob
~~~
-- focus will be working on setting up d1_integration tests for CN.
    - adding appropriate certificates
    - testing nodeUpdateCapabilities, and another 5 or 6 tests
-- will remove assertRelation from arch docs and libclient code (done)
-- finish up install of cn-dev-unm1.dataone.org.  waiting on certificate issue to be resolved.

Roger
~~~~~
-- extensive refactoring of GMN before vacation; will continue on that
    -- refactoring was for cleanup
    -- working on getting the tests working again

Skye
~~~~
-- working on the sync of index on cn-dev2, seems to be working
-- checking in the new cn-index component in buildout
-- removing mercury from the cn-buildout for now
-- Checking for references to old common_java in indexer components.
-- Further index component tasks.

Ben
~~~
-- Metacat release will be the priority
-- all tests passed except tier 4 (which are not valid)

Nick
~~~
-- Resuming investigating port 389 issues on cn-dev-3.dataone.org (Resolved)
-- playing with adapters on virtual hosts, looking for new tasks


Replication/Synchronization issue-

Proposal:
no locking/direct edits to the shared sysMeta map in d1_synch/d1_repl - only Metacat can do that.

Problems:
Exclusive locking needs to occur before a get is performed to guarantee consistency

Maybe the serialversion on additional CN api method calls would allay data consistency issues
 
Need to investigate why the get in Synchronization causes an event entry to be triggered in Replication.  May have added in logic to metacat to create, as a placeholder, an entry in the sysmeta map when the get is performed(???) 


CN Replication api questions:
Replication.setReplicationStatus - how can VersionMismatch be thrown? (fr. Chris, it can't.  Is it ok to remove the exception from the API?)
Replication.updateReplicateMetadata - description is confusing in arch docs - leads one to believe all replica entries would be replaced with the new singular replica.

Register.updateNodeCapabilities allows update to all of the Node document, except identifier (and maybe NodeType?).  Should change the description and maybe method name to be simply updateNode - 
        - keep the existing method name (see also mn.getCapabilities())