[wildfly-dev] discussion about JSR352 Batch and JPA

[Scott Marlow]

What are the requirements for supporting Batch section 11.6 [1]?  From 
looking at JSR352, I get that each "chunk" has its own JTA transaction. 
  I previously had heard that we only supported starting the transaction 
from the application processing code (via UserTransaction) but I think 
the Batch container/runtime should start a JTA transaction for each 
"chunk" that is processed.  What are we really doing for managing the 
JTA transactions for batch?


REGULAR CHUNK PROCESSING & JPA


For the JPA support for regular chunk processing [1], the following will 
give a new underlying persistence context per chunk (jta tx):

@PersistenceContext(unitName = "chunkNonpartitionedAZTT4443334")
  EntityManager em;


PARTITIONED CHUNK PROCESSING & JPA


For the JPA support for partitioned chunk processing [2], the following 
will give a new underlying persistence context per chunk (jta tx):

@PersistenceContext(unitName = "chunkpartitionedAZTT4443334")
  EntityManager em;

One concern that I have about partitioning is the performance impact of 
deadlocking and waiting for the JTA transaction to time out.  Depending 
on whether the work is configured to retry or not, hitting several dead 
locks in a batch process could defeat some of the performance gains of 
partitioning.  Avoiding deadlocks by always reading/writing/locking the 
underlying database resources in the same exact order, would help avoid 
deadlocks.

Beyond the basic JPA capability of ensuring that each "chunk", has its 
own persistence context, how else can we help the batch processing 
experts writing JSR-352 applications on WildFly/Hibernate?

Anything else to discuss about JPA & Batch?

[1] Batch spec section section 11.6 Regular Chunk Processing 
https://gist.github.com/scottmarlow/6603746

[2] Batch spec section Batch 11.7 Partitioned Chunk Processing 
https://gist.github.com/scottmarlow/6607667

[3] persistence.xml https://gist.github.com/scottmarlow/6608533