Something that might be worth cleaning up about this code is that it later does 
some checks and decides that no sync is needed. For example it does this one:
dr.getDataPolicy().withPersistence() && persistentId == null
and then in synchronizeForLostMember() it does nothing if we don't have 
concurrency checks enabled.
It seems like all these checks could be done synchronously before hand this off 
to the scheduler.
It also seems like getRegionForDeltaGII could be renamed (i.e. 
getRegionForLostMemberSynchronization) or turned into another precondition 
(that our region is a DistributedRegion). Better would be just to call a method 
on InternalRegion that does nothing on LocalRegion and is only implemented on 
DIstributedRegion instead of doing this instanceof check.


[ Full content available at: https://github.com/apache/geode/pull/2340 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to