[
https://issues.jboss.org/browse/RTGOV-329?page=com.atlassian.jira.plugin....
]
Jiri Pechanec commented on RTGOV-329:
-------------------------------------
The @Singleton bean in question uses container managed concurrency
From the EJB spec - By default, the value of the lock associated with
a method of a bean with container managed concurrency is a write lock (exclusive lock).
When synchronization primitives are used the @Singleton bean should use bean managed
concurrency using @ConcurrencyManagement(ConcurrencyManagementType.BEAN)
See 4.8.5.2 and 4.8.5.4 in EJB spec
Consider refactoring/cleanup of ActivtyServerLogger
---------------------------------------------------
Key: RTGOV-329
URL:
https://issues.jboss.org/browse/RTGOV-329
Project: RTGov (Run Time Governance)
Issue Type: Quality Risk
Security Level: Public(Everyone can see)
Components: Activity Collector
Affects Versions: 1.0.0.Final
Reporter: Jiri Pechanec
Assignee: Gary Brown
Fix For: 2.1.0.Final
Attachments: Performance.java
1) I consider the code to be overcomplicated - are we really getting any performance
boost with reusing the ArrayList of events? I found on the web that clear() might be even
slower than new allocation
2) There is unsynced access to fields _sequenceNumber and _failuresSinceLastSuccess
3) I am not sure if it is wise to mix synchronization on Java level - _timer in
BatchedActivityUnitLogger - and Java EE level - method synchornization on @Singleton
--
This message was sent by Atlassian JIRA
(v6.2.6#6264)