Details
-
Bug
-
Status: Closed
-
Major
-
Resolution: Fixed
-
2.0.0-M3
-
None
Description
I ran into a bug in AbstractQueryCache when using the timestamp QueryCache eviction policy. This is only a problem when running in a multithreaded environment.
The net of the problem is that in AbstractQueryCache.onTypesChanged(TypesChangedEvent ev) we call writeLock(), but when using the non-default eviction policy we never call writeUnlock(). As a result, any other threads requesting a writeLock on the cache will block indefinitely.
Attachments
Issue Links
- is part of
-
OPENJPA-738 QueryCache Improvement
- Closed