• Bug
  • Status: Closed
  • 1 Critical
  • Resolution: Fixed
  • ehcache-core
  • cdennis
  • Reporter: amiller
  • October 28, 2009
  • 0
  • Watchers: 0
  • July 27, 2012
  • April 19, 2010

Description

Currently, the Cache maintains both a MemoryStore and (optionally) a DiskStore. There is no synchronization used when acting on both stores in concert. That is mostly probably ok, but could create some weird race conditions against user threads and/or the disk spool thread. Should consider these races more carefully and possibly document why the race is not an issue as appropriate. It’s possible that some races *are* important a rw lock needs to be introduced for operations that cover both in some cases.

Since these races have likely existed for a long time, it’s probably unlikely that there are major critical issues here, so this can be a longer-range thing.

Comments

Fiona OShea 2010-03-10

This is going to happen as part of RMP-XXJ JTA for Ehcache standalone See RMP-526 - Main Esperance RMP