There is a workaround in place in the implementation that does a get and then a put.
I can't force the underlying cache to lock on the key when I do the get, so another thread could potentially put something in before I replaced the value in the workaround case (in other words, it's best if the underlying method gets fixed in ehcache)
Reported here:
https://jira.terracotta.org/jira/browse/EHC-894
There is a workaround in place in the implementation that does a get and then a put.
I can't force the underlying cache to lock on the key when I do the get, so another thread could potentially put something in before I replaced the value in the workaround case (in other words, it's best if the underlying method gets fixed in ehcache)