I just read an interesting article[1], which highlighted something that never really occurred to me.
And I think we are guilty of not clearing up the ThreadLocals upon application passivation, and hence leaking massively. This mail is from my mobile, primarily as placeholder for myself, to create a Jira issue about this. [1] http://java.jiderhamn.se/2012/01/29/classloader-leaks-iv-threadlocal-dangers-and-why-threadglobal-may-have-been-a-more-appropriate-name/ Cheers
