On 02/12/2014 22:40, Mark Kolich wrote: > On Tue, Dec 2, 2014 at 2:04 PM, Mark Thomas <ma...@apache.org> wrote: > >> On 02/12/2014 20:41, Mark Kolich wrote: >> >> <snip/> >> >>> If you have any thoughts on what I'm doing wrong, or if you can confirm >>> this is a legit bug, I'd be grateful for your input. >> >> There are known concurrency issues if you set an error status on a >> non-container thread. I'm literally in the middle of fixing them. >> > > Interesting. > > I think that explains it. > > In my test servlet, I changed response_.setStatus(401) to > response_.setStatus(200) and it has been running without issue now for > almost 30-minutes. > > >> >> I'll let you know when I am done and if you can retest with 9.0.x trunk >> at that point that would be great. >> >> >> > Sure, happy to.
Great. > Btw, do you have a pointer to the bug (er, bugs) tracking this issue? https://issues.apache.org/bugzilla/show_bug.cgi?id=57252 is what got me looking at this most recently. The tests (that I am in the process of extending) are here: http://svn.apache.org/viewvc/tomcat/trunk/test/org/apache/catalina/core/TestAsyncContextImpl.java?view=annotate This is part of the fix: http://svn.apache.org/r1643002 I'm currently working on the remainder of the fix which is adjusting the logic in the ErrorReportValve (generic error handling) and the StandardHostValve (application defined error handling) to avoid the concurrency issues. I'm close to a fix but I'm not quite there. I hope to commit the fix in the next 24 hours or so. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
