On Nov 4, 2008, at 1:24 PM, Mattias Jiderhamn wrote:
I noticed this bug was marked as resolved, so I tried reloading our
webapp running on Resin compiled from SVN. Same problem.
Familiarizing myself a bit further with YourKit and comparing with
the sources, I believe both
com.caucho.jca.UserTransactionProxy._threadTransaction and
com.caucho.transaction.TransactionManagerImpl._threadTransaction
should be cleared properly.
Although I have not yet found the path to the HttpRequest. I could
send you my YourKit snapshot if that helps in tracking this down.
Did you build from resin/branches/3.1 or the trunk? The trunk changes
aren't checked in yet.
The fix is only for the SecurityContext clearing. The comet changes
had messed up the automatic clearing of that ThreadLocal.
-- Scott
/Mattias
Scott Ferguson wrote (2008-10-30 16:11):
On Oct 30, 2008, at 6:39 AM, Mattias Jiderhamn wrote:
Although embarrassed to admit it, we have had long standing problems
with PermGen memory leaks. We have gotten used to restarting the
server
every time we redeploy, to avoid OutOfMemoryError. Since we would
like
to make use of - or at least evaluate - some of the new Resin
features
like WAR versioning, I thought I’d give fixing this another shot.
I've added this as http://bugs.caucho.com/view.php?id=3031
That ThreadLocal needs to be cleared in a finally block when the
request completes. I'll need to check on that.
-- Scott
In com.caucho.server.dispatch.ServletInvocation there is a static
ThreadLocal holding references to the current (original) request
(while
dispatching?). This means that as long as the executing (and
dispatching) thread remains in Resins thread pool, but is not used
for
another dispatch, a handle to the original
com.caucho.server.http.HttpRequest is held. In the request, there
is a
handle to a com.caucho.server.dispatch.Invocation, which in turn
has a
handle to - the application classloader! (And there is your PermGen
leak)
Isn’t the fact that the ThreadLocal is never cleared an obvious
memory
leak? Or why haven’t this been fixed even though Scott seems to have
noticed this (see
http://bugs.caucho.com/bug_view_advanced_page.php?bug_id=2883)?
If the leak actually is in Resin, why isn’t anybody else seeing
this???
Are we doing something unusual? (The only thing I could think of was
ServletRequest implementations inside the application being
dispatched,
however I can't find any such implementations in use)
/Mattias Jiderhamn
_______________________________________________
resin-interest mailing list
[email protected]
http://maillist.caucho.com/mailman/listinfo/resin-interest
_______________________________________________
resin-interest mailing list
[email protected]
http://maillist.caucho.com/mailman/listinfo/resin-interest
_______________________________________________
resin-interest mailing list
[email protected]
http://maillist.caucho.com/mailman/listinfo/resin-interest
_______________________________________________
resin-interest mailing list
[email protected]
http://maillist.caucho.com/mailman/listinfo/resin-interest
