search

vipx / google-guice

Automatically exported from code.google.com/p/google-guice
Apache License 2.0
0 stars 0 forks source link

FinalizableReferenceQueue leaks, preventing reloads in a Servlet container #227

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago 
Paolo Capriotti:
 "The FinalizableReferenceQueue thread prevents my application  from being unloaded. The thread 
is started from a stack-local instance, so I think there's no way I can stop 
it. "

http://groups.google.com/group/google-
guice/browse_thread/thread/1d022bfc3505f7fa/5cdd912e7b8b1095?
show_docid=5cdd912e7b8b1095

Original issue reported on code.google.com by limpbizkit on 4 Aug 2008 at 7:43

GoogleCodeExporter commented 9 years ago 
Related discussion thread:
http://groups.google.com/group/google-guice/browse_thread/thread/bac22d152fe2d25
7/531dcc0ffd08e11d#531dcc0ffd08e11d

Original comment by gili.tza...@gmail.com on 17 Nov 2008 at 10:32

GoogleCodeExporter commented 9 years ago 
This has been fixed in Google Collections. We just need to bring over the new 
version.

Original comment by crazybob...@gmail.com on 17 Nov 2008 at 11:01

GoogleCodeExporter commented 9 years ago 
Excellent. I'm eagerly awaiting this update! I've been suffering from major 
webapp
deployment problems related to memory leaks over the past five (!!) years. 
Being able
to develop webapps without having to restart the web server every five minutes 
would
be a dream come true :) Thank you!

Original comment by gili.tza...@gmail.com on 18 Nov 2008 at 5:21

GoogleCodeExporter commented 9 years ago 
I brought in the new version of Google Collections in r708.

Original comment by limpbizkit on 29 Nov 2008 at 10:22

GoogleCodeExporter commented 9 years ago 
Jesse, I don't think this is enough. You should probably keep this issue open 
until
http://code.google.com/p/google-collections/issues/detail?id=92#c6 is fixed.

Original comment by gili.tza...@gmail.com on 29 Nov 2008 at 10:34

GoogleCodeExporter commented 9 years ago 
Stuart's patch exploits the fact that he can reflectively change some private 
fields - this is very fragile stuff. 
Creating a new thread is a reasonable thing to do, and we shouldn't have to 
write hacks to make that safe. 
Especially since the hacks break portability between JVMs (and possibly between 
app servers).

So what we really need a safe, hack-free way to use background threads that 
works in both Java SE and EE.

Gili, could you talk to the Glassfish folks and see if they can address this on 
their end?

Original comment by limpbizkit on 29 Nov 2008 at 11:58

GoogleCodeExporter commented 9 years ago 
Jesse,

I filed a new bug report: 
https://glassfish.dev.java.net/issues/show_bug.cgi?id=6857

Original comment by gili.tza...@gmail.com on 30 Nov 2008 at 2:22

GoogleCodeExporter commented 9 years ago 
Yeah, those reflection hacks are totally flaky and I didn't expect them to go 
into
google-collections, but using doPrivileged when creating the Finalizer thread 
and
pro-actively setting the TCCL to null are IMHO a good idea.

The dangling inherited-thread-local can be fixed in the GlassFish worker pool, 
which
just leaves the access-control-context issue in the classloader. Again this 
might be
possible to fix in GlassFish, but I'm not sure as it involves Java security 
which is
a headache to follow at the best of times :)

Original comment by mccu...@gmail.com on 30 Nov 2008 at 10:58

GoogleCodeExporter commented 9 years ago 
Also be aware that Guice has its own copy of some of the Google-Collections 
code,
including the Finalizer thread - I coded up a patch to fix this when I was 
testing
the other issue:

http://peaberry.googlecode.com/svn/branches/spike/lib/build/google-guice-patch-2
27.txt

I still had to keep copies of a few classes from Google-Collections (such as the
ReferenceMap) to fix some visibility issues but I sync'd them with the version 
in
Google-Collections. Also the tests for ReferenceMap have an unfortunate 
dependency on
the relocated package name (because they aren't JarJar'd) but it's a start...

Original comment by mccu...@gmail.com on 30 Nov 2008 at 12:46

GoogleCodeExporter commented 9 years ago 
How can this issue be fixed if
http://code.google.com/p/google-collections/issues/detail?id=92 is still open?

Original comment by gili.tza...@gmail.com on 22 Oct 2009 at 5:03

GoogleCodeExporter commented 9 years ago 
There seems to be another issue report for it:
http://code.google.com/p/google-guice/issues/detail?id=288.

Original comment by syva...@yahoo.com on 26 Oct 2009 at 2:10

GoogleCodeExporter commented 9 years ago 
But you probably knew that already as you seem to have originally reported it :)

Original comment by syva...@yahoo.com on 26 Oct 2009 at 2:13