I believe the two most common culprits are the JDBC driver manager
and Commons Logging.
--- Pat
> -----Original Message-----
> From: Sergei Dubov [mailto:[EMAIL PROTECTED]
> Sent: Tuesday, November 29, 2005 12:18 PM
> To: Tapestry users
> Subject: Re: java.lang.OutOfMemoryError: PermGen space
>
> Just out curiosity, Pat, what kind of particular "JVM level singletons"
> hold onto classloaders?
>
> Thanks,
>
> Serge
>
> Patrick Casey wrote:
> > There's the 'yah but' though, which is that, in a web app, JVM
> > singletons can hang onto the classloader which, in turn, keeps
> references to
> > all the classes it has loaded. Hence even when the GC makes a sweep
> through
> > perm space, the "old" classes are still references (by the singleton)
> and
> > hence don't get swept. So the "old" classloaders are themselves never
> > getting swept.
> >
> > It's not a failure of the garbage collector, or of tapestry, it's a
> > failure of imagination on the folks who decided that JVM level
> singletons
> > were a good way to solve a certain set of problems.
> >
> > --- Pat
> >
> >>-----Original Message-----
> >>From: Sergei Dubov [mailto:[EMAIL PROTECTED]
> >>Sent: Tuesday, November 29, 2005 10:24 AM
> >>To: Tapestry users
> >>Subject: Re: java.lang.OutOfMemoryError: PermGen space
> >>
> >>Perm space is garbage collected when the classloader that loaded the
> >>class files is garbage collected. And this happens not only during hot
> >>redeploys. This is info is per JVM spec.
> >>
> >>Serge.
> >>
> >
> >
> >
> >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> >
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
