tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bradley Wagner <>
Subject Re: 100% usage and hanging on startup of Spring/Hibernate app in Tomcat 5.5 on Ubuntu 7
Date Thu, 10 Sep 2009 22:42:24 GMT
A couple of other findings:
- Tomcat 5.5.28 out-of-the-box with its default webapps has no problem
starting with the default (parallel) gc
- As soon as I remove the default webapps, add ours, and modify the
context.xml file to provide it a DataSource, Tomcat hangs. We're definitely
loading a lot more into memory and inducing garbage collection with
Hibernate and Spring in the mix. I think I'm going to try a Hibernate or
Spring sample webapp to see if it has similar problems.

On Thu, Sep 10, 2009 at 4:43 PM, Caldarale, Charles R <> wrote:

> > From: Bradley Wagner []
> > Subject: Re: 100% usage and hanging on startup of Spring/Hibernate app
> > in Tomcat 5.5 on Ubuntu 7
> >
> > The *only option* that did not cause the startup process to
> > halt indefinitely on startup was the Serial GC option.
> This is really weird.  Might be time to file a bug report with Sun.
> > - Does this indicate some kind of threading problem in this Java env?
> Hard to tell without a snapshot of what's going on inside the GC code
> (which is all C++, not Java).  Could also be a problem with Ubuntu.
> > - Does the option "DoEscapeAnalysis" still use a parallel gc?
> Should be independent of the GC algorithm; it's extra logic in the JIT that
> determines whether or not an object even needs to be allocated, or if it can
> simply reside on the method's stack.
> > - Is the experimental more similar to the parallel or the serial?
> My understanding is that G1GC is both parallel and concurrent.
>  - Chuck
> MATERIAL and is thus for use only by the intended recipient. If you received
> this in error, please contact the sender and delete the e-mail and its
> attachments from all computers.
> ---------------------------------------------------------------------
> To unsubscribe, e-mail:
> For additional commands, e-mail:

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message