Joel,
I put Tomcat 3.2.2b5 on a dual processor NTServer box. I've been beating on
it all morning with JMeter using a sequence of JSPs and servlets from the
examples webapp that ships with Tomcat. Several times while the stress test
is running I touch to update the timestamp on all the JSP files to force a
re-compile.
So far its been running fine and I haven't seen any thread hang problems.
I'd be interested to know if you can duplicate the deadlock problem on your
system using the stuff from the examples webapp?
> -----Original Message-----
> From: Joel Kozlow [mailto:[EMAIL PROTECTED]]
> Sent: Tuesday, May 22, 2001 6:58 PM
> To: '[EMAIL PROTECTED]'
> Subject: RE: Thread deadlocking in 3.2.2 beta 5?
>
>
> Hello Marc,
>
> You and I think alike :). I have already upgraded the single
> processor Server to 3.2.2 beta 5. I ran a battery of tests at it to
> simulate heavy load. So far, there are no thread deadlocking errors
> reported. Thus, I think you are correct in your assessment that this may
> have something to do with a multi-processor server. However, before we go
> too far down this road, I thought I would see if any one else is running
> Tomcat on a multi-processor box with NT 4?
> I would also be happy to help out in diagnosing this problem. Again,
> if we confirm that no one else has really tested a multi-processor
> environment, I can take the following steps:
>
> * Install Tomcat 3.2.2 beta 5 on a development server
> * Run the product with debug logging
> * If necessary, run the entire server in a JBuilder debug session
> * Provide any test results that you request
>
> Please let me know how I may be of further assistance. Thank you
> again for
> your help.
>
> Best regards,
>
> Joel Kozlow
>
> -----Original Message-----
> From: Marc Saegesser [mailto:[EMAIL PROTECTED]]
> Sent: Tuesday, May 22, 2001 2:27 PM
> To: [EMAIL PROTECTED]
> Subject: RE: Thread deadlocking in 3.2.2 beta 5?
>
>
> It is possible that running on a multi-processor machine has uncovered a
> thread synchronization problem. Could you provide a little more
> information
> about the application? Are JSP files changing and being
> recompiled often or
> is it mostly static?
>
> Could try running Tomcat 3.2.2b5 on a single processor machine
> and see what
> happens? I'll try to look into this but these are things are notoriously
> hard to track down.
>
> > -----Original Message-----
> > From: Joel Kozlow [mailto:[EMAIL PROTECTED]]On Behalf Of
> > midian
> > Sent: Tuesday, May 22, 2001 11:40 AM
> > To: '[EMAIL PROTECTED]'
> > Subject: Thread deadlocking in 3.2.2 beta 5?
> >
> >
> > Hello,
> >
> > I have an app I just ported to Tomcat 3.2.2 beta 5 that runs
> > fine for a
> > while, and then begins to run slower and slower until it dies.
> > From looking
> > at the log files, it appears that 3.2.2 beta 5 is thread deadlocking and
> > thus running out of resources. The log file shows "Full thread dump..."
> > sections where you can see the JVM is having problems releasing threads.
> > The really weird thing is that this is not a high volume server. It
> > received no more than 300 hits over the period of a week.
> > However, that was
> > enough traffic to kill it.
> > I have the same exact app running on another machine under
> > 3.2.1 with no
> > problems. If you look at the log file on the 3.2.1 machine,
> there are no
> > "Full thread dump..." errors and the app has been up for months. Thus,
> > something appears to have changed between these versions that
> is affecting
> > my app's thread usage.
> > There are a few environmental differences between the two
> > implementations of Tomcat. I have 3.2.2 beta 5 on a quad
> > processor WinNT 4
> > Sp6a machine w/128mb allocated to Tomcat. 3.2.1 runs on a
> single processor
> > Win2K server w/64mb allocated to Tomcat. Could the
> multi-processor server
> > be a factor? Both servers are using Sun JDK 1.2.2.
> > Since the log file to 3.2.2 beta 5 is so big, I have posted it to
> > the URL : http://www.poptarantula.com/logs/jvm.stderr_5_22_01.txt
> >
> > Does any one have any idea what could be causing this behavior? Are
> > there any known fixes? Thank you in advance for your help.
> >
> > Best regards,
> >
> > Joel Kozlow
> >
> > [EMAIL PROTECTED]
> > Sn. Developer
> > Hewtlett-Packard
