Mailing-List: contact tomcat-dev-help@jakarta.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "Tomcat Developers List" <tomcat-dev@jakarta.apache.org>
Message-ID: <004001c18dd0$8ea86880$1b7d0304@vz.dsl.genuity.net>
From: "Bill Barker" <wbarker@wilshire.com>
To: <mike@ds808.net>
Cc: <tomcat-dev@jakarta.apache.org>
References: <001901c18ccd$281f5190$944e4d0a@jwh78>
 <004501c18cf4$e657b0e0$1b7d0304@vz.dsl.genuity.net>
 <003001c18cfa$aee684a0$0b853f04@jwh68>
Subject: Re: http://nagoya.apache.org/bugzilla/show_bug.cgi?id=5390
Date: Tue, 25 Dec 2001 21:45:41 -0800
MIME-Version: 1.0
Content-Type: text/plain;
	charset="Windows-1252"
Content-Transfer-Encoding: 7bit

I've changed the (new) StreamHandlerFactory to punt on "jar" protocols.  If
you can build from source, I can tell you how to disable it altogether.  You
might want to try the 12/26 nightly.  Personally, I can't see this doing
anything but shortening the stack trace.  But then again, I can't see how
the inclusion/exclusion of Http10Interceptor can make any difference either.

I'm including the developer list in the reply, in the hopes that someone
else can spot what I'm missing.
----- Original Message -----
From: <mike@ds808.net>
To: "Bill Barker" <wbarker@wilshire.com>
Sent: Monday, December 24, 2001 8:13 PM
Subject: Re: http://nagoya.apache.org/bugzilla/show_bug.cgi?id=5390


> Hi Bill,
>
> Gee, it's xmas eve and we're pounding on the keyboard.  Hahaha - or should
I
> say Hohoho.
>
> I attached the log -- it's the whole thing because it's just one url
> request -- xxx.jsp and every error that went with it until I received a
> blank screen response.  At the end, tomcat is dead.  It will not serve any
> more requests.
>
> Let me tell you where I am.  I do have a work around although I cannot
> explain why it works!
> 1. I started tomcat using a console with the http10 connector.  Everything
> worked fine using this connector.
> 2. I then started up the web server and everything went fine using the
ajp12
> and ajp13 connectors.
> 3. I then removed the http10 connector still using the web server ajp12/13
> connectors and IT BLEW UP.
> 4. I put back the http10 connector back in and only used the ajp12/13
(step
> 2) and it works great.
> 5. I brought tomcat up as a service and only used the ajp12/13 and it
works
> great as long as the http10 connector is enabled.
>
> I repeated these steps for a few hours to make sure everything was
> reproducible.  It was.
>
> I put this fix in (enabled the http10 connector on localhost so internet
> folks can't use it) on my second test server and it also ran fine.  I'm
now
> doing a stress test on it now for a few hours to make sure everything's
cool
> before I roll it on the prod server.  I'm sure it's going to fly.
>
> If you can figure this one out, send me your mailing address and I'll send
> you some hawaiian chocolate macadamia nuts.  Aah, send me your mailing
> address anyway and I'll send you some munchies.  You've been a real pal.
> Merry xmas.
>
> Aloha my friend and thanks for your help,
> Mike
>
>
> ----- Original Message -----
> From: "Bill Barker" <wbarker@wilshire.com>
> To: <mike@ds808.net>
> Sent: Monday, December 24, 2001 5:33 PM
> Subject: Re: http://nagoya.apache.org/bugzilla/show_bug.cgi?id=5390
>
>
> It's likely that I'll only be interested in the lines below the presumably
> 1000 lines that start java.util.zip, but you can send me the whole log if
> you like.  Of course, gzipped is better.
> ----- Original Message -----
> From: <mike@ds808.net>
> To: <william.barker@wilshire.com>
> Sent: Monday, December 24, 2001 2:48 PM
> Subject: http://nagoya.apache.org/bugzilla/show_bug.cgi?id=5390
>
>
> > Yes, it's normal thru the 404 except the 404 never get's returned on the
> > response.  Instead that monster stack overflow takes over and tomcat
goes
> > dead (a blank response is sent to the browser).  If you'd like to see
the
> > full 1040+ lines of exception code associated with the log snippet I
> posted
> > to bugz, I'll be happy to email it to you.  This big exception occurs
> after
> > I type in xxx.jsp after a fresh restart of tomcat.  I'm not sure if it
was
> > cool to post that much stuff to bugz or to continue this discussion as a
> > thread.  The problem is consistent.  As soon as I take out the web.xml
> > entries for the jsp, everything runs normal.  I'll scale down my test
env
> to
> > tomcat standalone and retest.  I've already got the code running on my
> > production server and I'm afraid someone is going to try that jsp
> > load-onstartup and blow my jvm/tomcat system for my other server users.
> >
> > I'm in the office now and successfully reproduced the same error on a
> second
> > tomcat server using the code based from the 12/20 nightly build.
Anyway,
> do
> > you want me to send the big log to you or post it to bugz?
> >
> > Thanks for all your help and merry xmas.
> > Regards,
> > Mike
> >
> >
> > ------- Additional Comments From william.barker@wilshire.com  2001-12-24
> > 14:03 -------
> > Everything up to the "Status code:404" line is normal, and AFAIK none of
> the
> > changes to the App-Classloader should be able to cause this.  If you can
> add
> > the log as an attachment, I might be able to track something down, but
> since
> > my
> > simple attempt to reproduce failed to fail I don't have enough info to
go
> > on.
> >
> > If you have your own 404 handler, you might also want to check there
> > (although
> > I can't see how to write one buggy enough to blow up here :).
> >
>
>


--
To unsubscribe, e-mail:   <mailto:tomcat-dev-unsubscribe@jakarta.apache.org>
For additional commands, e-mail: <mailto:tomcat-dev-help@jakarta.apache.org>