Hmmm, good question...stack overflows are generally things that really *should* crash any java application due to their nature. But, then, we probably don't want them crashing tomcat entirely. But, then again, if the VM is out of memory, I don't exactly want it throwing the error, catching it, outputting the error, then throwing it again the next request because it's *still* out of memory, and so on. I'd say it's probably safer for tomcat to not catch it, because it really indicates that something is wrong with the code if it's throwing those types of errors, and heck, it motivates you to fix them! :)
Kris Will Glass-Husain wrote: >Okay, that's helpful info. > >Opinion question then-- is this reasonable for an error in one webapp to >bring down my entire app server? > >WILL > >-----Original Message----- >From: Scott Brickner [mailto:[EMAIL PROTECTED]] >Sent: Thursday, June 13, 2002 11:22 AM >To: 'Turbine Users List' >Subject: RE: Turbine.handleException: null > > >Remember there are three major subclasses of Throwable: Error, Exception, >and RuntimeException. The Error subclass is unchecked (like >RuntimeException), but it's *not* a subclass of Exception. >StackOverflowError is an Error. Nothing is catching it, even in the Tomcat >code. > >This is usuall reasonable, as Errors are supposed to indicate something >"really" wrong with the JVM. Out of memory, invoking an abstract method, >referencing fields or methods that don't exist, class files not verifying, >that kind of stuff. > >-----Original Message----- >From: Will Glass-Husain [mailto:[EMAIL PROTECTED]] >Sent: Thursday, June 13, 2002 12:14 PM >To: [EMAIL PROTECTED] >Subject: Turbine.handleException: null > > >Hi, > >I have some odd behavior happen occasionally when certain errors are >generated in my action code. Turbine reports the error (calling my custom >Error.vm page and my ErrorScreen). Then Tomcat inexplicably shuts down, >forcing me to restart the Tomcat process on my server. This is only true >with some errors. With most exceptions (e.g. an index out of bound), after >I get the error screen I can continue to use my app. My question is -- why >do some errors crash Tomcat? > >In my Turbine.log file, I see the exception and a stack trace (as listed >below for a stack overflow error). My Tomcat logs contain nothing useful. >I'm running on a Win XP machine with the Tomcat.exe NT service, and using >Turbine 2.2b1. (Although I'll migrate the app later to a LINUX box). >Tomcat version 4.0.2. > >Any ideas? > >Best, WILL > >[Thu Jun 13 08:59:17 PDT 2002] -- ERROR -- Turbine.handleException: null >[Thu Jun 13 08:59:17 PDT 2002] -- ERROR -- > Exception: java.lang.StackOverflowError > Stack Trace follows: > java.lang.StackOverflowError > at org.apache.oro.text.regex.Perl5Matcher.__match(Perl5Matcher.java:1043) > at org.apache.oro.text.regex.Perl5Matcher.__match(Perl5Matcher.java:1154) > at org.apache.oro.text.regex.Perl5Matcher.__match(Perl5Matcher.java:1131) > >etc for about 1000 lines. > > >-- >To unsubscribe, e-mail: ><mailto:[EMAIL PROTECTED]> >For additional commands, e-mail: ><mailto:[EMAIL PROTECTED]> > >-- >To unsubscribe, e-mail: ><mailto:[EMAIL PROTECTED]> >For additional commands, e-mail: ><mailto:[EMAIL PROTECTED]> > > >-- >To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> >For additional commands, e-mail: <mailto:[EMAIL PROTECTED]> > > -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
