You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@tomcat.apache.org by Sebastien Sahuc <ss...@imediation.com> on 2000/09/21 23:45:17 UTC

JVM core dump with JDK1.3 win32 due bad File Descriptor

Hello,

Here is a post I've sent to the Jboss list:

  Hello,

  I don't know if it was done yet, but I finally succeeded in integrating 
  Catalina in jboss. 

  It wasn't a piece of cake, since it generates core dumps on my win 
JDK1.3 
  JVM. 
  I was getting EXCEPTION_ACCESS_VIOLATION from all the place, and when I 
  forced to NON JIT, the error was a 'little' more explicit : '
  java.util.MissingResourceException can't find bundle for base name 
  org.apache.catalina.valves.LocalStrings' 
  whereas the resource was in the catalina.jar !!!!

  So I traced down the ResourceBundle which took me way too long... in 
  java.util.ResourceBundle I had to trace an exception that was caucht 
but 
  simply ignored :-( (line 944) 
  Finally the original exception is :
  java.io.IOException: Bad file descriptor
  
  Does anyone know what this mean ? Is there some limit in these file 
  decriptors ?
  
   Is it a JVM bug (because with JDK1.2.2 win it does work) ?
  
  So to make everything working, I had to extract from the catalina.jar 
the 
  necessary LocalStrings.properties (there is 10 of them) into a classes 
  directory, and after reshaping the classpath, everything worked fine. 
Any 
  idea on what's happening ?
  
  So beside this issue, I'm willing to share with the other my 
  'CatalinaMBean', and the guideline (way similar to Tomcat3.x). Thus I'm 
  expecting directive from the board :-)
  
  All the best,

  Sebastien


Has anybody an idea of what is going on ?

All the best,

Sebastien



Re: re jvm core dump with jdk1.3 win32 due bad file descriptor@1.sdm

Posted by Sebastien Sahuc <ss...@imediation.com>.
Geoff Soutter wrote :


> As Craig recently reminded me:

> If the JVM crashes and it's using zero native code, then the only
> possible
> cause is a JVM Bug.

> It may be related to this bug I found on the the BugParade which looked
> kinda similar:

> http://developer.java.sun.com/developer/bugParade/bugs/4369396.html

> looks like the 1.3 Win32 JVM is more of a beta release than an FCS
> release
> (I think they released it to co-incide with JavaOne didn't they?).
> -( Hopefully they'll refresh it soon...

Someone from the Jboss list wrote me that :

  Could this have something to do with the manifest entry?

  I know that I have had similar problems that went away when would
  extract everything from a jar file and rejar it without manifest
  generation.

And IT ACTUALLY SOLVED the problem. So with Manifest files in catalina.jar 
---> core dump , and without (M option in jar tool) --> No problem so far.

I should remind that the problem is more related to the JDK1.3 JVM on 
win32 OS than to Catalina code, but who knows how many people is going to 
report the issue ?

The note is here for archival, I might tell it to sun (if not done yet).

All the best,

Sebastien






> cheers

> Geoff

> ----- Original Message -----
> From: "Sebastien Sahuc" <ss...@imediation.com>
> To: <to...@jakarta.apache.org>
> Sent: Friday, September 22, 2000 7:45 AM
> Subject: JVM core dump with JDK1.3 win32 due bad File Descriptor


> Hello,

> Here is a post I've sent to the Jboss list:

>   Hello,

>   I don't know if it was done yet, but I finally succeeded in
> integrating
>   Catalina in jboss.

>   It wasn't a piece of cake, since it generates core dumps on my win
> JDK1.3
>   JVM.
>   I was getting EXCEPTION_ACCESS_VIOLATION from all the place, and when
> I
>   forced to NON JIT, the error was a 'little' more explicit : '
>   java.util.MissingResourceException can't find bundle for base name
>   org.apache.catalina.valves.LocalStrings'
>   whereas the resource was in the catalina.jar !!!!

>   So I traced down the ResourceBundle which took me way too long... in
>   java.util.ResourceBundle I had to trace an exception that was caucht
> but
>   simply ignored :-( (line 944)
>   Finally the original exception is :
>   java.io.IOException: Bad file descriptor

>   Does anyone know what this mean ? Is there some limit in these file
>   decriptors ?

>    Is it a JVM bug (because with JDK1.2.2 win it does work) ?

>   So to make everything working, I had to extract from the catalina.jar
> the
>   necessary LocalStrings.properties (there is 10 of them) into a classes
>   directory, and after reshaping the classpath, everything worked fine.
> Any
>   idea on what's happening ?

>   So beside this issue, I'm willing to share with the other my
>   'CatalinaMBean', and the guideline (way similar to Tomcat3.x). Thus
> I'm
>   expecting directive from the board :-)

>   All the best,

>   Sebastien


> Has anybody an idea of what is going on ?

> All the best,

> Sebastien



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




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

Re: JVM core dump with JDK1.3 win32 due bad File Descriptor

Posted by Paul Speed <ps...@progeeks.com>.

Geoff Soutter wrote:
> 
> As Craig recently reminded me:
> 
> If the JVM crashes and it's using zero native code, then the only possible
> cause is a JVM Bug.

	True, although a month ago or so I was able to get Tomcat
to consistently crash my JVM by leaving out the class designation for
a custom tag.  So there must be some strange ClassLoader issues there.
Probably completely unrelated to your problem, but I thought I would 
mention it in case it gives you something else to try.

	-Paul

> 
> It may be related to this bug I found on the the BugParade which looked
> kinda similar:
> 
> http://developer.java.sun.com/developer/bugParade/bugs/4369396.html
> 
> looks like the 1.3 Win32 JVM is more of a beta release than an FCS release
> (I think they released it to co-incide with JavaOne didn't they?).
> -( Hopefully they'll refresh it soon...
> 
> cheers
> 
> Geoff
> 
> ----- Original Message -----
> From: "Sebastien Sahuc" <ss...@imediation.com>
> To: <to...@jakarta.apache.org>
> Sent: Friday, September 22, 2000 7:45 AM
> Subject: JVM core dump with JDK1.3 win32 due bad File Descriptor
> 
> Hello,
> 
> Here is a post I've sent to the Jboss list:
> 
>   Hello,
> 
>   I don't know if it was done yet, but I finally succeeded in integrating
>   Catalina in jboss.
> 
>   It wasn't a piece of cake, since it generates core dumps on my win
> JDK1.3
>   JVM.
>   I was getting EXCEPTION_ACCESS_VIOLATION from all the place, and when I
>   forced to NON JIT, the error was a 'little' more explicit : '
>   java.util.MissingResourceException can't find bundle for base name
>   org.apache.catalina.valves.LocalStrings'
>   whereas the resource was in the catalina.jar !!!!
> 
>   So I traced down the ResourceBundle which took me way too long... in
>   java.util.ResourceBundle I had to trace an exception that was caucht
> but
>   simply ignored :-( (line 944)
>   Finally the original exception is :
>   java.io.IOException: Bad file descriptor
> 
>   Does anyone know what this mean ? Is there some limit in these file
>   decriptors ?
> 
>    Is it a JVM bug (because with JDK1.2.2 win it does work) ?
> 
>   So to make everything working, I had to extract from the catalina.jar
> the
>   necessary LocalStrings.properties (there is 10 of them) into a classes
>   directory, and after reshaping the classpath, everything worked fine.
> Any
>   idea on what's happening ?
> 
>   So beside this issue, I'm willing to share with the other my
>   'CatalinaMBean', and the guideline (way similar to Tomcat3.x). Thus I'm
>   expecting directive from the board :-)
> 
>   All the best,
> 
>   Sebastien
> 
> Has anybody an idea of what is going on ?
> 
> All the best,
> 
> Sebastien
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org

Re: JVM core dump with JDK1.3 win32 due bad File Descriptor

Posted by Geoff Soutter <ge...@whitewolf.com.au>.
As Craig recently reminded me:

If the JVM crashes and it's using zero native code, then the only possible
cause is a JVM Bug.

It may be related to this bug I found on the the BugParade which looked
kinda similar:

http://developer.java.sun.com/developer/bugParade/bugs/4369396.html

looks like the 1.3 Win32 JVM is more of a beta release than an FCS release
(I think they released it to co-incide with JavaOne didn't they?).
-( Hopefully they'll refresh it soon...

cheers

Geoff

----- Original Message -----
From: "Sebastien Sahuc" <ss...@imediation.com>
To: <to...@jakarta.apache.org>
Sent: Friday, September 22, 2000 7:45 AM
Subject: JVM core dump with JDK1.3 win32 due bad File Descriptor


Hello,

Here is a post I've sent to the Jboss list:

  Hello,

  I don't know if it was done yet, but I finally succeeded in integrating
  Catalina in jboss.

  It wasn't a piece of cake, since it generates core dumps on my win
JDK1.3
  JVM.
  I was getting EXCEPTION_ACCESS_VIOLATION from all the place, and when I
  forced to NON JIT, the error was a 'little' more explicit : '
  java.util.MissingResourceException can't find bundle for base name
  org.apache.catalina.valves.LocalStrings'
  whereas the resource was in the catalina.jar !!!!

  So I traced down the ResourceBundle which took me way too long... in
  java.util.ResourceBundle I had to trace an exception that was caucht
but
  simply ignored :-( (line 944)
  Finally the original exception is :
  java.io.IOException: Bad file descriptor

  Does anyone know what this mean ? Is there some limit in these file
  decriptors ?

   Is it a JVM bug (because with JDK1.2.2 win it does work) ?

  So to make everything working, I had to extract from the catalina.jar
the
  necessary LocalStrings.properties (there is 10 of them) into a classes
  directory, and after reshaping the classpath, everything worked fine.
Any
  idea on what's happening ?

  So beside this issue, I'm willing to share with the other my
  'CatalinaMBean', and the guideline (way similar to Tomcat3.x). Thus I'm
  expecting directive from the board :-)

  All the best,

  Sebastien


Has anybody an idea of what is going on ?

All the best,

Sebastien



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