Java crash with Tomcat
|
View:
New views
11 Messages
—
Rating Filter:
Alert me
|
 |
Java crash with Tomcat
by Chris Fowler-2
::
Rate this Message:
I'm running into an interesting problem with Tomcat 4.1.30 on CentOS 5.1
running Java 6.06. Each morning around 10am I get a SIGSEGV crash. Here is the crash report: http://support.opsdc.com/~support/crash.txt I've looked at the apache logs and this software has very little traffic and usually around 10am no one is using it. The only thing accessing the main page is a monitor that makes sure it is still running by looking for a login page. I'm not quite sure what might be causing it. Software on server: OS: CentOS 5.1 App Server: Tomcat 4.1.30 Java: 6.06 Apache: 2.0.55 Now I am running Tomcat as copied from a Fedora Core 2. I've been thinking I may need to recompile 4.1.30 from source using Jdk6. Originally I had this running on FC2,JDK1.4,Tomcat 4.1.30. I decided to upgrade to CentOS based on the extended support period. Chris _______________________________________________ ajug-members mailing list ajug-members@... http://www.ajug.org/mailman/listinfo/ajug-members |
|
|
Re: Java crash with Tomcat
by Thomas, Dave
::
Rate this Message:
You shouldn't need to recompile Tomcat, simply use a binary release,
latest 4.1.37. If possible, use a newer Tomcat, such as 6.0. Check what else the server is running (such as a cron job). If all else fails, test & replace the RAM. -----Original Message----- From: Chris Fowler [mailto:cfowler@...] Sent: Monday, June 23, 2008 10:57 AM To: ajug-members@... Subject: [ajug-members] Java crash with Tomcat I'm running into an interesting problem with Tomcat 4.1.30 on CentOS 5.1 running Java 6.06. Each morning around 10am I get a SIGSEGV crash. Here is the crash report: http://support.opsdc.com/~support/crash.txt I've looked at the apache logs and this software has very little traffic and usually around 10am no one is using it. The only thing accessing the main page is a monitor that makes sure it is still running by looking for a login page. I'm not quite sure what might be causing it. Software on server: OS: CentOS 5.1 App Server: Tomcat 4.1.30 Java: 6.06 Apache: 2.0.55 Now I am running Tomcat as copied from a Fedora Core 2. I've been thinking I may need to recompile 4.1.30 from source using Jdk6. Originally I had this running on FC2,JDK1.4,Tomcat 4.1.30. I decided to upgrade to CentOS based on the extended support period. Chris _______________________________________________ ajug-members mailing list ajug-members@... http://www.ajug.org/mailman/listinfo/ajug-members ********************************************************************** This email, its content and any attachments is PRIVATE AND CONFIDENTIAL to TANDBERG Television, Part of the Ericsson Group. If received in error please notify the sender and destroy the original message and attachments. www.tandbergtv.com ********************************************************************** _______________________________________________ ajug-members mailing list ajug-members@... http://www.ajug.org/mailman/listinfo/ajug-members |
|
|
Re: Java crash with Tomcat
by lawsonma
::
Rate this Message:
Interesting... Do you have a Cron running around that time in the morning? Maybe, another program is competing for the same memory location.
-------------- Original message from Chris Fowler <cfowler@...>: -------------- |
|
|
Re: Java crash with Tomcat
by Chris Fowler-2
::
Rate this Message:
lawsonma@... wrote:
> Interesting... Do you have a Cron running around that time in the > morning? Maybe, another program is competing for the same memory > location. > Cron is running but I see nothing that it may be doing during that period. I also see no log messages in syslog for anything either. _______________________________________________ ajug-members mailing list ajug-members@... http://www.ajug.org/mailman/listinfo/ajug-members |
|
|
Re: Java crash with Tomcat
by Sony Antony
::
Rate this Message:
SIGSEGV can only happen from native code and almost certainly as a
result of a dangling pointer. ( and not from other external processes ) you can print the stacktrace from the resulting core using gdb, to see which subsystem generated the core. --sony On 6/23/08, Chris Fowler <cfowler@...> wrote: > I'm running into an interesting problem with Tomcat 4.1.30 on CentOS 5.1 > running Java 6.06. Each morning around > 10am I get a SIGSEGV crash. > > Here is the crash report: > > http://support.opsdc.com/~support/crash.txt > > I've looked at the apache logs and this software has very little traffic > and usually around 10am no one is using it. > The only thing accessing the main page is a monitor that makes sure it > is still running by looking for a login page. > I'm not quite sure what might be causing it. > > Software on server: > OS: CentOS 5.1 > App Server: Tomcat 4.1.30 > Java: 6.06 > Apache: 2.0.55 > > Now I am running Tomcat as copied from a Fedora Core 2. I've been > thinking I may need to recompile 4.1.30 from source using Jdk6. Originally > I had this running on FC2,JDK1.4,Tomcat 4.1.30. I decided to upgrade to > CentOS based on the extended support period. > > Chris > > > _______________________________________________ > ajug-members mailing list > ajug-members@... > http://www.ajug.org/mailman/listinfo/ajug-members > _______________________________________________ ajug-members mailing list ajug-members@... http://www.ajug.org/mailman/listinfo/ajug-members |
|
|
Re: Java crash with Tomcat
by Jesse Barnum-11
::
Rate this Message:
This line:
> =>0x095e5000 JavaThread "CompilerThread0" daemon > [_thread_in_native, id=25029, stack(0xa0767000,0xa07e8000)] Makes me think that the problem is with a JSP page, which is being compiled in response to a request that comes in at around 10 am. The compiler is calling native code, which is crashing. Does the site get updated overnight, which would cause the JSP pages to have new modification dates? Also, check the Apache httpd logs around that time and see if you notice a pattern of a certain request for a JSP that causes the crash. --Jesse Barnum, President, 360Works http://www.360works.com 770-234-9293 FBA Platinum, FileMaker 8 Certified == SuperContainer: A Better Container Field == On Jun 23, 2008, at 1:57 PM, Chris Fowler wrote: > I'm running into an interesting problem with Tomcat 4.1.30 on > CentOS 5.1 > running Java 6.06. Each morning around > 10am I get a SIGSEGV crash. > > Here is the crash report: > > http://support.opsdc.com/~support/crash.txt > > I've looked at the apache logs and this software has very little > traffic > and usually around 10am no one is using it. > The only thing accessing the main page is a monitor that makes sure it > is still running by looking for a login page. > I'm not quite sure what might be causing it. > > Software on server: > OS: CentOS 5.1 > App Server: Tomcat 4.1.30 > Java: 6.06 > Apache: 2.0.55 > > Now I am running Tomcat as copied from a Fedora Core 2. I've been > thinking I may need to recompile 4.1.30 from source using Jdk6. > Originally > I had this running on FC2,JDK1.4,Tomcat 4.1.30. I decided to > upgrade to > CentOS based on the extended support period. > > Chris > > > _______________________________________________ > ajug-members mailing list > ajug-members@... > http://www.ajug.org/mailman/listinfo/ajug-members _______________________________________________ ajug-members mailing list ajug-members@... http://www.ajug.org/mailman/listinfo/ajug-members |
|
|
Re: Java crash with Tomcat
by Chris Fowler-2
::
Rate this Message:
Jesse Barnum wrote:
> This line: > > >> =>0x095e5000 JavaThread "CompilerThread0" daemon >> [_thread_in_native, id=25029, stack(0xa0767000,0xa07e8000)] >> > > Makes me think that the problem is with a JSP page, which is being > compiled in response to a request that comes in at around 10 am. The > compiler is calling native code, which is crashing. > There is only one JSP page and it is not being called at that time. > Does the site get updated overnight, which would cause the JSP pages > to have new modification dates? Also, check the Apache httpd logs > around that time and see if you notice a pattern of a certain request > for a JSP that causes the crash. > > I've checked the access_log of apache and I've check syslog. The only thing that hits those pages at 10am is a perl program that simply parses the login.htm page to verify that tomcat is still serving up the application. Every morning around 1am tomcat is restarted. --Jesse Barnum, President, 360Works > http://www.360works.com 770-234-9293 > FBA Platinum, FileMaker 8 Certified > == SuperContainer: A Better Container Field == > > On Jun 23, 2008, at 1:57 PM, Chris Fowler wrote: > > >> I'm running into an interesting problem with Tomcat 4.1.30 on >> CentOS 5.1 >> running Java 6.06. Each morning around >> 10am I get a SIGSEGV crash. >> >> Here is the crash report: >> >> http://support.opsdc.com/~support/crash.txt >> >> I've looked at the apache logs and this software has very little >> traffic >> and usually around 10am no one is using it. >> The only thing accessing the main page is a monitor that makes sure it >> is still running by looking for a login page. >> I'm not quite sure what might be causing it. >> >> Software on server: >> OS: CentOS 5.1 >> App Server: Tomcat 4.1.30 >> Java: 6.06 >> Apache: 2.0.55 >> >> Now I am running Tomcat as copied from a Fedora Core 2. I've been >> thinking I may need to recompile 4.1.30 from source using Jdk6. >> Originally >> I had this running on FC2,JDK1.4,Tomcat 4.1.30. I decided to >> upgrade to >> CentOS based on the extended support period. >> >> Chris >> >> >> _______________________________________________ >> ajug-members mailing list >> ajug-members@... >> http://www.ajug.org/mailman/listinfo/ajug-members >> > > > _______________________________________________ > ajug-members mailing list > ajug-members@... > http://www.ajug.org/mailman/listinfo/ajug-members > _______________________________________________ ajug-members mailing list ajug-members@... http://www.ajug.org/mailman/listinfo/ajug-members |
|
|
Re: Java crash with Tomcat
by Chris Fowler-2
::
Rate this Message:
This may shed some light.
when we originally moved from 1.4 -> 6.06 on CentOS 5.1 our RMI code would experience the same crash. Whenever we would start the RMI registry and the server it would immediately sigsegv. Since we were planning to remove all RMI code anyway, I went ahead and rewrote the class in the application that called the RMI server and had it get the data needed another way. The RMI server ran on the same server as the application so it was not really needed to begin with. Even recompiling the RMI code did not solve our problem of crashing. This was crashing inside of a virtual machine. Jesse Barnum wrote: > This line: > > >> =>0x095e5000 JavaThread "CompilerThread0" daemon >> [_thread_in_native, id=25029, stack(0xa0767000,0xa07e8000)] >> > > Makes me think that the problem is with a JSP page, which is being > compiled in response to a request that comes in at around 10 am. The > compiler is calling native code, which is crashing. > > Does the site get updated overnight, which would cause the JSP pages > to have new modification dates? Also, check the Apache httpd logs > around that time and see if you notice a pattern of a certain request > for a JSP that causes the crash. > > --Jesse Barnum, President, 360Works > http://www.360works.com 770-234-9293 > FBA Platinum, FileMaker 8 Certified > == SuperContainer: A Better Container Field == > > On Jun 23, 2008, at 1:57 PM, Chris Fowler wrote: > > >> I'm running into an interesting problem with Tomcat 4.1.30 on >> CentOS 5.1 >> running Java 6.06. Each morning around >> 10am I get a SIGSEGV crash. >> >> Here is the crash report: >> >> http://support.opsdc.com/~support/crash.txt >> >> I've looked at the apache logs and this software has very little >> traffic >> and usually around 10am no one is using it. >> The only thing accessing the main page is a monitor that makes sure it >> is still running by looking for a login page. >> I'm not quite sure what might be causing it. >> >> Software on server: >> OS: CentOS 5.1 >> App Server: Tomcat 4.1.30 >> Java: 6.06 >> Apache: 2.0.55 >> >> Now I am running Tomcat as copied from a Fedora Core 2. I've been >> thinking I may need to recompile 4.1.30 from source using Jdk6. >> Originally >> I had this running on FC2,JDK1.4,Tomcat 4.1.30. I decided to >> upgrade to >> CentOS based on the extended support period. >> >> Chris >> >> >> _______________________________________________ >> ajug-members mailing list >> ajug-members@... >> http://www.ajug.org/mailman/listinfo/ajug-members >> > > > _______________________________________________ > ajug-members mailing list > ajug-members@... > http://www.ajug.org/mailman/listinfo/ajug-members > _______________________________________________ ajug-members mailing list ajug-members@... http://www.ajug.org/mailman/listinfo/ajug-members |
|
|
Re: Java crash with Tomcat
by Barry Hawkins-2
::
Rate this Message:
Chris,
You aren't by chance using a Tomcat package prepared by CentOS, are you? I ask because in the GNU/Linux packaging world, packages will often be built using Free (as in freedom AND beer) runtimes for Java, and you can occasionally have odd things happen when said package runs on the Sun JVMs. If this is your scenario, I'd recommend a manual install of the appropriate Tomcat binary from Apache and see if you observe any difference. That way, you can rule out the Tomcat you are using. I have also observed the JPackage team (those who package lots of Java stuff for the RedHat free alternatives, including CentOS) *omitting* certain classes which were problematic to have compile on Free runtimes. When I was the Debian Lucene maintainer, people kept griping at me for how long it was taking, saying that JPackage had it for months ahead of me. For kicks, I learned the basics of RPM formats and unpacked their Lucene package. They had omitted the one class that had been giving me fits. It was the one RMI class in Lucene at the time :-). Regards, Barry On Jun 23, 2008, at 8:13 PM, Chris Fowler wrote: > This may shed some light. > > when we originally moved from 1.4 -> 6.06 on CentOS 5.1 our RMI code > would experience the same crash. Whenever we would start the RMI > registry and the server it would immediately sigsegv. > > Since we were planning to remove all RMI code anyway, I went ahead and > rewrote the class in the application that called the RMI server and > had > it get the data needed another way. The RMI server ran on the same > server as the application so it was not really needed to begin with. > > Even recompiling the RMI code did not solve our problem of crashing. > This was crashing inside of a virtual machine. > > > Jesse Barnum wrote: >> This line: >> >> >>> =>0x095e5000 JavaThread "CompilerThread0" daemon >>> [_thread_in_native, id=25029, stack(0xa0767000,0xa07e8000)] >>> >> >> Makes me think that the problem is with a JSP page, which is being >> compiled in response to a request that comes in at around 10 am. The >> compiler is calling native code, which is crashing. >> >> Does the site get updated overnight, which would cause the JSP pages >> to have new modification dates? Also, check the Apache httpd logs >> around that time and see if you notice a pattern of a certain request >> for a JSP that causes the crash. >> >> --Jesse Barnum, President, 360Works >> http://www.360works.com 770-234-9293 >> FBA Platinum, FileMaker 8 Certified >> == SuperContainer: A Better Container Field == >> >> On Jun 23, 2008, at 1:57 PM, Chris Fowler wrote: >> >> >>> I'm running into an interesting problem with Tomcat 4.1.30 on >>> CentOS 5.1 >>> running Java 6.06. Each morning around >>> 10am I get a SIGSEGV crash. >>> >>> Here is the crash report: >>> >>> http://support.opsdc.com/~support/crash.txt >>> >>> I've looked at the apache logs and this software has very little >>> traffic >>> and usually around 10am no one is using it. >>> The only thing accessing the main page is a monitor that makes >>> sure it >>> is still running by looking for a login page. >>> I'm not quite sure what might be causing it. >>> >>> Software on server: >>> OS: CentOS 5.1 >>> App Server: Tomcat 4.1.30 >>> Java: 6.06 >>> Apache: 2.0.55 >>> >>> Now I am running Tomcat as copied from a Fedora Core 2. I've been >>> thinking I may need to recompile 4.1.30 from source using Jdk6. >>> Originally >>> I had this running on FC2,JDK1.4,Tomcat 4.1.30. I decided to >>> upgrade to >>> CentOS based on the extended support period. >>> >>> Chris >>> >>> >>> _______________________________________________ >>> ajug-members mailing list >>> ajug-members@... >>> http://www.ajug.org/mailman/listinfo/ajug-members >>> >> >> >> _______________________________________________ >> ajug-members mailing list >> ajug-members@... >> http://www.ajug.org/mailman/listinfo/ajug-members >> > > > _______________________________________________ > ajug-members mailing list > ajug-members@... > http://www.ajug.org/mailman/listinfo/ajug-members _______________________________________________ ajug-members mailing list ajug-members@... http://www.ajug.org/mailman/listinfo/ajug-members |
|
|
Re: Java crash with Tomcat
by Chris Fowler-2
::
Rate this Message:
Barry Hawkins wrote:
> Chris, > You aren't by chance using a Tomcat package prepared by CentOS, > are you? I ask because in the GNU/Linux packaging world, packages > will often be built using Free (as in freedom AND beer) runtimes for > Java, and you can occasionally have odd things happen when said > package runs on the Sun JVMs. If this is your scenario, I'd recommend > a manual install of the appropriate Tomcat binary from Apache and see > if you observe any difference. That way, you can rule out the Tomcat > you are using. > I have also observed the JPackage team (those who package lots of > Java stuff for the RedHat free alternatives, including CentOS) > *omitting* certain classes which were problematic to have compile on > Free runtimes. When I was the Debian Lucene maintainer, people kept > griping at me for how long it was taking, saying that JPackage had it > for months ahead of me. For kicks, I learned the basics of RPM > formats and unpacked their Lucene package. > They had omitted the one class that had been giving me fits. It > was the one RMI class in Lucene at the time :-). > > Regards, > Barry > Actually no, we don't do that. We create or compile all packages needed for the software and install in /opt. We compile perl from stable. We download java from sun. In our case we downloaded the tomcat binaries instead of source. _______________________________________________ ajug-members mailing list ajug-members@... http://www.ajug.org/mailman/listinfo/ajug-members |
|
|
Re: Java crash with Tomcat
by Mike Millson-2
::
Rate this Message:
Take a look at section 2.2.3.1 of the Troubleshooting and Diagnostic
Guide[1]. Try using the -client option to force the HotSpot Client VM or exclude the org.apache.catalina.core.StandardWrapperValve.createFilterChain() method in .hotspot_compiler in the Tomcat working directory. Regards, Mike [1]http://java.sun.com/j2se/1.5/pdf/jdk50_ts_guide.pdf Chris Fowler wrote: > I'm running into an interesting problem with Tomcat 4.1.30 on CentOS 5.1 > running Java 6.06. Each morning around > 10am I get a SIGSEGV crash. > > Here is the crash report: > > http://support.opsdc.com/~support/crash.txt > > I've looked at the apache logs and this software has very little traffic > and usually around 10am no one is using it. > The only thing accessing the main page is a monitor that makes sure it > is still running by looking for a login page. > I'm not quite sure what might be causing it. > > Software on server: > OS: CentOS 5.1 > App Server: Tomcat 4.1.30 > Java: 6.06 > Apache: 2.0.55 > > Now I am running Tomcat as copied from a Fedora Core 2. I've been > thinking I may need to recompile 4.1.30 from source using Jdk6. Originally > I had this running on FC2,JDK1.4,Tomcat 4.1.30. I decided to upgrade to > CentOS based on the extended support period. > > Chris > > > _______________________________________________ > ajug-members mailing list > ajug-members@... > http://www.ajug.org/mailman/listinfo/ajug-members > > > _______________________________________________ ajug-members mailing list ajug-members@... http://www.ajug.org/mailman/listinfo/ajug-members |
| Free Forum Powered by Nabble | Forum Help |