sounds like something is not being cleaned up perhaps? What is your application actually doing? ex jms, db etc
We use JMS also. We have a connection pool to Postgres.Our application is coded in EJB & servlets.
probably best to isolate the problem by removing either your jms or db calls. If you can get jboss to run properly without that processing it will hopefully locate the problem
i noticed that while using jdk 1.4.1 or 1.4 some threads were not getting killed
they were set to setDaemon(false) ???
it works ok with 1.3.1
I am also facing a very similar kind of problem, most of the time the jvm crashes and sometimes java process are live but there is no response. In that case we have to explicitly kill the java process and restart java. This can be because our processes are also crossing the limits mentioned in the process table. How do we increase the max process limit ?
We have a total of 1 G.B RAM and we have set the following parameters while starting JBoss
We get the following error when JVM crashes :
HotSpot Virtual Machine Error : 11
# Error ID : 4F530E43505002BD
# Please report this error at
# Java VM: Java HotSpot(TM) Client VM (1.3.1_04-b02 mixed mode)
# An error report file has been saved as hs_err_pid10524.log.
# Please refer to the file for further information.
Can anyone conclude, what exactly is the problem, Pls don't suggest upgrading the jvm to 1.4 as our application is not fit for it.
you may want to try the command ulimit
For example, ulimit -a gives you all current resource limit settings.
I think this 1024 limit is actually the number of concurrent open files that are limited, you could try augmenting it with
ulimit -n 2048
Thanks for your help.
We installed JDK 1.4.1 on Linux and then deployed our application on JBOSS using the same.
Initially around 117 java processes got created.
When we started load testing, count went upto 148 and as soon as test is over it came back to 118
When we stopped the server all the java processes got killed.
Hence the problem was with JDK 1.4.0_01