JNI Problem in Solaris : Getting desperate!!!.

[Tim Panton - Westhawk Ltd]

If you are sure it is a file access problem, try raising the file ulimit,
(see man limits).

Is your legacy code built with libthread? If not you might want to
synchronize access to it. 
Alternatively you could try the green threads version of the JVM.

If none of this helps, try running the
whole thing under truss, and see if
you can work out what's going on.


magesh at tpi.techprov.com wrote:
__________
>
>
>Hi gurus,
>I'm running a Servlet on Solaris with  Jrun webserver and servlet engine.
>The servlet accesses through a jni wrapper some legacy code which returns
>an account object by reading two files.
>The following error occurs at high load 50 concurrent users or more.
>Segmentation Fault
>    si_signo [11]: Segmentation Fault
>    si_errno [0]: Error 0
>    si_code [1]: SEGV_MAPERR [addr: 0x1ca0f084]
>        stackpointer=FA4FB320
>*** panic: JIT signal handler did not take the signal
>Test for suspension on thread: 48 in process 5614 ...
>
>However, when I synchronise the doGet and doPost methods of the Servlet,
>this does not occur. I'm suspecting multiple file handles opening at the
>same time that are not closed causing the problem on Solaris.
>Has anyone encountered this before. Any tips will be very helpful.
>How can I interpret the trace above????
>
>
>Any help is greatly appreciated.
>thanks,
>mmp.
>
>
>---
>To unsubscribe, mail advanced-java-unsubscribe at xcf.berkeley.edu
>To get help, mail advanced-java-help at xcf.berkeley.edu
>
>
>
>
>---
>To unsubscribe, mail advanced-java-unsubscribe at xcf.berkeley.edu
>To get help, mail advanced-java-help at xcf.berkeley.edu
>
URL http://www.westhawk.co.uk/

---
To unsubscribe, mail advanced-java-unsubscribe at xcf.berkeley.edu
To get help, mail advanced-java-help at xcf.berkeley.edu